 Hi, here we go we're gonna kick off this open ed tech workshop and I'm having a severe deja vu to last year How many people here weren't here last year actually in the same room at the same event Okay, about half of us those other half were here or they didn't hear me so We it's very this is a very open flexible thing It's a workshop and the the point of the workshop is to for us to be thinking around open ed tech the I I'm hoping by the end you'll have a better understanding of what it is all about Why I keep talking about this open ed tech brand instead of the Moodle brand But it is all very related and very connected and And also you'll have an understanding of how you can hopefully be part of it going forward So it won't won't just be today. It'll be everything afterward and You'll also have I really it won't be just one person speaking We're gonna have all a lot of us all speaking. So, you know be ready to participate here the the rough outline That I have for today is I will start off with a bit of news So I'll give you a bit of an update all the things that have been happening in the past Month or years months and years Regarding open ed tech as an association as an organization I'll Show you all of the places online where things are And give you a bit of an introduction to where things are and what's happening. I I Have two main activities after that the first one Well, the big one is I want us to brainstorm Around this concept of an AI assistant a learning assistant or a life assistant And that's all just that one It's there are many many issues around it and we're going to brainstorm the issues and then we're going to brainstorm the solutions and try and push Our thinking forward around that concept Hopefully we'll get through a lot of that in the first half maybe the second we've got two halves here We've got two hours roughly with a break in the middle for lunch So is it lunch or coffee a coffee break? So we'll have a little break. I'm sure your brains won't take a break, but you know bodies can and Then I want to leave a decent amount of time at the end for you to present things so Anything that's related with open education technology if you have a project that you think is related Is some open education technology? Maybe maybe you got frustrated because you couldn't do a presentation at the mood you missed a slot And you just want to quickly show something or tell us about something. I'll probably give everyone like Three minutes to five minutes, maybe make sure it's relevant but you can come up here and Show the group and let's have some sharing of initiatives because I know a lot of you are working on some really cool Projects and there just might not have been another place to talk about it yet or or maybe you just want to highlight Something that that's happening elsewhere. So that'll be really helpful come in other drama It's a drama from another mother and Still I'm still recovering from the adrenaline of the music jam last night. That was so much fun Anybody else who is that the music jam last night a few some of you that was great. Oh, that really worked I've been trying to have that thing for years and that really worked out so So yeah, that's the kind of structure so look if you are thinking of something that you think might be good for this group throughout the first part and And you can get it together It's for towards the end get that prepared slowly sort of in your mind in the background and we'll get to that So, let me let me start with some news I Mean does that sound okay? Let's have some feedback here. Is it all right? Yeah. All right. We'll take it as we go So the first news to say is that open ed tech is now an actual association Cheers for that I I I've been banging on about this whole concept for like literally years I mean the first meeting we had was in 2019 and we had a lot of people together around that and We we built a lot of the basic concepts and we built some of the relationships and and then there was a Pandemic and a lot of things are on hold and a lot of us all have full-time jobs or two So there's not a lot of time to dedicate to it. So it's been creeping along slowly And I've just been trying to push it very slowly over time And then we started to create an actual association because it's very key for this whole concept that there is a an Association that is separate from any particular software project Like Moodle can't do this on its own Because it's too associated with the Moodle World the Moodle product the Moodle brand and There's there would be very justifiable skepticism from people that this is just a Moodle thing and I it shouldn't be just a Moodle thing What we're talking about here is an open world of quality lifelong learning That means lots and lots of organizations Should feel some ownership and contribution to that And if it's led by one particular product it starts becoming too much of that thing That said Moodle is a avid supporter of this and will be part of this along with big blue button and many many other software projects and Institutions and more we'll get into that So we had to create the association and you know, I thought that wouldn't be very difficult I created an association before the Moodle users Association and That's gone for years since and that didn't take so long to set up honestly in Australia. It was pretty easy however, trying to set up one in Belgium from Australia is a world of hurt and It took a long time and it was expensive and there's a lot of lawyers and there's a lot of to and fro and a Lot of paperwork and anyway, we got there So the official founders of the open ed tech association is Fred and myself And we are the first directors of the board to get it going and it has a proper Rules of association and it's a completely independent functioning entity and It should be and it's not dependent on any persons if if As over time people will come in and out of that it'll run it. It's runs like a proper association. It'll have fully transparent and clear agenda So that was finally formed only a few months ago in June. I want to say June 6 or something like that It's in the news Right. So what we're looking at right now is the open ed tech site and that's been developing. It's a WordPress site You where we've been we keep working on the messaging and the the well content here and So you can find out here about the vision Now it's all obviously Completely it's developing and and so on at the moment. The only people who have time to work on these things is Myself in my spare time My and my two children I'm hiring my son Tommy you'll see Tommy do Guillaume surround He's 19 and he's one of the best programmers. I know already like he's just a full-on hacker He lives on github quite a bit and and he's just I mean he has other interests. He's 19 But he's picking it up very quickly and it's honestly not something I forced on him But he just kind of he's he gets it So he he's good at some things. He's working on our first project Ed you bought and Some other things and my daughter to he do Guillaume is to you I so you'll see her around and I'm actually paying them both out of my own pocket pretty much to to work on open ed tech and And help with research and the website and the content. So just when you see all these do Guillaume's is around you know So we have fun at home We have a little office and the three of us have an open ed tech Day every week and we we do this to start pushing along this little project. This isn't sustainable by the way This is just to get it going and I'll explain where we're going forward So the purpose of open ed tech is Very firstly to be a very strong voice for open education technology as a preferred platform for education, I mean imagine if Parents putting their kids into a school or a university were thinking, you know what I I I want this Institution to be an open ed tech institution is an institution that believes in open education technology As a kind of as a brand as a concept as a summary of the ideals and the principles around Education technology and so this is the voice and the brand for that concept to make it easier to talk about You can say open source, but it's more than open source. It is obviously open source, but it's This whole particular Framework that we that we're working on together So we do that through research and curating standards into a blueprint into a framework There are many