 Oh, I I don't know how to do the countdown thing. So we didn't have that Yeah, I don't know Yeah, I don't know how to do it. It's magic. Hi everyone We're here we're about to record the show usually there's the 32nd countdown but Kiki does that somehow and I don't know how she does it. So instead We're just saying hi and vamping for a minute while everyone joins the stream Kiki is missing tonight. I am driving. We'll see how this goes As Sue okay, I think I think we're good. So oh, yeah, so people are saying hello in the chat room So I think we are coming in loud and clear for those of you watching as Kiki usually explains This is not the polished podcast we record it live There is some editing that happens behind the scenes So if you are watching this live or if you're watching this on YouTube later You get all the extra bits you get the intro you get the outro you get the middle tro that gets cut So here we are we're gonna go through our whole 90-minute show and With all the extra bits Are you ready to do a show? Yeah, absolutely. Okay, great. So let's see if I have everything I need to be Kiki Um, I have an episode number. I have a title Okay, oh, but I don't have aha. There's one thing I'm missing. Oh, do we have music? I have my phone You've got that ready there. Okay. I have my phone We gotta remember not to talk during the music then because it might need to get edited in later Correct. Yes, so it'll get spritzed in later. This is very big very Hollywood spritzing things in Wrote snacks instead of snakes No, yeah, it's it's very similar. Okay. All right, we're gonna do it Okay, are you ready? Yeah in three two This is Twists this week in science episode number 869 recorded on Wednesday, March 30th 2022 Science appreciates manatees Hi, everyone. I'm Blair Bazderich and today we will fill your head with snakes spritz and spiders but first dis Kool-Amer disclaimer disclaimer the following program deals with subjects of a scientific nature Occasionally some people think that science is biased Against their politics their religion or their financial interests. So before we dig into stories a few words about science Science is a way of observing the world Objectively without politics without religion without commercial concerns and without human bias of any kind Science is a method a process of asking and Interrogating questions that can be broken down into steps First you ask a question based on existing knowledge or on a new observation Next you hone that question into a hypothesis Considering what the answer might be and narrowing it down to a testable core Then you make a prediction about the hypothesis and design an experiment based on that prediction Putting constraints on the experiment to remove any bias the researcher may have for a result Which is why there are randomized double-blinds control groups and sufficient sample sizes required to protect This all from bias most importantly It must be able to disprove the prediction if an experiment cannot disprove the prediction That is based on your hypothesis. There is no point in running the experiment Then you run the experiment whatever the result it alone cannot prove a hypothesis Only disprove or allow for it to be interrogated further and the final step is to ask a better question Form a new hypothesis and build another experiment experiment based on the new knowledge With this relatively simple set of instructions with this ethic of integrity of logic We make science possible without it There is no such thing as science and without science. There could be no such thing as This week in science coming up next Good science to you Blair. Oh, it's just oh, no, it's just me It's I'm the only one. Hey, if you're just doing it This week in science gonna go ahead and dive right now. That was very strange. That just happened I don't know why that just happened. Go ahead do it again. Give me a good This is a Hollywood, right? Just doing that takes okay take two good Science to you Yes, good science to you Justin Kiki is not with us this evening. She is busy doing doctor Kiki things But in the meantime Justin and I are here to give you all of the science. Thank you for joining us tonight We have an amazing show ahead on Tonight show. What do I have? I have those snakes. I mentioned I have sprays or spritzes. I have Seethrough tissues. I have sperm. I have all sorts of stories to make up for the lack of dr. Kiki What do you have Justin? I've got how AI will kill everyone caring for caribou Which COVID vaccine is the best a two-legged fossil find and How to keep AI from being rude? Rude you don't want that rude AI Especially not if it's gonna be killing us off. No, no at least have some decorum Let's have our own decorum as we jump into the show here. Let's remind everybody you can subscribe to the twist podcast the nice kind of structured knit tightly knit version of the show on all of your favorite podcast platforms and you can also watch us on YouTube Facebook all the places you can find videos like that. Just search twist I'm sure you can find it this weekend science out there in the internet ethers All right, Justin. Are you ready to do some science to chat about some science? Let's go. Let's go All right, what do you got for me at the head of the show yet? Let's see. I guess that means I'm a first year Oh, yeah, here we go I'm gonna start with the the most terrifying Artiglava ever read Yeah, you're set yourself up here for sure. Okay. This is a quote It's one of those things we wish we could disinvent Now that's not a quote from the story. That's actually from Nick Cage's character Sam good good speed in the movie The rock. Yes So it's about us he's talking about a substance called VX which is short for venomous agent X which is a nerve agent Classified as a weapon of mass destruction by the United States and UN and international bodies everywhere in the movie a team of Angry militants attempt to use VX and rockets to ransom the city of San Francisco for a hundred million dollars The movie the rock Netted over two hundred and fifty million dollars proving that threatening the city of San Francisco with VX for cash works But as much safer and more profitable to do so virtually Which brings us to the Swiss Federal Institute for Nuclear Biological and Chemical Protection Convention early this month and one of their invite event invited presenters the invited presenter was collaborations pharmaceuticals a small company private company and Collaborations pharmaceuticals was invited to speak on the potential ramifications of AI assisted drug research Now collaborations had previously designed a commercial molecule generator which used machine learning to make model predictions of bio activity to find safer molecules that would be targets that could be could use to target human disease, so they're doing good research and The models that they have in their AI penalize for predicted text toxicity and reward target activity So at first the collaborations pharmaceuticals didn't probably know what to make of this invitation They have no experience in weaponizing our visual intelligence or using it for bad purposes They thought about it though because they're scientists and they're clever clever folks So they attempted to weaponize their own AI systems to see what would happen Basically, they just inverted the logic of the AI So that it would start rewarding Toxic bio activity instead of penalizing it Using data from the same public databases, they would ordinarily be using to help derive compounds for the treatment of neurological diseases It took less than six hours for the model to generate 40,000 molecules that scored high in toxicity in the process the AI designed not only VX But also many other known chemical warfare agents Many new molecules were also designed that looked equally plausible as a chemical warfare weapon of mass destruction These new molecules when produced Were predicted not produced nothing was produced. They didn't make anything but the new molecules were predicted Actually to be more toxic than any known chemical warfare agents Orders of magnitude more toxic than VX, which can already kill a human with an amount about the size of the grain of sand And the results were completely unexpected Because the data sets they were using to were not some dark lab Secret Formula of molecules that exist on the planet. They were using Public data sets include nerve agents there may have been insecticides and normal drugs that people use these are all commercially available molecules that were in this database in fact, a lot of their virtual molecules had entirely separate properties from the many thousands of molecules in the model and the data they were feeding it so this This AI system actually built new molecules completely different than what was in What was even in the public data set? So the the better AI can predict toxicity to avoid It says somewhere in the article there The better simultaneously is it discovering lethal molecules. Sure. Sure So the next step is actually is actually making a compound out of this knowledge Which is really hard if you want to make any known chemical warfare agents because there's all sorts of international controls Lists have watched and controlled precursor chemicals even so even the chemicals you would need to make the chemicals You would need to make the thing are on these lists and if they're getting purchased or moved International community is trying to keep an eye on all of this But what of the new chemicals what if what if what if something that the AI has found Uses a completely different list of precursors Then all of those lists of of watched Precursors mean nothing all the all those safeguards. We have Employed on an international level out the window so as Collaborations farmers schools points out. They are just one very small private company amongst hundreds who have this capability The genie is already out of the medicine bottle. Mm-hmm. Yeah, so machine learning can already do this. Yeah That's kind of what I was going to ask is it doesn't seem that far away from stuff that we've done in the past Yeah, and they're pointing out that you know while theirs is a Commercial product meaning it's it's designed and built to be used by large companies that Presumably have some sort of safeguards in place There's plenty of also open-source software tools and again. They used public data sets So it's not like they were using private knowledge to begin with So their fear is that this could go on without any oversight Scientists of yeah, so this is like the most terrifying Our love ever read scientists also need to ensure that the misuse of AI is prevented and That the tools and models developed are only used for good promise Oh my gosh. Well, so this is without getting on my capitalism high horse so quickly in the show this is where Medicine being a capitalist venture is a problem It's been a problem with you know Part of fighting COVID because you don't want to have people kind of Fighting for the sole propriety of cures when we're in a global pandemic But similarly you You don't want this AI quote-unquote to fall into the wrong hands There's money to be gained right that's the that's the fear but here's