standards in the world There are open standards. There are non-open standards, but even amongst the just the open standards There are lots of conflicting standards. There are lots of proposed standards. There are lots of De facto standards, so you know that the Moodle course backup is Become a sort of de facto standard of an online course backup, but it's not a not an approved standard Really, it's just kind of a standard because lots of people use that format And there are lots lots of standards bodies that try and certify things and work on things and there's We don't we don't want to always be building standards at Open Ed Tech or hardly ever We want to be saying we like that standard. We like that standard and here's the that's the curating part The second activity is about funding people need to survive and eat so we need to pay people to work on things and so we want to Source funding from around the world and there is you know, there's money everywhere It's just a matter of directing it towards these kind of activities And that's the two major activities And we actually have an address in in Belgium If you want to go and visit it you're welcome to it's a mailbox Eventually it will be an office So at the moment We have a lot of ways to get involved. Let me let me jump straight to the membership before we go any further. Oh I'm gonna have to Where's that there we are view the membership options, so we have these membership options here and And These are so you can just become a member for free We just call it a follower level and that just gives you more access to to things Edit rights on documents things like that If you're just a low-key supporter, it's about 200 euros a year An associate member is more for organizations who really want to support things It's a thousand euros and a full member which gives you More voting rights on the board the ability to be elected into board positions Is five thousand euros a year Now a plan is to through this mechanism to get enough to hire two People to work full-time here and the first two people to work full-time will be one will be more focused on the say marketing activities the Keeping up websites and articles and doing all the communications and things like that And the other one will be more will be focused on fundraising and project management so they'll be they'll be Connecting with fundraising opportunities say Erasmus European projects They'll probably be based in Brussels or able to go to Brussels quite often They're able to do all the schmoozing and interacting and and Promoting and talking and and you know someone with real abilities and reputation who can Do this work? It I mean it won't be me it needs to be someone who's just full-time focused on this and That person will be responsible to help push the next level of funding which will be You know working with like these large organizations and raising more money and as we develop the blueprints for what we want Some of it will be out there and already exists Some of it will still need to be developed and so the funding will be pushed towards those development projects to create the missing the missing gaps and That's how we work our way towards the framework That exists and is usable So that's the overall the overall plan. So we have thought the website now. There is a news. There is news here This is the main news channel There's lots of ways to follow us. There is also a newsletter to sign up We are connecting this this is like the source of truth of news Connecting this to mastodon so we have a mastodon Account you see it over here on the right Just this week wordpress announced a wordpress mastodon integration, which is terrific And the best thing about that is it's a two-way integration. So it uses activity pub The we don't have to hack it through RSS anymore It means that wordpress can publish directly to mastodon and any replies and activity that happens in mastodon Gets pushed back into wordpress as well as comments on on blogs So this is terrific news. We've been waiting for it for a while and it's just coming out now So we'll implement that very soon Is anybody here on mastodon already? Can I have some hands? Not that many of you get on to it get off this X BS And you know get on to some mastodon mastodon is absolutely Awesome. I'm loving it. It's Like you don't need to be connected to everything and every one and all the chaos and all the advertising and all the Algorithms and all that stuff going on most social media Mastodon is a lot more low-key and you will develop a smaller community of people who you actually like to listen to and Honestly, you know, you don't need to be spending more than an hour a day scrolling through Messages right you got a life to live. So okay, and that make that hour a quality hour right, so you're just actually getting news from sources and people who are interesting and It really does work. So one of the things we have already is an open ed tech Mastodon server Because mastodon is a federated distributed social platform through using activity pub if you want to get one of the highly coveted Addresses on our open ed tech server you have to be a member And I think that's fair It does cost us to run that server and if it had thousands of thousands of people on it It would cost us a lot more So we're just restricting that to members. So there's another reason to become a member The and yes, you can keep up through mastodon and it's quite quite efficient way to do it the The other thing that we have is a matrix Server as well. So there's an open ed tech Server on matrix and I'll just switch over to a Matrix client and you can just see what that looks like anybody here on matrix already I'm saying, oh, that's good. I'm saying like 20 25 people You know matrix is an alternative to Messaging any other messaging systems and it's like a really good alternative. It's also distributed federated people run their own servers You have an address like an email address and you can talk to anyone else on any other server just like you can email anyone else on any other server and Our open ed tech server has a bunch of rooms These are mostly working groups that are still in startup phase not that much activity But there is the general one at least you should get into the the general one is the general discussion There is there is this you can see here You probably don't see this one if you're just lobbing in this that's been it's a private one because in there We've been throwing a lot of like You know keys to access servers and Legal stuff so we have to have some sort of channel for that You will see in here a Edu bot playground Edge you bought is the first Little project from open ed tech and it's been how long about a year now It was Yeah, it was at least a year so it's a chat bot that you can talk to and it's using G it was using GPT 3 and then now it's on GPT 4 But we've built some software that helps make it a better participant in a group chat And we really want to make it a almost a facilitator. So we're just experimenting with concepts about having an AI member of a chat and It really is Already quite useful in my opinion. So some of the things it does for examples whenever anybody posts a link It it goes off gets the content of the link and summarizes it and posts it back Which means you're not always leaving the chat. You can just immediately sort of see what the link was about And that's like a very little feature, right? You can also give it certain personalities. You can give it instructions on how to behave and where we want to go with this in future is make it much more of a facilitator When you have 50 people in a chat and they don't know each other and they don't know what's going on The bot can say hey, it's a bit quiet in here. Let's play a game Let's play a word game and start getting people you know icebreaker a bit of an icebreaker game and start getting people active And help that sort of facilitation as an automatic feature of a chat room that it just sits there and and helps chat Be more efficient and more useful. So that's kind of exploring the educational possibilities in chat And so if you want to play with it, you can play with it in the edge you bought playground here Sorry, I've got my font size turned way up so you can read it So it's a bit sort of cramped here, but you can sort of see a lot of people are playing with the edge of bot in there So that's matrix and you know the beautiful thing is once you have a matrix account You can connect to so we use it at moodle HQ or as our official chat system all of our company talks through matrix We have I'm also say connected with we have the moodle community there if you didn't already realize big maybe you missed it somewhere else, but you know there are Community chats for moodle and so here's people posting some videos from the the music night last night And you can see some of the rooms here and there is a back channel for this very conference Moodle moot global So if you're not in there already and you want to find out about you know keep in touch with what's happening during this conference Get to this moodle moot global Chat and I will I might as well anybody need to know the the address If you get into matrix and search for moodle moot you will probably find this Yes, there's links on our website to the open ed tech stuff I just thought I'd show the moodle moot stuff along the way actually meant to in the keynote, and then I forgot all right, so That's the beauty of the of matrix though is that you know here I am connecting to the element community to talk to the element is the here's the matrix community Got stuff with my family etc. etc. So it's a really really good Messaging system and I meant I'm going on about it because it's the most open messaging system I could find and It's a good project. I mean I'm in contact with the CEO of it, and he's doing a great a good job He's frustrated by the speed of development as we all are on every project But you know it's it has a few a few wrinkles, but it's getting better and better quite quickly. It's it's developing well Okay, and I Think that's I know the last thing I want to show you is that we have a document system So we're just using etherpad for our documents now. We were using Google Docs, but you know that's not very good. Is it so I mean it's good, but it's not good So I'll just zoom in here. So this is where we're going to be working today So it's at docs.openedtech.global, but if you want to go to this QR code You will get straight into this document. I Did say we at the moment. It's completely open to the world We haven't hooked it up to the WordPress membership system yet that will be quite soon and so In a short time you'll only be able to edit these documents if you remember But it will be open to the world So yep, thank you for that. You can delete that message now. So we're gonna make some space further down Everybody in etherpad if you haven't used it if you click the top there you can see everybody coming in so Change your own name at the top if you want to whatever you want it doesn't have to be you know, whatever you are comfortable with it's basically anonymous, but it'd be nice to know who you are in this room and You have your own color and when you edit things it will show with the color in the in the settings you can You can turn the the colors on and off Yep on the left hand side, we've got the outline of the document and What we're going to be doing and what I'm going to be starting off here is starting to have a little bit of a brainstorm Around the AI personal assistant I'm also going to create headings for the rest of it so you can start putting your things in there as well So let me get on with that Well, actually no someone else type for me. You've got all you'll you can all do this. Can somebody make me a heading Under at the bottom probably heading level to please don't mess up the document. Okay, so this is the commons here Keep it tidy a level heading to for Presentations short presentations at the end of the document and if you have an idea for something you want to present Just please add it there And if you have a link to something put the link in the document as well and we'll you know during the next couple hours We'll build that up and when we can work on it. Thank you for that. Look. It's magic. I'm already got an AI happening here and I'll let me put this in here. Let me just so Gonna just Tidy that up a bit. Please don't mess that up Okay, I want to have a subheading here about problems and I want to have a subheading about solutions Actually, no, we'll stick them under each one Yeah, do that great so No, no, no, please don't write anything there yet under problems. We're gonna we're gonna talk about this in the room Thank you. So Let's let's have a we've got some microphones here Let me just lay out the scene. So at my keynote. I was talking about the I believe a central part of a future learning system Could be and should be and probably will be just because of the convenience factor will be an AI assistant that you access on your own device That will be a major interface to the rest So it's gonna sit on your device Let's just start with that concept Racha this morning who I love that presentation by the way because the UNESCO framework is Quite well thought out I think in terms of rolling out stuff the Policies and the the structure of things and the reason for things. I really want to my keynote from the first day I actually want to fit those things into that framework better actually and use more of the UNESCO language around it, but Something she was talking about was an issue with AI assistants is How do we target it to the lowest the people who have the least Access to education like not just say oh, it's for everybody. Oh, yes, but you know, you do need a $2,000 phone Right, so there's an issue right so issue issue number one. I'm gonna suggest. I'll just to kick this off is access to the lowest tier of Let's say learners So can somebody type that in for me? Um Maybe someone just type that in under problems, please as a sub on as a bullet point great. There we go. Okay I'll let you fight it out Is anybody else got any other suggestions around this concept for like a problem? We're just gonna have a list of titles and then later on we're gonna talk about solutions and brainstorm It's like what are problems? What are roadblocks with this concept that you can see? Yes, sir Let's can someone run around with the mic. Yep. Thanks Tiff Actually, you can probably distribute a few of them to other people we can get around faster, too Thank you. As you mentioned in your keynote data privacy protection Yes, it's okay on a stone data privacy protection Okay, so we can we can prove our assistant, but we need people to trust the assistant Yes communication and whatever we put in has to be kept private Yep, I think that's a valid problem. Anyone disagree Okay, anyone did anyone disagree with that? I think it's already solved No, okay problem to discuss later Number three, I think it was you over here I know you're over there. Sorry. No, Tiffany. You do the running you do the choosing Tiffany You have to make yourself look attractive to Tiffany just a quick question Are we looking into technology problems or overall people problems anything? Okay? Absolutely limited economic ability literacy skills Availability to things like electron electricity Should be covered way before having an AI personal. I think that's what I meant by the first one. Yes, actually, okay Maybe someone could add some of those sub points devices electricity What was the other thing? literacy, yeah Okay, great Yeah, yeah, correct. Yes. I think does that make sense to you that it's kind of related to the first one that's I didn't mean the stupidest learners. I meant the yeah when I said lowest I kind of met lowest Yeah Yeah, yep That's what I had in mind actually. Yeah, totally. So I think that's that's in one thing there No over here. Yes, it's another accessibility issue but from a different angle Generative AI is quite good at big languages like English Spanish French and so on But if someone wants to learn in Icelandic or something Yeah, it doesn't have enough data to actually speak the language so that adds to the accessibility issues I think yeah, very good. Yeah, anybody disagree with that Don't think so So what do we call that like? Language ability language language support Yep, oops, sorry, I really got to turn off the screensaver Come on. Why are you doing that? Okay, where was it? Sorry, I'm trying to remember how to turn off the the headings because I want to turn them off There's a way You want to remember? There we go. There we go Okay, language support good one. What's the next one? Yes motivation What motivation? We are adults motivation. We