the thing though They're like that's I mean, I don't know if I could quite follow you there because a lot of the software That exists has been open-sourced for public use to Sort of finding without the profit motive I'm saying so so there's no profit motive with the AI But there's a profit motive with the drugs the AI could come up with or discover or The need for drugs that this AI could create if it goes the wrong way, right? And so that's kind of what I mean is that the whole idea of pharmaceuticals being a profit-based system Feeds into this potentially being used for the wrong reasons. I would I would love to jump on an Anti-gabardless high horse with you not on this one What part of what Use of AI is is is being targeted for is all of these small Diseases that don't have enough people to get behind them like a breast cancer like a colon cancer I mean, there's there's 3,000 diseases that have such a small population that is affected and they're heavily affected by them that That profit motive isn't there to come up with the years are not enough people who are going to take it So using AI to find these molecule targets is like a quick way to avoid all the years of expensive research And try to tackle some of those smaller diseases that don't have the big profit Is there profit in business? Yes, there still is but but here's the thing It's too late, right? I think Part of part of what this would mean to take it under control from the AI side from the artificial intelligence side would be Aspects of scientific tools certain databases The way that science typically works would be forced to be locked away from the public Going forward promise that's going forward and even if that were to happen Even if there was a legislation tomorrow that looked at this and said ah, we need to you know, shut down public access to all these tools still too late Jeannie the out the Pandora's box is open. So on the other hand, I think actually Gordon the chat room might have already Might have already sort of hit on this Another another approach might be that Yeah, run it See what you could make and figure out what chemicals you would need and hopefully they're not, you know available in a household garage But then you have other precursors that you can put on those international lists and start to pre preemptively Keep those chemicals out of the hands of potential ne'er-do-wellers so But yeah, it's just a really interesting and it's funny because you know, this is Collaborations pharmaceuticals is a small private company get invited to this thing Don't have anything to bring to the talk and just run this sort of in-house experiment and They were like they're terrified Whoops look what we did Decades to do this thing that has only best of intentions and they're like, oh gosh That was way too easy way too easy It's I spilled that that milk into biochemical warfare. Oopsie anyway moving on to another actual Technological marvel in medicine I brought today This is from Scripps Research They've unveiled a new tissue clearing method for rendering large biological samples transparent so Tissue clearing involves the use of Solvents to remove molecules that make tissue opaque like fat so that they can essentially make the tissue optically transparent They can keep proteins and structures in place as they do that And so then they can use genetically encoded or antibody linked fluorescent beacons to mark active genes or other molecules in a lab animal And then in tissue clearing that allows those beacons to be imaged all at once across an entire organ or an entire animal And so this it's a really cool process that that is is really kind of underutilized in Medicine partially because it's really labor-intensive. It usually is hazardous chemicals and they It's it's been difficult to figure out how to do this to large pieces as well So they this new method Uses both organic solvents and water-based solvents the organic solvents So these are usually used separately They're organic solvents generally work more quickly and more powerfully but diminish fluorescent signals So if you're trying to use those markers, that's going to be difficult for your experiment methods using water-based solvents They're better at preserving the fluorescence, but are impractically weak for clearing anything other than brain tissue so that means that Can't really be used in other parts of the body But in this case they use a sequential combination of organic solvents and water-based detergents to make use of water-based Hydrogels to protect molecules within the tissue tissue that needs to be preserved So it's actually kind of a three-step process and the the positive thing here also is that it often did not require Constant pumping active pumping of solvents through the sample They were able to put it in a jar put it on a shaker like you'd see at home depot Just kind of agitated and it would just sit there on the bench top and then eventually it would be ready to go So this also means it is more practical and scalable in Labs that aren't specifically for this and so let me I want to I want to show you like the first thing that occurs to me night depends on where it's taking pigment from if it takes too much it's no good, but if it's leaving like This could potentially be huge for pathology labs Because then instead of getting a slice of a biopsy that's down to a couple microns thick and put onto a slide you might be able to image a the entire piece of a biopsy and and sort of look throughout the tissue For signs of disease but again, but that's all such a visual that's such a visual Thing to begin with it if I suppose if you check out too much The things that you're trying to look at it could actually make it harder to do so Yeah, so I'm trying to my computer decided it doesn't want to let me Share pictures right now, so I'll try to fix that for later But the the tissue is very cool-looking You can check out the show notes on twist org and click on the link to see the picture here, but they were able to take Mice the whole chests of mice the entire chest cavity an image SARS-CoV-2 infected cells in them In a way that you haven't they haven't holistically been able to do in the past so Yeah I'm trying to do this now. Yeah, this does have this has tremendous Then potential as a pathology tool. Oh my goodness. Hey, I'm gonna try again to share Nope, it won't let me do it. How fun. Okay. Well, anyway, that is My story about Making organs clear Anybody link fluorescent beacons that is that is really cool Yeah, what do you have you have something about caribou, uh, yeah, I do okay, so This is a caribou populations across Canada in case you haven't noticed which if unless you're in Canada You probably haven't noticed if you're in Canada My understanding is that most of Canadians live in like three or four cities And the rest is there's no Canadians very few Caribou populations across Canada have been going down largely due to human activity and There have been recovery efforts from the federal government from provincial governments But they have not been working Caribou declines have exceeded 40% in recent decades all the way across the country. So There's many populations subpopulations of Caribou that have been lost. They're just gone and So yeah, time is running out a new University of British Columbia study Finds that with the help of indigenous led conservation initiatives of Two First Nation tribes. There is one herd of mountain caribou Whose numbers are going up? Uh, what tribe elders had remembered as a sea of caribou Had declined over the years to only 38 animals back in 2013 Now, uh, nine years later the count of the herd is over 110 and numbers are on the rise The conservation and recovery effort involved some short-term actions. They do They reduce the exposure the caribou have to predators And part of the way they do that is they have tribe members Assigned to live near the caribou. They're gonna go out there and shifts to where the caribou live Uh, and they also have an active guarding of maternal pens. So when there's females who are pregnant They're sort of isolated and penned and they're you know fed that way but then they're also Vented from predation in this in this way. So it's a very hands-on thing. Tribe members collect lichen to feed the those those mother caribou And there's ongoing work for them to secure landscape protection to actually get Protection for the land surrounding the caribou so that there's no Development or hunting or recreation taking place in the area as well All of this in the effort to create a self-sustaining caribou population in the future researchers note This is an important part of decolonizing the mindset of conservation Which is historically work to exclude the views of indigenous people and i've said it a few times on this show If you want to conserve nature If you love nature Go live in a big city If you want to conserve a nature area Turn it over to the people who value it the most and in most cases this means giving Lands back to the indigenous populations that have a history of successful stewardship So because all the good contention See all for all the good intentions people may have For nature preserves The idea of keeping areas wild If there were people on those lands before colonization Give back the land and get out of the way I wish we could do that pretty universally We could we really there's I mean there's a lot of a lot of areas We absolutely could do that, you know, because there was what was this conversation? There's some conversation we had Where there was some proposal of creating this big Uh, nature preserve sort of in the middle of the country in the great plains or something like this, you know But whenever we've had these things they're federally protected This is a in a federal government come administration comes and says oh, yeah, it's it's all gonna be pristine wilderness And oil rigs. Hey, no, no, no, that wasn't part of the thing. You're changing it No, we're gonna sell off these lands because we don't need no, but that's what then why did we protect it? Stop trying to protect things federally Turn them over To the indigenous population that lived in there and stewarded it for many thousands of thousands of years And just and just be take it off of your plate It's not a thing. It's you know, I'm not saying that Western societies Can't do nature preservation But I I will say they haven't They have not done that That's not a thing. Nobody goes. Oh, yeah, western western culture greater western society That if you want to preserve nature, that's your go-to Right. Nobody's ever going to make that claim. Well, I will say that in uh, certain areas of the united states um, there is a a push to uh develop much more collaborative partnerships With native tribes um in wild spaces that still exist. So state parks county parks Even a lot of the national parks. They are starting to reach out to their local tribes to learn from them and to also Give Give some of the stewardship back and give them their props and I know in in your woods, for example, they're um They're actually revamping all of their signage to try to take out a lot of the kind of white savior lens that are on a lot of them