are very short of time So what are we going to find in this for us to be worth it? Yeah, okay Anyone disagree with that or want to modify it? No, I think it's a good motivation. Let's put motivation up there. Thank you. Wow. It's just coming in real time now perfect Oh, we're right behind it. Yes. Yes, it's maybe it's just a sub point for the first one but accessibility for the non-typical Non-neuro or non-body-typical people so if you are blind or disabled Alright, okay. Yep, so physical disabilities or Accessibility to What's the what's a good phrase and we want to suggest one? Kind of a full spectrum accessibility or something Yeah All right. Oh, okay. Yep. Keep that inclusion. Yeah inclusion. That's good Yeah All right. Good one. Yes so it kind of goes back to like trust if the Personal a I assistant is targeting like children or kids like Transparency can the parents see what's going on and that flips over to his adults to the trust do we can we see what's going on? but Best time to learn is when you're younger. Yeah, so Transparency for the adult for the parents of the kids. Okay. Yep, beautiful Could I say trust and transparency should I combine them? Shall we Trust and transparency. All right someone put that in Who's doing most of the typing here? Can I have your hand up? Who's who's who's subscribing? Okay? Thank you Thanks. Thank you. Love the way that just happens Thank you, Anna Can someone put yeah trust and transparency is the next one where where is it? Oh, right. Oh, yeah If you could put them in order actually So just so we can remember the order. I mean kind of mentally where Unfolding them in a time. I think it'll be easier to keep them in memory if they stay that way in the time order What is the next one? Yes, okay, please It's not a problem actually but for an assistant system We should like talk about the mission of the system its goals and in which context is going to be used Because we need to think about the accuracy of the answers given to the users so is that more about the accuracy and the trust of the Behaviour or about the you're saying about it should state its mission and goals. Is that a solution or is it? Yeah, we should think it's going to be used in which context and can we afford that like chat GPT is going to say Something wrong or not Okay, so So for that we need to think about that because there are multiple technologies and it depends on the context of use Yeah, I'm just trying to see if that fits in the first one here or is it a separate We're kind of loading a lot in there on trust But I maybe maybe trust is just one issue. How do we trust it? Yeah? Yeah, okay? Maybe it's just one issue. We can discuss all the solutions later Someone more in the back get around a bit Some So I was thinking about ethics would it be required for such system to have some kind of ethical behavior So that not every project and every person is supported in doing what either they want to do Yes You mean Is that ethics in terms of some of these other things? Or is it I don't I don't see it that way. I was thinking more about an Asimov's Robotic laws. Yeah Okay. Yeah, so built-in ethics Hard-coded guardrails, right? Yeah. Yeah, okay. Yep Building ethics into the system. I love Isaac Asimov by the way He is one of the most fantastic people ever ever To exist if you think about all the things he He explored Earlier last century he was already deep in these issues and Is anybody watching foundation on This TV series anyone else watching foundation? It's incredible. This you know, this guy was thinking about human civilization for the next 12,000 years plus across the galaxy and One of the things that happens in it Ironically is that the robots get so powerful That there's a war with humans and humans eventually win at great cost and they actually ban all robots From the galaxy. I I hope we don't get there So maybe put no No robot wars No, no, don't I'm joking But yes ethics into the system. He had three rules Does anybody remember the three rules one was don't hurt humans? Through any action or inaction What's the other two? The Yeah, obey human orders. That's right unless it contradicts the first rule. Yeah Yep, and you And protect yourself That's right. Yep at last. Yeah interesting set of ethics Who's next yes, yeah, and I was thinking of provide good quality interactions Providing good quality. Can you expand on? like interactions When you interact with the machine and machine machine and human machine Human human. Yeah. Okay. Yeah, let's put that down. I think quality The the issue is how do we make sure it is a quality interaction? I think I Mean the quality is probably quite easy to Define in terms of each person will decide whether they're going to work with this thing or not if it's low quality Don't just stop using it So it has to be Really useful and high quality Simos Yeah, I think Listen with the ethics and to the mission the goals. I think identity Because always when we choose when we want to collaborate with someone, you know, or or hire someone, you know We have a process of choosing or discussing, etc. So with this type of AI Assistant we need somehow to define, you know, the characteristics the identity the identity. Yeah, okay. Is that Do you see that as Okay, no, I won't we'll discuss it later. I think so what is the problem is what is the Out of the box identity of the of the system If you look at the movie her She came straight away with a voice and a personality and a whole thing That may not have been a good identity for everyone who opens that system Or was it customized for him? Or does the identity change over time? All good questions we can discuss, but I think yeah, let's put the something about identity in there the the the Out of the box personality, let's say, yep Up someone's put it up under trust again We're cramming a lot of things into trust, but I you know everything is trust trust. I saw a quote someone said Trust is a hundred times information Like you can get good quality information, but like trust is worth a hundred times good information Because if if someone you trust to gives you information, that's how you know it's good All right, is there any more we're gonna have a big discussion here Yes, Alex. Yeah, I think we should definitely mention something about the model that we are using since this is Personal AI assistant good. Yep. I mean I see this as something that we wish something that can be used as a framework Yes, so how we want things to but how it works and regarding trust and every ethics and everything else. It's it's It's on the model choice of base model choice and and should it be model agnostic? I mean yes Can I yeah, that's a choice of base model? I think that covers it. Yeah, very good one I like how we're getting a bit more technical too because we Also do need to talk about some of that stuff choice of base model I think I'm gonna stop soon because we're gonna have one of time to get into them So maybe two more two more Who feels I have a big one got a big one. It's a big all right I guess I think about like critical thinking ability and like soft skills So, okay if the user is it dependent on just their critical thinking ability How is it programmed or reprogrammed? You know for accessibility? So, yeah, is are we going to have pre-programs? How does somebody choose that? What about their level of competencies and how are we really gonna go about encouraging that? So, okay, so we've summarized that as Soft skills soft skills. Yeah soft skills not a phrase. I usually like but it is Emotional intelligence, maybe yeah, yeah, I put a cue Can I add one actually? Proact level of proactivity How proactive should this be chat chat GPT if you've used it is not proactive at all zero proactivity My assistant Liz is very proactive my phone goes ping and I'm like, oh, that's right like Let's discuss proactivity. How proactive should it be? How adjustable should that be? And that that's for me. That's an interesting one because it speaks to how autonomous is this thing? One more somewhere You've already had one someone who hasn't had one anyone who hasn't had one. I haven't had one Okay. All right. This is it. We'll stop here. There's plenty more You can keep discussing in this document and for years. So yes, let's start. Okay So if we think of existing products built on large language models, they all basically Exploit the human all the human content output, you know Okay internet and talk to any people working on a creative sector nowadays. They're super concerned about that So yeah how to prevent this kind of exploitation with these with these assistants. I think that's a big problem. I Think that comes under choice of base model because the base model is some technology There is a technology of the generative AI and then in most of it is about the training data and that you've put into it Was that be it would that be okay with you? Or do you think it's separate? More what sorry oh, so yeah, yep I Think it's still the base model but because the base model is literally like this is the training data We've used but maybe not new Alex you look very agitated Okay, all right Okay, we yeah, let's let's separate training data in general. We just say what data The issue is what training data is being used. I think actually yes, actually now I'm thinking there's more and more issues actually because I wanted to listen to my conversations 24 hours You might not want to That's also a training data issue. So yeah, I think we can expand a lot on that. So put training data there Look, we are focusing on the learning aspect, but I honestly think why would you have two? Like why would you have one for your life and one for your learning when your learning is Determined by your life. I mean that's a logical step. I made I mean we can But let's yeah, maybe there's another one that handles your flight bookings Let's say but let's yeah, we'll focus on the learning So imagine the movie her and you've got a device Like that that you're interacting with all the time and it's focused on the learning aspect of it to start with yep Okay, great All right, look, I really want to get into actually Talking about these things now. So let's let's I think we could all if you've got a computer and you want to jump in And you have we can start putting sub bullet points under these But I think we'll have a discussion in the room as well starting from the top and We'll start adding some flesh it out a bit with some solutions What can we what can we see? So Do you want the colors or not who wants colors I can turn off the colors who wants colors? Two people who doesn't want colors? More all right. I'm gonna do that Don't like how that menu disappears when I'm small. There we go okay, so Accent, how do we make sure this? What what are some solutions some ideas? Just ideas that could be Sketchy they can be well thought out. Maybe you know examples. How do we get a technology like this to? the people who have the least ability or access today Now we've got some ideas already a minimal offline mode available. I don't think it needs to be minimal Why can't it be completely offline? For a start if you haven't got internet why can't it be something you could live a device you could literally just send them You don't need internet for AI that is a fact right like I said in my keynote if you missed it and I can show you here Actually, if you want to try it, I'll put the instructions up in this document later if you get mlc chat It's called mlc chat It's an app and then do a search for Actually, if you look I do a search for dr. GPT or one word and You open the app and then you download the dr. GPT model into the app and then you can run the model on your phone 100% offline. Oh sure. Well, that's the other issue. Yeah, that was like the only thing that solves is lack of internet So, yeah, there's other solutions. So Let's go. What what other solutions are there to get this to the as I said, you know We're focusing on the the people who need it the most and it's hard. It's hard for us We are all wealthy. We wouldn't be at this thing if we weren't relatively wealthy You know in the scheme of the world, we are all wealthy here We're all privileged and it's hard to think about everyone else, but We need to right well, I can suggest one thing It's a bit far-fetched Like a think of a couple but look here. I think of the easy one first What about approaching device makers? You know trying to shame them into getting more devices out there You know working with the people who make things and try and work out win-wins Somehow get funding towards buying these things and pushing them out there. Maybe I mean The far-fetched one is building robot factories that build these very very cheaply Like a open-source hardware and you have robot factories churning them out Anyone else anything else? Okay, yeah libraries good one good solution to books. Yeah libraries Like that. I think we should add open source It's basic, but you know you shouldn't be paying for any software licenses certainly right and Just like internet cafes open at tech cafes hosted by AI so you can learn by living with that Yeah, okay, so like delegations of People going and bringing bringing this stuff Yep Yeah, so is that different from okay way? Yeah, we'll Put that there. There's two more microphones here if anybody else wants to you know grab a microphone and help Help Tiffany out here. We can pass them around Just so everybody can hear the comments so that yeah, thank you Alex for about the licensing Protecting the licensing as much as we can enforcing or open source. Yes Custom-made Raspberry Pi devices someone's added here great Richard have you got one? Oh, okay. Thank you Ah Yes Like courses they are supposed to follow UDL guidelines. I think it's going to be a Good idea if with the chatbot itself follows UDL guidelines. It follows. What sorry UDL UDL. Yeah, universal design for one. Yes. I know UDL, but does that how does that apply when you're talking directly with somebody? Yeah, the chatbot is supposed to like send text answer with text with videos with Different kind of media and like that we have the different means of representation and Is going to be engaging so it's going to be respecting UDL guidelines Does that help the lowest tier of learners? Oh god, here we go. Let me turn this off Does is that for this section for the solution? Yeah For accessibility for accessibility inclusion or if we follow UDL guidelines, we are okay inclusion aspect. Yeah. Okay. Yep. Got it Okay, so let's add Good good design principles five minutes to the break. Okay. Good Yeah, let's add UDL for well, let's say UDL inclusion Guidelines yes I'll do this later the next after the break All right Someone typing that in. Okay. Thank you All right, I'll move on to the next one. We'll quickly go through. Oh, no trust Let's start with a Literacy literacy skills actually Only because trust is a big one. I think we'll come to it after the break Literacy literacy skills so a lot of people And I know some people who would not even be able to know where to start with a Voice assistant maybe Well, you know in text maybe but if it's talking to them Maybe maybe that's already a big start However, it's got to talk to them in their language in their dialect I Know some people who speak of some very complicated mix of languages at once You know, maybe three languages and slang all mixed together I don't know what are some solutions here. How do we make sure that it works out of the box with anybody? Any language any ability of literacy? training Someone saying training. Yep direct training They will yeah, they'll be multimodal. I'm sure It might be a bit further out to get it on device and match all the other things But yes, eventually they will be able to They do now you can already have a eyes that will flash up images and videos and as you go So, yeah multimodal. Yeah, put multimodal down. I think that's important. I Think it would be beneficial if there would be some human peers Group so that you always have someone outside of the technological order to interact with first, right? Yeah, so community learning centers That kind of thing. Yeah Which is I was in Indonesia recently and they have this beautiful lifelong system lifelong learning system. They've built called Prachaja Indonesia is a massive country. It's like 280 million people over like 6000 islands And they need distance education to work and they have the government gives everyone credit money actual money And there's a website you go to which is a training provider portal And you pick what you want to learn and then your credit is applied and the training provider gets paid and So that all works out in the villages Where they're you know smoking fish like some of the things they need to learn When you're smoking a fish