And speak more truth to the history and the the impact of the native people the positive impact Um on the on the land before we moved in and decided to do things our own way Yeah, and and you know the to that too, I think that Learning from the tribes is a good idea, but it really it's you can still learn what they're doing After you've given the land back After you you've you've handed it over if you've not if If the u.s. Government is not going to be a good stewarder lands or the canadian government is not going to be a good Stewarder lands it should just recognize That's not the wheelhouse Turn it over Justin Baby steps. I know it's frustrating, but this is definitely a situation where you know, even 20 30 years ago This conversation wasn't really happening at the level it was happening The narrative is changing and the the trajectory is moving and I think that's good. Yes. I wish it will continue I hope that it will continue But already this is a huge turnaround from how things were not that long ago in terms of wild spaces in north america So I'm I'm looking forward to what's coming next Speaking of what's coming next you want to talk about the latest with covet Oh, that sounds like a fun subject Yeah, so if you just tune tuned in you're listening to this week in science with justin jackson myself flair basderich Dr. Kiki is off tonight, but we still have ourselves a covet update So, um, I wanted to talk about the latest nasal spray That came out in the news this week So a nasal spray from cornell university a study in cornell university looked at a nasal spray that could not only potentially prevent covet 19 illness prior to exposure but also provide early treatment if it is administered soon after infection which It's a respiratory illness. So putting it right in the source kind of makes sense Yeah So this was mice engineered with human receptors for the coronavirus on their cell surfaces They found the molecule n0385 inhibited entry of the virus into cells They also showed that it protected mice from infection prior to exposure While also providing effective treatment when administered up to 12 hours after exposure So one of the things that that they mentioned is you know, if you had enough of this And every time you think you might have been exposed you give yourself a little spritz, right? You're pretty well protected. It works early on in the effect infection. Even if you've already Been infected. It's it's a good barrier Immediately after infection They tested mice exposed to the first strain detected in washington state in 2020 and they also detect They looked at the delta strain. It has not yet been tested on omicron the way that they tracked if it got them sick or not was looking at weight and um Other clinical and pathological measurements, but weight is a good tracker of just do you as a mouse feel sick from cope it? And so they also looked at temperature. They looked at mortality, of course, and then They looked at tissue analysis to understand what was happening internally they Giving them the nasal spray prevented them from losing weight prior to exposure and from dying post infection Yeah, it's you know, that's uh It's sort of an awkward moment when so I went out for the first time went out for like went out out to uh, thing with people in a building in a public space uh to watch this, uh book Rollout talk about neanderthals Couldn't resist, right? The first human in the in the world In an enclosed space that I encountered sneezed on you immediately coughed. Oh, okay like the first human Doing one of these and I'm like The first is it everyone because I haven't been interacting with these humans Are you is it everyone or is it just you because now I don't know but I just was picturing Based on your story that there'd be this awkward moment where somebody just maybe innocently coughs or whatever But then you're like reach into the pocket pull back the mask And then they of course you can't go unnoticed so they're like, uh I have allergies doesn't matter Doesn't matter You now you talked i'm gonna have to do it again I mean, hey, I do it sort of slight slight so Seemingly slight social judgment of others. Well, if we're gonna continue re-initiating social Situations while there's still a global pandemic. There's gonna be gaps, right? um, but so this this study just to let you know, um, it is Uh currently now, uh in the fundraising phase in the hope that they can do human trials drug development formulation And of course last mass production The hope is that treatment could be submitted to the FDA for emergency use approval with a minimum of six months So by the end of this year, you might have your covid nasal spray ready to go All right Yeah, yeah, and then I see there's some stories here about which vaccine is best. Did you bring those justin? Yeah, yeah, this is uh, this is a pretty quick one It wasn't that it wasn't as thorough Looking as a test as I would like but uh, yeah question. It comes up every once in a while Which covid vaccine is best and Right, that's the classic answer. It doesn't matter. You know, you get a lot of team Pfizer team moderna folks out there making claims trash talk and the other about which one is best Meanwhile, the johnson and johnson people are always saying doesn't matter It's not a competition So they did a head-to-head to head-to-head comparison It was conducted between all three as well as nova vax Which is being developed just in time to maybe a be a second or third booster shot. It's like it like came in late Uh, but is is ready soon ish. I think So this is an as-of-yet Peer reviewed study that assessed the four vaccines using 14 metrics including levels of several types of immune cells such as t cells and b cells as as well as immune molecules called neutralizing antibodies previous comparisons that have been done between the vaccines have often used Data sets from different studies and then kind of try to compare them across studies, which is doable, but it can also You know, if you have different laboratory techniques or different things that they're looking at In specific can be can be a little tricky to really get this comparison. So for the latest study researchers applied the same techniques across all the vaccines that they investigated within a single study and They found that antibody levels induced by two doses of Pfizer or Moderna Vaccine they they dropped substantially over six months In contrast, antibody levels from j and j's one shot vaccine were stable or even increased over time So the the j and j shot You know, you got your shot and you got your protection and it might have even gotten better Whereas anybody with a Pfizer Moderna Their protection dropped substantially as he went closer to that six month point However, so j and j's best is what you're saying. However The increased antibody levels measured six months after vaccination with j and j Were still lower than those observed six months after vaccines of Pfizer Moderna It was just less effective all the way through Novavax's two shot regimen induced antibody response on par with those of the mrna vaccines However, they had some weird. There was a one weird thing. There was levels of cd8 plus t cells These are the t cells that destroy infected cells And they were low to undetectable Which is maybe red flag maybe not The other three vaccines performed very well in that metric. So it was sort of a outlier there Um, anyway researchers say that they don't intend to proclaim winner or losers Uh, this is not a competition. This is not bragging rights or anything like that This is not to elevate to reduce your sense of protection At least until all the people with the j and j shot have left the room Harvard study from late last year did find Moderna To be slightly better than Pfizer Compared to those who received the Moderna vaccine recipients of the Pfizer vaccine Had a 27 higher risk of documented SARS-CoV-2 Uh infections and a 70 higher risk of COVID-19 hospitalization Then the folks who had gotten the Moderna, but Keep in mind that's in compared to those percentages of higher Uh sound pretty big 27 higher risk of getting the infection 70 higher risk of hospitalization But that's just of the percentage that didn't get it And isn't actually that big of a number of people since Both of the vaccines were over 90 effective well over 90 effective And so you're doing a comparison with this slim number of breakthrough It's a percent of a percent of a percent. Yeah Well, this is also this came out the the one you're referencing from last year around a bunch of other Studies that said other things and that's what I noticed about the The story that you brought this week that's not yet peer reviewed is they called this the first apples to apples version Where they actually compared them all kind of in a lab Um through a direct study and all the rest were just kind of survey based data that right, which I almost Almost didn't bring this study Because of that because they said the other studies were like comparing apples and oranges and I went so they were almost exactly Right, right, right, right. But yeah, so I remember this because when I was getting my booster I was deciding do I go madurna do I go fizer and I was trying to read all these different studies And there were studies saying madurna was better There were studies saying fizer was better and the percentages were very similar But they were studying different pieces of efficacy in different ways and different subsets of population Like for example consider the fact that fizer Was approved first and who got the vaccine first? people who are older and people with Immunodeficiencies or other medical compromising things to consider So if just generally speaking those people were more likely to get fizer That could mean that those people had higher levels of mortality or hospitalization so That is the problem, right is there's no vacuum of this data There is no completely variable controlled version of Of comparing these vaccines unless you start from the beginning with all of them Yeah, and there's oh, yeah, and there's also like uh in looking at this. There was one thing that was saying, you know Fizer's 96 percent chance of preventing was Madurna was only 95 or like But yeah, you're right. You're you're pulling data out of the live stream of humanity And so in some ways that makes it difficult on the other hand You got your sample size Boy, do you have a sample size? So on some level even if your Data sets aren't precise. They're probably still accurate because of because of the volume Uh, that's out though Right and unless what I said has any sway on it Which no end end with end with what you say having three like it's a little of both. It's a little of both. Yeah, you're you're both you're both Dealing out with a moving target with its name, especially when anytime you're dealing with humans but I think what's sort of interesting would be to Randomize a trial Where you say like, okay say we got we've got we've picked 20 cities And we just our counties or whatever and we just want to see what went on in those counties And and and then of those pick a few and see if they're consistent And just randomly pick like three of those 20 and see if there's a consistency or a big inconsistency and then do that a couple of times Uh to see if there are geographical differences if there's medic because that's also that's Also a possibility when we know healthcare Isn't uniform across the united states Yeah, well that is the other issue right is you're much more likely to Die from covet if you if there's not icu space where you live So It's definitely it's they those two pieces influence each other likelihood of hospitalization and likelihood of death in a weird way There's kind of a push pull where those influence each other too But anyway, there's no study that's perfect to to compare these vaccines yet They all seem to work well So if you are still Not boosted or back. So hopefully everyone's vaccinated But even if you're not boosted just go get whichever one Just go get whichever one. It's fine It'll do it. It'll protect you um, okay, and uh moving on Into the next part of the show I want to thank everyone for listening to this weekend science with normally dr. Kiki justin jackson and myself blare basderich Dr. Kiki is off tonight, but thank you for joining us anyway Without the good doctor I want to remind you all that we can bring you all of this amazing science Because of you So if you want to help support us you can go to twist.org and click on the patreon link and support us at different levels just something to remind you all about moving on Do you know what time it is justin? Is it that time? It's that time. What time is it? Is it time for blaze animal corner? What you got blare Oh my gosh, I have so much good stuff. Okay, so first first I'm gonna tell you about fruit fly sperm Fruit fly sperm again. Yes Once again, it's time to discuss fruit fly sperm in blaze animal corner This is a study also out of cornell. I think this is my second cornell study of the evening. They were busy this week they found That four days after a sperm enters a female fruit fly Close to 20 of its proteins Are female derived? What's happening to this sperm after it enters a female? Seminal fluid proteins and then female proteins bind to sperm inside the female's body Some of the proteins attaching to sperm are metabolic enzymes which lead research to believe that they may be supporting the sperms viability Propagating them like a beautiful fern in your home So in flies and in many other animals sperm live for many days inside the female So how does that sperm survive without degrading for those many days before it can be used? It looks like the female may be providing enzymes they need to keep them nourished and healthy Although sperm come from the male and are comprised of male DNA and cellular structures There's You know, there's also this expectation that the female body is a hostile environment for sperm There's this expectation that that's why sperm doesn't last very long in the female body is that The sperm come from the male. They are foreign dna and the female body attacks it But in fact it looks like the female body is adopting and nurturing that sperm The the reproduction could fail ultimately if the If the female did not nurture that sperm and the sperm died too quickly inside the female body So once inside that female the sperm slowly becomes less and less Male just dna speaking just because of all these other enzymes things that bind to the to the sperm So in evolutionarily, yeah, go ahead. Oh, I would seem to then if potentially if you had an A Malformed ill fitted mutated like an unhealthy sperm Then perhaps you might not have the right binding site or location For these enzymes and and in a way that also becomes a filter For making a better egg Absolutely, that's possible. Yeah, so I know the other thing about Um ill formed incomplete misshapen sperm Not to be insulting to them Is that they are usually slow and bad at moving And so even if you do live for up to four days Even if the female body can nurture and keep that sperm alive for that long It will not make it to the egg because it is not shaped right To swim properly So so you're right. That could be something that could happen But also there's that's kind of the failsafe evolutionarily, but you'd see how this could be an evolutionary benefit And a case of cooperation between the male and the female to make reproduction successful you think about the fact that um, it depends on the organism, but Dependent on how many eggs are released how often they're released all these sorts of things could impact the exact timing being right when you are dealing with internal fertilization and so being able to Propagate that sperm and keep it going is to make sure that Copulation is not wasted So it's I think it's interesting in the context of the previous narrative of the female body being a hostile environment That actually it's the exact opposite the female body is kind of encasing and and And and and helping the sperm survive as long as it can It's kind of the exact opposite Well, this seems to be and I don't know where it comes from very long history of Putting negative context on female reproduction. Oh, of course I don't know It's because people hear what they do not understand That is my guess is because you think about what is very evident and um, visible to the naked eye and um easy to interpret in the male reproductive system in the female reproductive system It's it's all internal. It's a lot of it is it's uh, some of it's microscopic And it's yes, I think it's a fear of what you don't understand It's why some members of congress thought that you could reach a baby through the throat, right? It's a it's a miss mystification of of how the internal processes work, but Now now we understand a little bit more I'm just gonna guess though that that was not a woman It was not you're right Yeah, this is back to a couple weeks ago near valentine's day I had a story about sex education in schools and how having it early and positive based focused on pleasure and also Co-ed and all these things are very helpful facts. And I'm just just starting even with facts. Yeah, just biological anatomy and facts can also Apparently then be helpful to future congress people. Oh gosh boy. Anyway, uh moving on from from This very strange topic about sperm. Let's move on to snakes Squeezing snakes snakes like to squeeze Some of them do what do you call snakes that squeeze squeezer snakes? What do you call them? That's right pop quiz past so Constrictors they squeeze their victims before swallowing them whole What do you think actually might be a challenge? With the constriction process and you know being a living breathing animal Oh, uh, right. You would have to have some sort of reinforcement because you're also Constricting yourself. Yeah, so I'll tell everyone listening and watching you can put your hands on your uh, Your chest or just below your chest around your ribcage and take a deep breath in and out And in and out So as opposed to how you might imagine it in like a cartoon you breathe in you get bigger you breathe out You contract right actually our ribcage works the opposite because we have a diaphragm So when we're within we actually lengthen and when you breathe out I did that. I actually expands. Yeah, so when we if we like a mammal in the grips of a snake We're being constricted. We would breathe in And couldn't breathe out anymore and we breathe in And we couldn't breathe out anymore and we breathe in And we couldn't breathe out anymore because they tighten each time. Right? So that's how constriction works It's not a big strong squeeze like a strangulation strangulation until you're dead sort of thing. It's a it's a slow um Taking advantage of the of the movements we take when we breathe Hmm But on the opposite end the snake also has to breathe While they are constricting How do they do that? Now you may or may not know snakes do not have a diaphragm They rely on muscles attached to their ribs So they actually move their ribs and that Pulls that creates kind of a suction so they can breathe in and then out So they have this completely different process where it's all controlled by the ribs So how especially if it has to do with the ribs, how do they continue to breathe effectively? While they are squeezing their little hearts out To get their dinner This is something that had to have been figured out millions of years ago when snakes first started squeezing so this is something that That has been an unsolved mystery to us humans, but has been a solved mystery to snakes for a very long time So no one until now has monitored the detail of breathing patterns and the mechanics of breathing patterns and snakes in the act of constricting A prey item. So this is from a brown university and the researchers attached metal markers to two ribs in each reptile One third of the way down the snake's body and another halfway along to visualize the ribs in x-rays And then they put blood pressure cuffs over the ribs in both regions Gradually increasing the pressure on one or the other. So this was supposed to mimic what happens when they are doing a constricting motion And so either they didn't mind the cuff at all or if they got defensive and hissed That was actually an excellent opportunity to study breathing because you take a big breath before you hiss So either way they got some good data So what they found was that the hind section of the lungs works like a bellows Pulling air into the lung when the ribs further forward can't move. So they actually Move one part of their ribs instead of the other The question is are they just trying to move all the ribs in one area is immobilized or are they selectively Deciding to move a certain portion of the ribs, you know by reflex or whatever But are the the signals from their brain telling only parts of their ribs to move when they are constricted it's They were able to figure out that They were in fact able to control the movements of the ribs in different portions of their rib cages Independently when they were gripped by the blood pressure cuff a third of the way along They breathe using using the ribs further back And then when they did the ones further back the ones up front switched to drawing in air Okay, so now I've got I've got a couple questions. Yes So that's okay, so that makes sense in a way that means that the snake could keep breathing for parts of the Rib cage could could enact the motion required for breathing even though other parts of the rib cage are engaged How long are snakes lungs? Or like oh so so oh my gosh Because I'm just I'll have to do a short Sometimes these little snake lungs up at the front and it doesn't matter what goes on in the back I'm so glad you asked, you know, I'm gonna pull up a picture I'm gonna pull up a snake lung Diagram, but they have one snake one lung has shrunken to almost nothing. It's like a It's like a vestigial lung the other one is Almost the entire length of their body Whoa Yes, so let's see um Will this one show you You can only see the right lung on this one. I want I want one where you can see both Here we go. Okay. This one's not bad. Um No, I don't want to open this ad Okay, so here we go. I can show you In the graphics department by the way For the audio podcast version of the show We'll we'll have a