Don't stand in front of the smoke because you're breathing in smoke and you're gonna get cancer So they're like, how do you smoke fish in a more healthy way? that's a literal thing that is a huge problem that needs to be taught and So they have put community centers that start Introducing people to this idea of lifelong learning and how to do things better because they'll go Hey, we've always smoked our fish this way and my granddad lived to a ripe old age of 35 Yeah, so that's it's pretty good Sorry storytelling. Yeah, absolutely. Yep storytelling or Maybe Maybe to be broader adapting to local Methods of teaching to boot to bootstrap them in yet. Sorry or what? Yes Yep Yep, so build that into the model so the model Evaluates you in a way Adaptive adaptive learning for the model. Yep Yeah, you know when you when you install a new operating system on you might not know how on an iPhone You might be not be an iPhone user but on iPhones it starts saying hello to you in lots of languages So it could boot up and go start trying to find out about you. Okay, you know tried it. Hello. Bonjour, and then you know help you to Connects. Yeah. Yeah, absolutely. Just does some of this bleed into the language support Sorry, what does some of this bleed into the language support the next point? Yeah, probably probably all the same thing and they all probably in some ways relate to the training model of the LLM Yes, because because localizing story-tapping storytelling the the language in Indonesia and and then the example of of Iceland thinking about some of us from Quebec how our French is different from yes in a French from France That it that those are all sort of the same level of thing. I think yeah, no good point. There's Maybe say Link to below Good, look, it's a break time. I'm not gonna go get a coffee. You have a have a break and We are back in how long Tiffany 15 15 minutes We'll see you back here if you don't need a break feel free to keep typing stuff in this document and This is really going well actually. I'm really really liking where we're going with this so We have quite strict time limits and I know the coffee situation is not good that they just shut off the coffee really quickly So I'm getting someone to bring some coffees in maybe but We should also get going. I mean, I really need coffee because like I think we all do a lot of late nights and a few drinks And a bit coffee can quite help. So I Feel sorry about that. Let's see if we can get some in but we'll push on Only because this really has to end in I think we have 40 minutes So I'll spend another 20 minutes Maybe no has anyone added. Let's see. Let's see what we've got on the presentations. Let's see what we're working with none, okay Is anybody think of something they want to show if not, we'll just keep going with the first bit Anybody got anything? Apparently not. Okay. Well good. So I mean are we are you happy with the discussion? We're having is this good useful? I mean I've this is very very useful for me because I'm gonna give it to an AI and it's gonna design the system for us Okay Let's deal with trust. I I really want to look at the trust and I would love us To talk about the training data as well, I mean I want to talk about everything But you know this can we'll continue after today Let's keep going with trust alright So how do we increase trust? Of the system there is a lot of private data in there There is a lot of Steerage or steering going on this thing is literally trying to teach you right Imagine you know Donald Trump built a system to teach you For example not not not a system Most of us would be happy with So, you know, there's a lot of very bad possibilities and there's a lot of good possibility. How do we trust that this system is Good for us and the world if I can summarize So someone's written one here, which is Constant access to personal data and the possibilities to erase them. Is that a solution? Probably I mean, I think so actually probably is I mean I The reason why I wanted to access all my personal data is I wanted to understand everything well and then I can trust better It's suggestions because it knows me If some random system says you should learn this is like, yeah, okay. Yes correct I well like yes, that was one of the one would would create their own model so like a let's say a university or some organization can download the open source code and Well, I mean, yes, so good. So, yes, let's be clear. We're here. It's open ed tech. Yeah, so this would be Hypothetically the system that open ed tech would recommend and help to implement and it would be we would be saying Yep the combined thinking of the association is suggesting this is a System that solves these issues And and works something we could all feel comfortable Pushing and recommending and getting out there. So Yes, others can make their own AI's and there'll be all sorts of things happening everywhere But the open ed tech will be recommending this one Yeah, but and it would be important for the trust of the system like if to if if Donald Trump could Yeah, download the open source code correct his own model and we will have a Donald Trump Personal learning assistant. Yeah, we will be we'll be fighting the truth bot. Yeah, yeah truth social bot So yes, exactly. We trust Donald Trump then I can use his assistant Yeah, I hope open ed tech becomes a beacon of trust. Yeah Okay Is that clear? Yeah, okay Yes here There's two more microphones actually if anybody wants to just Pass them around. Yep. Go on. Yeah, so the trust and a currency of answers. I think they depend on the technology used So yesterday we had a presentation about two types of chatbots IBM Watson assistant and chat GPT So with IBM Watson assistant the developer the chatbot developer will answer the answers So that will do what to the answers answer the developer will add the answers to the chatbot Ah, so they're programmed. Yeah, it's programmed But you don't need like coding skills to do that. Anybody can do that. Okay So here we don't have a problem of accuracy at the same time It's going to be interesting if the chatbot is able to generate Text answers, but using data or documents or files or whatever We decide to do to use and to add the train to the training. Yeah, so that's also possible Yes, I think we are I'll say we're more focused on generative AI approaches. So it literally it is generating answers From its brain. Yeah, I think generative is a nice approach But with our documents or whatever we decide not really open because I don't know if the answer is not appropriate, let's say it's going to create a problem So you're saying the training set should be or the core curriculum for the for the bot Yeah, is a limited and decided. Yeah, and who decides that? Core curriculum the community the community. Yeah. Yeah, is okay. Well, let's have some note taking here. So Obviously, I think I think training data is very key as well Yeah, I was thinking one key property for the models we want to recommend is that they are Reproducible, so we know how they have been trained and potentially you could train it again And also traceability when they say something that it's possible to see where they got the information from Rather than and then each person can you can decide what data they want to get in or something? But it's important that all of them are traceable. I think Yep These are both in Moodle's AI principles. I will add Can someone else help do the typing I can't time it's hard to type and talk at the same time Without coffee. Oh, thank you No, that's all good. So I think one of the key aspects for trust is decentralization Because if you have one instance, we all know the example of X formerly Twitter If you have just one instance of something it can be a high-checked. So I think Decentralization should be essential for the system even so saying the system Kind of contradicts decentralization, but still yeah So your personal assistant should be truly personal Yeah on I mean, I think we're I'm trying to be open with the discussion, but I think it's quite clear that eventually it needs to be on your own hardware There's no other way as soon as it's out there. It's risky right Unless I mean there might it could be ways to have it on a cloud and still trust it. Is there? Yes Yes Yeah So, how do we give people trust that the down the download Even is safe Going back on trust. I mean, what are our