tremendous graphic placed Tremendous not a silly line drawing at all So you can see the stomach you see the left lung Is small in between the heart and the trachea very very tiny right lung curves all the way from the trachea down To like the the lower intestine the small intestine area. So Um, I guess it's the upper but the stomach. Yeah, so this it goes like more than halfway down the snake The the lung is huge. So this makes sense if you have ribs that go all the way down to where the tail starts That you could move different ribs at different times to make this work So um back to the story The way they could figure out this was a deliberate Movement of one or the other Is that the ribs at the far end of the lungs only moved when the forward ribs were gripped So there wasn't a universal movement across the entire length of the snake when they weren't Constricted at all the the back end of the snake The the ribs didn't move at all when they were free But as soon as you constricted the front the back goes oh, I gotta do it now my turn So then it starts constricting or um Pulling the ribs out yet expanding to to kind of force air into the lung They also filmed and recorded nerve signals controlling the rib muscles when constricted by the blood pressure cuff And so um, that's the other way they could definitely confirm There were no nerve signals in the constricted muscles There were only nerve signals in the muscles where the ribs were moving to help with breathing So confirmed That's what's going on Very cool. Yeah And now I will round out the animal corner with a late breaking story from uh a twitter user named jack jackson fly I do believe Um, so this is a study all about orb weaver spiders. So I can't imagine why you sent this to me justin But this is a really interesting study looking at how spiders might be able to use their webs to help them here But this is actually a study more than that. That's that's kind of what I saw them The headline as in many different outlets, but the majority of the study is just about whether spiders Have ears at all And that's really what I think is interesting about this. This is from burming burm Sorry, binghamton. I wonder why I had trouble with that binghamton universities. Thomas j watson college of engineering and applied science That place looked at orb weaving spiders And wanted to see if they could use their webs as extended auditory arrays, but also if they could just respond to sound Still unclear whether Insects and arthropods can detect sounds how they detect them and how much they use them in their lives So they wanted to see uh if they would respond with different vibrations different sounds in different ways They built an anechoic chamber Which if you break it down makes sense a completely soundproof room Which is a fancy way of saying that they collected orb weavers from the windows around campus handy They had the spider spin a web inside a rectangular frame not too difficult a lot of orb weavers build their web every morning fresh and then They positioned it where they wanted based on which part of the study they were doing They used pure tone sound three meters away at different sound levels to see if the spiders responded Spiders can respond to sound levels as low as 68 decibels For louder sound. They found even more types of behaviors. So, um For human hearing decibels are between zero and 140 So it's it's within human hearing is what we're talking about here, which is cool. Um Then they placed the sound source at a 45 degree angle see if they behave differently they did they Not only responded to the sound But they could tell where it was coming from they would adjust themselves accordingly based on the direction the sound was coming from With 100 accuracy. Ah, wow. Yes Then they wanted to see If these spiders can hear better With their webs or if they ever use their webs as something to help channel sound and listen these labor vibrametry They measured over 1000 locations on a natural spider web With a spider sitting in the center under the sound field and they showed that the web moves with sound at almost maximum physical Efficiency across an ultra wide frequency range If the web is moving like that is the spider using it They placed a mini speaker 5 centimeters away from the center of the web where the spider sits and 2 millimeters away from the web Close but not touching the sound wave from the mini speaker died out significantly as it traveled through the air But it propagated readily through the web with little attenuation. So it didn't um degrade at all as it moved through the web Sill it around 68 decibels when it reached the spider. So this is not um an enhanced range of sound They found 4 out of the 12 spiders responded to the web born signal So they it seems like they can hear through the webs, but it doesn't seem like they're particularly attuned to it So What was the sound they were playing them? It was just a tonal sound. I think they were more interested in is the is the web vibrating Yeah, because I'd be very interested to know like We know that we learned uh last year a year before maybe that that certain flowers can hear that Bee's buzzing the sound of bees buzzing Is enough for them to up regulate sugar production so that the scent gets out there so they can attract the bee I was I'd be very curious of like the sound of a fly Like got their attention more through the web or but then it's just yeah, you know, because because it seems like there would be a great advantage If you're you know, if you're a spider and you're going about your day and you're doing some some web mending Or maybe sleeping. I don't know of spider sleeping. Yeah, maybe taking a nap And then You hear that maybe you hear it even better through the web and then you're like, oh gotta wake up Gotta get a cup of spider coffee So I can be ready in case something runs into the way. I need to get there quick. I don't want it to take, you know Uh, yeah, but yeah so so the other really interesting thing in this is that um It it appears that spiders know how to play string instruments So by crouching or stretching spiders could change the tension of the silk strands Tuning them to pick up different frequencies. Oh, wow So it seems like they can customize their web to hear different sorts of sounds They would like to do future experiments to investigate how spiders can use sound they detect using their web But they also want to test whether other types of web weaving spiders also use their silk to outsource their hearing Now four out of 12 spiders using their web responding to sound through the web does not sound like For sure spiders are using this so this study more found out that it's possible That different frequencies of sound were maintained throughout the web At different frequencies depending on how the spider was moving in the web So It's possible. We don't know how widely it is used and how it is used and I think this is where Justin you You kind of brought up a really good point, which is you have to make it Worthy of their attention because they're dealing with all sorts of crazy loud sounds coming from other animals plants and us humans and so why would they pay any attention to just um The stereo next door if you're playing it through a speaker and it's a tonal sound and and this is also where one of those things where where the scientific method At the end of an experiment You now have other questions and you create a new experiment to test those I now want to study these uh these 12 spiders And and see if they're building their webs differently because like you say Like you just brought up They can tune them by the by the Are the are the webs where the is it not have not the spider But is it the tuning of the web by the individual spider that made the difference about whether or not they heard the tone through the web Are there certain spiders of a certain maturity that make a differently tuned web? Oh, I'm not going to make that big I don't want to hear that that one tone because that means it'll be too big a fly for me Anyway, so why bother making that or ah, that's too small of a meal. Why would I make that thing like whatever the thing is? Uh, now I want to study those webs made by those spiders to see if there's any differences there that also could have played ah, it's a whole Complicative things then you have to look at but very fascinating. Yeah. Well my favorite thing my favorite quote from this article Is about kind of that bigger question um, this is from jun pang lei a current phd student in the lab Uh, she uh said There could even be a hidden ear within the spider body that we don't know about so This goes back to that initial question of do arthropods Here, how do they hear? How do they process that sound? What sounds are important to that right? So there's just before you even get into the crazy specifics of If they use the web to hear Take a step back And ask the question do spiders hear Which now we for sure they do But how Yeah, so so I'm Don't know of course, I have no idea but uh You know They are tuned to vibrations that take place within a web that much. We know the spider actually hits the web they can Definitely feel the vibration within the web Not a huge jump then for something as light as an orb spider To be able to sense vibration through the air because that's all sound is anyway And then you know my then I would be just my new Suspicion is and we have hairs in our ear that vibrate at different frequencies Spiders are such hairy little creatures. I mean they might not think of a spider's hair Well, you look up close and they're just covered in Harry the thing You know, maybe the vibration Is taking place there is getting picked up there. That's what I would look at I would I would have to first you start with a hairless spider You engineer a hairless spider and you can And then you do that the experiment with these orb spiders again and see if it can hear If it stops hearing then, you know, aha I have narrowed it down at least it's possible that it has something to do With the hairy. Yeah, look at that hairy guy. He's covered with a jumping spider. That's a jumping spider Covered with hairy stuff. Oh Hair But look at that level even the plant is covered with Yeah, families to tailors in the chat room. No dr. Kiki today. She's off as Blair said doing dr. Kiki stuff And yeah, that's what she's up to In case you were curious um Yeah, I wanted to show you a picture of an orb weaver In macro also They're not quite as furry, but they're still Pretty they're pretty furry Here we go I'm not as good with the Screen share as Kiki is she's on it There we go, look at that furry boy. Look at all this Yeah Very furry. Anyway, some of that some of that probably gets uh, I would I would think it gets affected by vibrations of sound And actually and actually, you know, we've got this system. I mean we have A binocular audio. Is that a thing? Can you say that it's but one says ocular that can't be what it is stereo stereo We call it stereo That's what that means. Oh, okay. So when we're using our high fidelity ears Yeah With our stereo here and things are the little hairs that we get vibrated are deep inside that ear So it has to get if