key performance indicators? Like how is the system judging its own? Performance so like what are its indicators? How are we measuring KPIs? What's really important as far as performance goes like I think we would need to have to develop those so that With anything with building trust its consistency, so how do you build trust with humans? It's well, we see patterns So how can we develop patterns with AI that it can be trusted? And I think that that comes with time and being able to actually identify that we're measuring this consistency of Result let me push back a little bit on that. How do you trust a person? Do you get them to tell you the metrics or do you build the metrics? How do you trust me? Do you want me to tell you how much I'm trustworthy or do you decide that you head? That's what I'm saying like so the user did they develop their own KPIs and they say this is how I'm going to measure the success of the personal assistant how you would do in real life correct and Does the actual program have its own internal set of KPIs of how it knows what like going back to? Don't get cancer from putting your face in the smoke, right? Well, how does the the personal assistant know that it's been successful? So how is it measuring its own success and how do we measure its success? Okay, so we're talking feedback mechanisms On every activity on every action. So yeah feedback loops and KPIs. I would say for sure. Yeah, we need to measure it Yeah, okay That's a good one. I hope someone's typing because they did I need a typer We need a typer. I got my hands full Good one. Thank you. Yep. I Would like to connect back to the trace ability Yes, the problem with the many AI systems is this what you call black box It's some kind of magic that happens and you don't know what and our results comes out I think the trust issue is You want to see what? what The results comes from for what what has the process What what is the process that has led up to the answer or the result? I think that the Personal assistant interpret that you showed At your keynote the it was pretty good at Showing what what we what it was doing doing. Yeah, I think that's our way to go to to build trust for an AI system Yeah, okay. Yeah true. That's while it's working. It's showing its steps step by step that step by step thing improves accuracy as well That's the chain of thought techniques and tree of thought techniques and And then the ones that even build on that Yeah, so breaking problems down into small concrete little things, but I mean You're talking with it, right? So if it's over explaining everything all the time, it's going to get very exhausting Maybe that should be a An optional feature. I think actually chat GPT itself already does this because it has its own Interpreter like open interpreters similar to something chat GPT has which used to be called code interpreter And now they've changed it to something else. Anyone remember the new name Anyway, when you switch it's a switch in the settings and it will show it's working But it hides it under a little a little arrow. I just show it I Have something to say Martin I think we should clarify what we mean by trust what we are trusting in my opinion We are trusting the LLM model because and if open a tech Assume some kind of responsibility then you assume the responsibility of suggesting an LLM or I mean You're not supposed to to train one LLM from scratch Yeah, I think I think that makes sense that we would recommend an LLM as the base And it would be there by default Yes, because then it would certify it certify an LLM and then you have to not choose a Donald Trump or for that Matter Karl Marx, yes, right and Then you have the training data for me the training data Since we are talking about personal assistance then the personal part of of this comes from the training data and the training data This is in my opinion what needs to be what what's personal and needs to be protected? Like a lot because I should be able to keep my training data in case I want to change or you suggest another LLM for example Yes, I think that's it's always going to be part of the solution because the way neural nets work is Then they're not ever looking up information, right? It's not a look up. It's It's Statistically putting words together and just like we like I might be doing my absolute best to be accurate and trustworthy and I'm going to misspeak and I'm gonna make errors and you we all are like that's just how Life is we make errors. We're not so we're not untrustworthy in a sense of We want to be we just are because that's how brains work They're a bit random sometimes and they remember they forget and they mix things up and they have misunderstandings and you know There's a lot of gray I think that's always going to be there in an LLM but the way of mitigating or preventing that is to have checks and reviews and feedback and That's how we solve things in our organizations to well we have teams we have processes we have management we have reviews and That's how we fix the problems So but I think in the future maybe a lot of that will be built into the LLM like it's so it will be quite Trustworthy, so you probably maybe you're right maybe it's just So look I've got off the track there so That was about Some can we put something in there about you got it certified LLM. Yeah, someone's putting that. Okay, great The traceability I just want to get back to that actually because This one thing to say step-by-step during the process All of interacting, but what if you get a result and you want to go back and Ask it to derive get backwards derive things So I can only think of stupid examples. I haven't had enough coffee, but if I ask you to draw a Schnauzer right a type of dog and I ask you to draw one Right, you'll have a go. You've probably seen a schnauzer before maybe not recently You probably can't quite remember what a schnauzer looks like and you'll have a go And I'll go and maybe you give it floppy ears. Maybe it has pointy ears. Does anybody know pointy or floppy is I don't know right what what it is of a schnauzer like so You're gonna Draw a picture and it's an it's got an error in it and I'm going how did why did you say pointy ears and You are now trying to think back to all your experiences and like your learning How you got that impression It's that right we want the AI to be able to do that really well and go well It I know it has pointy ears because blah blah blah blah blah blah and we have to then then we can trust it It's probably going to have to look up things. I think Is the AI going to have to be able to use the internet? The internet of course is going to become a cesspool of misinformation Which it already is and it's getting worse and worse a Little side projects. I will never do but I dream of doing is having an AI Look at Wikipedia get all the names of the articles and make a new site called my Wikipedia Where I just get an AI to write all the articles again based on the title and put it online No, I will never do that But anybody could do that right and that would just start getting in the search results and you'd go You'd come up and it might be third on the list and it would be yeah He's a really good article about everything and it would be like 90% correct and How will our kids know the difference? They won't know so yeah, there's Trust comes back to The stuff we put into it how because yep. Sorry. Can you start again? Yeah, because It's not only that if we can trust the source of information. It's also the fact of The gene AI that the system is choosing was the right answer and the level of depth of knowledge in that answer So are we going to stick on a very superficial? Superficial surface level of knowledge because when we look up something in the Wikipedia we just go for the first answer we get and Sometimes we investigate more sometimes we don't so the system is going to be fed by this recent search and this recent training and this recent answers and it's going to Be just going more and more superficial because there's no one investigating Yeah, so are we going to be able to trust? What the answers is this going to be all the knowledge? Is there anything I? Don't know how to say it, but is there anything when I ask this question Anything that I should know that it's not going to be said in this first In this first answer. I'm receiving is there anything if I have this critical Thinking