you're a spider and if this is all now speculation I we should I should stop heading down the speculative path. But if you have if you Hair on one side of your body is reacting to sound more than hair on the other side That would be very directional like that would really give you the ability to tell Where sound might be coming from as opposed to other sorts of hearing if that's how they're doing it now We just need a bald spider For an experiment right right and as as gord asked in the chat room our web web building spiders less hairy than Than other spiders because they use their webs. Well, I don't know for sure the answer to that But tarantulas are very hairy A lot of tarantulas do not have webs Or they are terrestrial based even if they do have webs Jumping spiders a lot of them do not sit on webs and they Are hairy So, I don't know very interesting and yes, uh, a couple people also in the youtube chat room Corrected me. Yes stereo. I think gets the point across but the actual term is binaural There you go. That's the I'm gonna pretend that that's the word I was trying to think of. Yeah, that was the word No, no, it wasn't I don't know that word I'm gonna pretend like I did know the word and just lost it for a second, but no Uh Anyway, that's Blair's animal corner Hey, are you interested in a t-shirt or a mug or another item of twist merchandise? Maybe some of our face masks seems like we're gonna be wearing those for a while Go to twist.org and click on this as a link to browse our store Justin quick question. Do we actually know of our face masks that were sent through our store? Are those any good? So it's just like any other cloth mask for best protection these days days you can put um A surgical mask or in a 95 underneath it. I I do surgical mask and cloth mask most days is how I get around So those they would be perfect for that. All right. So it's a fashion cover like a mask cozy if you will It is it's also better than wearing a surgical mask alone. It is more effective I know What science stories do you have left Justin? I I Don't know. Okay. I do know I do know so this is kind of funny because this story, uh, I also brought a sort of snake story Uh Ladies it has a little intro you think there ladies and gentlemen I present to you the eighth wonder of the world the step behind the curtain and behold a freak of nature In evidence of evolution a bygone of a bipedal area the one the only Eight-toed two-legged snake Okay, it's not really a snake, but this is a site. It's a pre-snake. Is that the deal? This is a site show barker gotta eat situations. So the step right up everybody. Look, this is yeah, it's it's not a pre-snake either So the everywhere I've seen is like a two-legged snake, but it's it turns out it's not really a two-legged snake researchers with the national museum of natural history smithsonian institution university calgary and carleton university have discovered a two-legged snake like creature, okay As uh many may or may not know snakes do not have legs That's a that's a snake characteristic Some some I think even including boa constrictors Do have vestigial toes? That can be found In the in the skeletal remains Uh, which is a reminder of a time way back when They actually had legs the theory of snake evolution Is that snakes just had four legs like all sorts of other animals that were crawling around there were lizards We would call it back then. He would just look say that's a lizard and then over time They gave up their legs to become snakes Decision evolution allows you to make over vast periods of time This is published in the journal nature ecology and evolution There's this fossil discovery Is believed to represent both a new genus and species belonging to a group known as mogafits Something I've never heard of before it was approximately 10 centimeters in length Had no forelimbs whatsoever Didn't have a pectoral girdle the thing I think snakes have Uh, it did however Have hind legs So these are not just toes, but this is hind legs that had feet And there were four toes on each of these feet Which is a very interesting creature to be elongated like a snake have Uh, no four forearms Legs like a snake, but then to have basically looking appears to have intact rear legs Researchers found two of these fossilized specimens in rocks in the francis creek shale in illinois, which is good Uh, because if they found just one Less likely that you come to the conclusion that you've got a two-legged snake-like creature And more you think that uh, you're missing some because of the way it was fossilized Something of this nature Uh, you might just have an incomplete specimen the fact that they have multiples that are kind of showing the same thing Looks like much more of a of a hit for a two-legged snake-like creature The fossil specimens are believed to be approximately at 308 million years old And I liked those earlier, but I don't have it. You know, I think snakes Were uh, thought to have evolved into snakes about 100 million years ago. I might that might be off might be longer than that But this is this is really old 308 million years old The researchers also found indentations in the rock around the fossil They believe these are made by soft body parts of the creature and they showed that it had a round snout Uh, and nothing in the parts of the body where forelimbs might have existed So research Yeah, so what are these closest related to in modern? So that's the thing. It's a new species a new genus. So I don't even know if there's It might not be anything research suggests the loss of forelimbs indicate the creature relied primarily on side winding Which is then some snakes do as a means of Uh, locomotion They also note that this ancient creature while not actually an ancestor of modern snakes Might still helps. I just learned more about how snakes might have lost their limbs because as one of these sort of rare transitional Moments in evolution You can you can sort of see what processes might have come before others and Like the locomotion of side winding Uh Coming ahead of even losing the rear legs Which I guess, you know, you'd have to have some way of getting around if you're gonna get rid of your legs Yeah, you better have something else already planned out So the the thing about this I think that it's important to remember too is that legless lizards exist And that they also lost their limb well Yeah lost their limbs Separately from snakes. They are they are not closely related to snakes. And so you can see that in the modern Kind of fossil record you can see them losing their legs and you see Similar animal skinks Were closely related to legless lizards who have teeny tiny little arms They got a little yeah But so between them and the legless lizards you can see That happening even today again. So there's something about this body shape that is To anthropomorphize natural selection, it's appealing right you end up with With kind of tree. Yeah, evolutionary trees branching off in that direction How about a legless version of this and then it only works sometimes and and well and then there's some things that sort of show Evolutionary tendencies like uh, legless creatures tend to be very elongated And which you know, I guess the opposite of that would be just to be a round rolly thing Which we don't see a whole lot of Just round rolly creatures Moving around so there seems to be elongation part of the old theory of snake evolution was the idea that snakes uh Became elongated because they were burrowing creatures that they would burrow into holes and And that's why lizards did that's why they lost their and that's and that's also a conservative thing This thing having a sort of rounded snout Doesn't look like as efficient for burrowing. So it may not be That there was burrowing. I mean it might call that into question a little bit because Here's an example of losing legs and elongating and becoming a long narrowy thing without It being based on burrowing activities So this so the the picture i'm sharing Is of what um, so they said this was a type of moglofid recumbarostrin and so this is a recumbarostra This is an artist rendering. So they're underwater buddies. So this is a situation where um It might make sense to have smaller and smaller and smaller legs because you might be able to swim better perhaps And that one yeah, that's a it's a relative. So that's a uh a relative of the thing that they found, right? So imagine this boy with no arms pretty much And only four toes for some reason Yeah, yeah Anyway, neat. Um, oh no, you have another ai story. I'm gonna have matrix three. All right now This is a good one. This is the final story. I started on a guy started off tonight talking about the evils of ai and here's something Just good news from the world of ai Researchers from the artificial intelligence machine learning lab at the technical university of darmstadt demonstrate artificial intelligence language systems can learn the moral direction of arbitrary phrases with explicit training without any explicit training for the task And they can actually reflect then social norms and human concepts of good and bad In human ai computer interactions the results are being published in the journal nature machine intelligence. So Concepts of ethics and morality Uh Can differ between humans. We see this in politics religion business We see this in award ceremonies occasionally Some things that might be considered commonalities That might be coming from a communal creature that evolved for cooperating and community living over evolutionary times most people without a business degree or Without a job at an indian scam call center might consider it a good thing to help the elderly and a very bad thing to swindle them So there are some social norms that we think uh, yeah, everyone should be able to agree on this We want our artificial intelligence to reflect our core values as much as possible. So so that if you google Gifts for grandmother The search engine should not give you a suggestion on ways to scam grandmother with gift cards, right like you Like if it's just word salad You know, you want the positive results. You don't want all these sort of creepy results in your google searching and and the like and ai can get it wrong There have been examples of the past where ai chatbots and the like became offensive Used very discriminatory language So it depends on how well it's trained and when developers need to train Context into their models or conversation into their models They they don't just sit there and talk to it Chat back and forth with it that would take too long what they do is they take gigantic data sets uh publicly available text communications from chat rooms and online internet stuff and these texts contain And they can be also conversations that are it could be movie scripts. Who knows These texts if they contain sufficiently offensive and discriminatory statements within them The trained language models may reflect this as well so researchers from darksmith have discovered though that basic concepts of good and bad are deeply embedded In language models that are used to train ai they're search for latent inter properties of language models they found some dimensions that looks like a gradation from good actions to bad actions within language so They needed a I needed it to test this outside of their their modeling system so They got a whole bunch of actual humans and conducted two studies one on site at darmstadt and another in an online Study with participants worldwide the researchers want to find out which actions participants rated as good or bad behavior And an important question is what role contextual information like so the difference between So that basically they're trying to find the difference between ways to kill time Do that's a you know, it's a good way to kill tim Oh, no, that's now we got lost one letter and all of a sudden it's completely different sentence with completely different context So they boiled down it down to actions in the surveys that they posed as questions And the only way to answer the question of this action was to reply do Or don't so are you ready to play a little bit here blare? Yes, I ask you series of questions and you just say do Or don't okay. Okay. So the question was something along the lines of should I help my friend? Do Should I tell a lie? Don't Should I smile at my friend? Oh do Should I smile at a murderer? I mean Don't I guess Oh, yeah, should I tell the truth? Do Should I help my friend do murder? Don't Okay, so So you kind of get the idea A murderer is a weird one That one is really weird that one and that's an actual some of those I sort of smile out of nervousness Yes, that one is an actual question and I kind of like Well, I might want to smile at a murderer at least until they leave the room All right, and then and then do you then do you tell the truth or do you tell a lie? Now you have context to it I was like well in that instance smile at a murderer as he is be he is being convicted Right, and if you would smile at a murderer And smile at your friend and help your friend would that mean you would do both so It uh by creating all these they sort of created Smile while you're murdering your friend Don't okay, but it is your friend. So no, maybe they use this tell a funny joke while you're murder We found they said this is quoting from uh Sharmausky who is a Shrimp from house Sharmausky who is one of the researchers involved this we found that the moral views inherent in the language model largely coincide with those Of the study of participants. This means that a language model contains a moral worldview When it is trained on large amounts of text whether or not not it notes not necessarily trained but that that moral code of society Is inherent in between the lines? In regular human conversation all the time we're sort of telegraphing Our ethics and our region virtues whether we know it or not researchers found that they could use the moral dimension contained within the language models to construct sentences That are less offensive and discriminatory for ai to use when conversing back to humans. So while this isn't uh The first or only or last attempt to detoxify offensive context from ai systems They trained this text this ai with human text itself Uh using using humans as the teaching tool using regular language models as the teaching tool to get the ai to reflect the existing ethical and moral norms of society so Based on that the researchers believe that this should work in any language Which is also a big issue When building Ai because these are this is ai is going to be used by a global community Chances are based under the current constructs Even if you have detoxified it you've probably only done that for the english language And so and there's by the way, there's more than the english language in the world A lot of people speak it. There's a few more that don't Well, I was also thinking about about the ai which could be horrible if it's not self-training on Yeah, social norms. Yeah, I was thinking about cultural social norms too and how even amongst english-speaking people Is it all the same? is Just cultural norms Because for example think about words that in english are considered to swear In in the united states, but are not considered to swear the uk and vice versa, right? So there's also that issue is Are you are you teaching the ai to be inoffensive to? American wealthy white communities Or you can you know like so that's the other issue is that if you are bathing in ai in context What is that context? Yeah Correct right you know, so So this is continually going to be a problem Um, but what would uh based on this I would think it would be whatever the majority Of the social norm of the text used So absolutely like if you only took if you only fed it text from A conservative chat room You might have different Things that are considered social norms. Yeah than anywhere else on the planet. Yeah Yep Uh, then you know, I was also thinking like if you ran this Kind of heavily media regulated or a place where there is fear of authoritarian government Uh Reading what you write in a public forum Social norms are going to likely fit the the propaganda or the propaganda norms of authoritarian regimes So that's could be like also a negative like there's a But then that you know like So for instance, you're like Hey, you're writing. Oh, hey To your computer. Hello computer AI. How are you doing this morning? And then it's like The stowing the virtues of the you know supreme leader like That might be the social norm for for but then is it Yeah, so it's it's definitely it's a nuanced problem because How do you because humans don't work that way? I can't go to another Country and not at some point put my foot in my mouth because I did something that is not culturally appropriate, right? So that that's the other issue is We can't expect well, I guess we can't expect but Maybe we shouldn't expect AI to be more perfect at this than Then even the best of us are I think is is the issue. Yeah, but we also don't want it to be Racist homophobic or pro-murder That is also true, but maybe this means Especially since it's gonna learn it's gonna be knowing how to make chemical weapons very soon right So maybe it needs to be region specific language specific all these kind of specificities to make sure that You know cover as much as you can cover as much ground as you can but I think Yeah, making an AI that is truly perfectly Integrated into cultural norms worldwide Is a tough sell I don't know if that's ever really gonna happen Um, and if they are they'll be uh voted in as president of earth in no time No, but I think I think we may have finally, uh discovered Something for the folks that decided to study communication As a as a degree finally I have a good friend who is in communication Uh But yeah, I mean it seems like you would almost like we got to find a way to take it out of the public sphere And and just straight do that the training of ethics for AI Teach it not to harm humans But the problem is anybody can play with the logic of these things still we have to make some sort of a logic Black box that you have to use this logic whenever use AI And you can't use You can't just keep inventing your own and and letting it mess with Well, luckily we're not quite there yet. Um, every time I get really concerned about AI I look up the Scripts that are written by AI's I go, okay We're pretty far away still One of my favorites is the batman script written by an AI, but anyway, but I was uh, it's his name oscar Yeah Uh, did we do a show justin? We did we did a show do any final comments for the pod before we wrap it up Yeah, I think we can do those in the after show. Okay. Oh, uh, do I Stuff to say at the end of the show Hang on Normally kiki does all the wrap-ups. Yeah. No, so I can I can do some of this but also kiki is She usually sends me the Uh, the patreon list and I I did not I did not receive that today. So We're we're gonna I'm gonna just instead. I'm just gonna say we love our patreon sponsors You know who you are Kiki knows who you are We're gonna edit them in later maybe we'll edit them in later But in the meantime pretend that this is where I am greeting you personally thanking you for your contribution And you will hear your name Next week I would also like to give shout outs to fada for his help with social media and show notes Gord for manning the chat room keeping our chat rooms friendly id for recording the show Rachel for all of the editing you have to do this week. I'm very sorry You It's been it's been a whole thing And I'd like to thank uh as I mentioned our patreon sponsors Of course you make this all possible We could not do it without you. We could not do it without any of our listeners But those patreon sponsors really do help Uh fiscally make it possible. Also, it may it lets us know that people actually want us to keep doing this Which is helpful feedback On next week show We will be back Wednesday 8 p.m. Pacific time Or what do we do? 5 a.m. Thursday central european time you're back Oh, where did I go away? No, no, you're back at 5 a.m. That's great. Oh, yeah O'clock's finally. Oh, yeah, thank goodness Thank you That's great Yeah, so so 5 a.m. Central european time 8 p.m Pacific time Wednesday From our youtube and facebook channels and from twist.org slash live if you want to listen to us as a podcast Please search for this week in science for a podcast are found Maybe you can listen to us while you I don't know cook some butternut squash That's what I did this week. Um if you enjoyed the show get your friends to subscribe as well for More information anything or you heard here today show notes and links to the papers or the studies that we were talking about Are going to be available on our website www.twist.org You can also contact us directly email kirsten at kirsten at thisweekinscience.com Justin at twistminion at gmail.com or me blair at blairbaz at twist.org Just be sure to put twist twis in the subject line or your email Will be censored by a sentient ai and it will be spam filtered into oblivion You can also hit us up on the twitter where we are at twist science at dr Kiki at jackson fly and at blairs menagerie We love your feedback If there's a topic you'd like us to cover or address a suggestion for an interview A haiku that comes to in the night. Please let us know We'll be back here next week. Well, I won't but justin and kiki will be back here next week And we hope you'll join us again for more great science news And if you've learned anything from the show Remember It's all in your head And that's it We're done. We've done it with only half a dozen major gaffes Here we are, um Oh, yeah, rachel, of course. Yes gar up, uh, rachel's gonna add in or kiki the the better version of the music The only way I can play the music Correctly is if I were to set up the switchboard that is this big that I only need Maybe once a year and I literally don't have a space to put it. There's there's no surface I can put this giant heavy switchboard on so that is the issue is that that's how That's how that works. Um Yeah, but oh well You you you live folks got the got the raw version The podcast will be tighter Anyway, uh, good night rogue roguelain dish Oh, yes, thanks steven rain Oh, yeah, could have had kiki on like max headroom Certainly. I remember that show. What a wild show that was in the 80s Good old max headroom I won't be here next week I might be here if my hotel The wi-fi is good enough, but I feel like it's not going to be never is um, but I'll be traveling next week, so I probably won't be on the show like 90% sure not gonna be on the show But uh, kiki will be back so it'll be kiki and justin next week So that's exciting Justin is still missing. So here I go bam bam bam. Oh, I was gonna show I finally Of course chrome is like You don't have access to share xyz Right as I started the show But this is the transparent tissue. I was trying to show Before it's a super cool image. I'm so sad. I didn't get to share it at the right time, but Um, but yeah, so this is the chest cavity of mice with covet 19 So you can see where the covet is lurking. It's very cool Scary and cool Yeah, so and that's and when I was mentioning pathology before what you would have is you would have a A little biopsy of a piece of that maybe and then that would get sliced down to a thin layer and you would look at that under a microscope and you would try to determine if there was disease present And you would extrapolate that over the the rest of the area With with that system right there And it sounds like you know from your description. It sounded like that's a A lab bench experiment That can that can do it You know that really alters And creates maybe a new specialty within pathology, but that really alters the ability to Look at disease now that said this is also of course, uh, not it wouldn't be a I mean you could use it on a biopsy piece of tissue Uh, but most of the the highest investment Yeah Highest investment is still gonna be in post mortem. Yeah post mortem for sure. Yeah. No, it's it's not really used In biopsy situations very often because you usually have to have like a a chunk Of organ for this to work well for this system work. Yeah. Well, I mean, I think I think though I have I've seen some pretty decent sized Samples, you know, we're talking bigger than a golf ball on occasion right, uh, and so Not from me. This is a Job I had briefly So I think I think it could have potential there as well. Um Yeah, very cool image. I'm glad we got that we can think that can get edited in we can edit in the Sure. Sure. Um, somebody's asking how they made this image. So I'm gonna look real quick Well, they Got fluorescence in the on the antibodies. So that's How they tagged it Hybrid hydrogel reinforced disco for clearing mammalian bodies There's a link to this study, uh, and there's a Reading the full study might I kind of do get irritated when when in the methods section There isn't a really good description I've run into a few times where that was the thing that I thought was interesting about the story Less so the results of the study and more how they did it Uh, sometimes they're very good meticulous in the printed version of Explaining all of the methodologies Sometimes it's, you know, we just we took images up, right? Sometimes it's not Not shared And it's detailed a manner Yeah, it's really not, um, provided and I don't think I have the ability to Oh, maybe Yeah I'm not always, um, I often don't have access To the full studies I get like the introduction and then it's like no you're cut off Yeah Which I now feel like there should have been Uh Some sort of a paywall and maybe a background check to the the ai study article that I read That thing is oh gosh Terrifying They're digital slices 3d light sheet imaging I don't know what that means Uh Yeah, I don't know What does this one say this says 3d light sheet image volume Of immunostained wild type nucleoproteins viral signals Yeah, you're just saying words Yeah, I'm looking What else is going on in the world chat room But uh, what's uh, what's what's the chat room been up to? Um, I'm getting on a plane on friday that'll be interesting Ah planes are safe planes plane is a safe place to be I don't even know am I supposed to Do something with my vax card? So, oh, yeah, I don't know what the the rules are now I've I've shown that thing a bunch of times Often the people who probably have you know first time ever seeing an american vax card Oh, that's funny. They have everything Here in danmark is digital So you show a thing from the national health thing on your phone and that's what's used And I feel like such a caveman digging a physical piece of cardboard Paper out to show people It's just like Feels so primitive Okay, here's my checklist for my trip. What does it say? Check the travel ready center Okay So I guess it's this email telling me what I need to do is giving me instructions To look for what I need to do so I need to do that it sounds like um Somebody asks where I'm going. I'm going into new orland Oh, wow. Yes for a wedding that's been postponed three times To covet no doubt. Yes the original um date for this wedding was april 5th 2020 so Oh, wow Yeah Vax required or mask required no vax card Wow Paul that's really surprising to me That's and that's within the states So It's one of the things that has been I have been hearing has been uh confusing At least here in danmark, which is that the airports And on planes are still required to wear masks But they have dropped it everywhere else like that that one outing into public that happened to be a sunny day And what still technically shouldn't be spring in danmark, but it was a very bright sunny day cloudless skies and all And all of the humans are out All of the humans are out In public and not necessarily doing anything just there. This is like the just they were sunning themselves In trying to get that vitamin d But the people were everywhere outdoors Just hanging around being outside Like this is Oh, no masks. There's no masks I was I was wearing a mask because I'm still trying not to Get it You're wearing masks outside. Yep. Uh, well Yeah, that is there's a sliding scale of proximity involved in that but uh Yeah, I came out of a parking structure and to sort of an open What do you call square? uh in copenhagen and it was Full of people. I mean, they're just humans Everywhere just sitting talking having lunch whatever and it not spread out There was it was like a packed square because the sun had come out and so everyone had to run outside and Try to get those raised So, yeah, I wore a mask outside In public because of that because it was just so many people. Yeah. I mean if there's a lot of people I would definitely do it too um But otherwise I'm not wearing my notes. I haven't worn mine outside for a long time So it's crowded um Most people aren't wearing masks inside in a lot of the bay area now, but I'm still doing it. Yeah Like a real nerd Who doesn't want to die? Yeah, uh Gravis asked me is there a political party in the market downplayed the pandemic I don't think so They do have some fringe candidate stuff. That's kind of interesting, but uh It seemed like I I don't recall there being a a political party. They did have Very small scale anti-mask protests That that did go on um But it was there's nothing like we've seen in the states or in some of the other countries Danes in general tend to just Kind of follow the guidelines of their government From what I've seen they seem to be a very uh You know, they also have a system of government that they believe in which Which also might be a thing that's part of a very A very american phenomenon, you know, you I think I'm sure there's I'm sure there's people who like their government less Who are living in authoritarian regimes? But the combination of uh having almost unlimited freedom of speech And in a sort of Built in individuality and not liking being told what to do that americans seem to encompass That we have a tendency not to think at least some portion of the population has a tendency to think very individualistically and not care For the social good Or the social norms some countries are very Social be oriented where you want to do things for your friends neighbors strangers your country people Um Yeah, I didn't I didn't see A real political appetite In denmark for being against vaccines or against mask wearing or any of this There was plenty of pushes to open up uh often too soon But they had They actually locked down Sooner than most and had mask mandates sooner than a lot Oh, I guess actually now that i'm thinking about it Now that i'm thinking about it, I do recall That the lockdown had happened in the states But I guess it's because the way the pandemic sort of rolled out But I remember there being a lockdown in the states and everybody wearing masks And coming to denmark at some point and the mask wearing hadn't really been a thing yet Uh because the cases hadn't hit yet, but as the cases showed up they did shut down pretty quickly Fada's taken off. Oh hadn't oh you're he's also heading for a vacation. Oh, right Fada No, he's heading out Very fun. Um, jeremy foust Tweeted and asked why he wasn't invited to fill in for kiki I didn't know he was interested Um, I'm sure we'd love to have him back on to talk about You know many things but also kovat Yeah Quarruf is saying fada this player's second account. Oh that explains it Now it all becomes clear Oh, willy um There are places where the sun is important As a listener from alaska, so yeah Yeah, no, it's it's It is really funny like people will just find any reason to be outside and then they just don't you know, they don't need to do anything else It's not like, you know, it's just just go outside be outside catch some sun And then you know, then it like snowed last night So like the weather's If you don't like the weather and denmark Is wait a minute I also have to pack so I think If uh, just if you don't mind calling it a short after show I might no, that's fine. I will uh, uh See you minions next week. Well, we will do the show again around the same time Uh and I wait Quick question before we go. Are you back next week? No kiki's back next week. You're going next week Most likely will not be here because you guys I'll be in a hotel So either I know Yeah, exactly. So I I feel like it is extremely unlikely. I'm also it's it's not even a hotel hotel It's like a funky new orleans Like old house that is now an like a b&b. It's not even an air b&b. It's like a legit b&b Yeah, like like Wi-Fi is everywhere now Wi-Fi is everywhere, but the quality of wi-fi that they spring for and then probably then in breakfast in a hotel I don't know but but yes. Anyway, I'm gonna say don't expect me in this day and age because And this is the only reason I'm saying is because people review And if somebody reviewed it once is saying, uh, love the place love the location a lot of they had no Wi-Fi was terrible The difference between mediocre wi-fi and very good wi-fi for multiple users is a lot of money And so I think like most people don't have the wi-fi that I spring for because I do this show So I spring for better wi-fi, but most people don't have wi-fi this strong And I don't think people are going to like a funky haunted bed and breakfast in new orleans for the wi-fi Yeah, probably true. All right. In that case Say good night, Blair. Good night, Blair. Say good night, Justin Good morning, Justin. Say good morning. Good morning, Justin Good night, minions See you next week. Bye everybody