that someone was mentioning before is there anything I should be aware of That's as it's not been answered before it's going to end being forgotten Yeah, very good questions what you were really triggering my neural net here. I was thinking what about this model to solve that What if the on-board AI? doesn't have a Lot of detailed information because it can't right it's not going to have the whole internet in here But if we focus on the thing that's here being like your super nerdy critical friend who Has a good general knowledge that we define in the base model a general knowledge for easy stuff is very good at knowing where everything is and Can do look-ups and open ed tech also creates we have an OER group the open education resources database of Knowledge that we approve We trust through processes we debate sort of like Wikipedia, but But serious. Yeah, like really yeah focused on on curriculum almost like the key things that we The basic education stuff that we're trying to get up a level of education like the core curriculum That's what we're working on at moodle net actually is this kind of idea of quality in the OER then if that's kind of the ongoing resource that our on-board bot can say Look, I don't know enough about that But I know where to look and I'm going to the open ed tech database and that's what this says Does that make sense to split the responsibilities a bit? Yeah Yes, don't you have it? Yes, you want to know more let's go to some serious resources you can consult and we will direct you there Yeah, because we want this to be serious and reliable. I mean so when you say I've been using This system someone is going to think okay the answer is reliable Yeah, that makes a lot of sense that seems like it makes sense to me and and you and then you could say in terms of traceability you could ask and say Who decided that on over there and it go well a hundred scientists with climate change credentials Stamped it and it's pretty pretty likely right So that's where all the data and the ongoing improvement happens And you don't need to update this every day because you can't it's gigabytes But it can reach out that but that does mean you need internet But I mean that's not too unreasonable to need internet. I think All the time, but yep Okay, I'll someone jot that down. I quite like that actually that's a Who's got it Yes, I have a question about the fact that Generally the system not always have the Availability of the internet data. Maybe we can introduce some knowledge a knowledge for example logic based knowledge for to say to control for example a programmer can say I already know that This kind of answer maybe is not too good So me as a programmer if I have the right knowledge I can introduce some rules To control this this kind of Answers also by introducing some constraint for example should be not possible that For example the answer I make a stupid example. The answer should be without I can say if I expect the Italian language should should not be give me back the English language I mean some control that the programmer can introduce In some way some kind of algorithm or rules logic based at the rules So you Yeah, just so I'm trying to understand actually so the logic you're saying we add more Like hard logic programming logic programming. It's more assurance of certain things for some parts Yes, it's not all generative not some core of logical a logical artificial intelligence symbolic artificial intelligence not machine learning or Yeah, a bit old-school AI a bit of that in there as well. Yeah, yeah No, because we are working in this kind of artificial intelligence our company also. Okay. Yeah. Yeah, we can oh someone put that down Yeah, it's like yeah, think about mixing AI models. Yes to get more trust. Exactly. Yes mixing different kinds Okay, that's a good point. Thank you. Yeah Every time I say anything just imagine I'm saying thank you after it because Like I thank you all the time, but I I'm told that I forget to say thank you a lot And it's just because I'm thinking about the next thing already. Sorry Yes, I don't want to open a box of Pandora's But I'm just thinking about if we want to approach the lowest tier and we are thinking of trust We need to think of trusting the other users because if somebody uses our assistant to build up a nuclear bomb or Makes the library of child abuse porn We will lose people we want to to reach with that. So we need to build a Protection shield Yeah, you're right. I mean the more we're talking the more I'm thinking about limiting limiting this AI actually rather than making it too capable like to to get to achieve some of these solve some of these problems it should be very a Bit like the the one on the mlc dr. The dr. GP t-chat I pointed out it cannot write poetry very well And but it does know how to do medical diagnosis very well, and it's very possible to a very special purpose So I mean I it that limits those damage, right? You won't be able to generate child porn with it because it just doesn't even have that in its training set It's not able to So that makes it safer. So safety is part of trust. Yeah, you're right. You're right I mean if you if a kid gets this thing because their parents go, oh, yeah, open a tech great here Take the AI and then they start doing crazy crap with it like this not no one's gonna trust it So you're right needs needs some safety But if you you can't overdo safety to like if you look at llama They overdid the safety and you would ask it questions like You know are there more men in the world or women Statistically and it would go I can't answer that because it is a it's a it's a it's a way a gender issue or something You're like it misinterpreted What the question was about as being you know Some sort of you know hitting a guideline a guard a guard rail So yeah the training data, so we got 13 minutes, I believe have we? Yeah, okay I was do we keep going on trust hands up to keep going on trust or to move to another issue Hands up for trust keep going One are you got a question on the boil? All right, we'll do yours and yours Yeah, okay trust board. Oh, did you say trust board? Oh nice trust committee and open ed tech Yeah, okay Yes, actually. Yes, that's actually what a lot of organizations end up with so good idea We'll have yours and then we'll move on to the other one. I said then so yeah, well, I'm not sure if it's trust or information and Training data perhaps, but I was thinking of the more I was The more I was thinking about what you said if it's if it should Connect to the internet and the nightmare that it's becoming to know what is true or not Perhaps it's performance issue but The AI has a better opportunity Then as humans to take a lot of data and make a comparison between the data and along with the traceability and Openness of how it works it would be possible to present like like we're talking about with critical thinking to present an answer that isn't Straight up Simple answer. Yes comparison of different sources that a little doing a meta research Kind of Yeah, I yeah, I haven't thought it through 100% but it's something a Comparison wise I think you're right. I I think that could be Doesn't need to be on the device here, but it could be in that if there's a open ed tech approved Dataset of OER Maybe there's this well, there will be AI bots crawling all over that helping checking reviewing Comparing and using tools to verify etc. That's what we're doing it at Moodle net in Moodle net for We're starting because of the quality aspect the first things we're doing are using AI to Fill out the metadata So it's going to analyze something coming in and fill in the metadata fields for you summarize make thumbnails extract work out which learning outcomes it fits and Presented to you on the next for the next form will be up. Here's what we think the metadata is Do you want to change it tweak it? You can review it and then you save it and then we're also using AI to start Helping with the recommend recommendation of what's popular. What's what's what's good. What's bad, etc And like it's just starting, but I think that Then you can have very big systems with a lot of training and a lot of reviews and a lot of repeating and a lot of feedback and and they can be in a Central trusted place somewhere Existing on an open ed tech cloud that nobody controls so, I mean open ed tech would set the The standards, but then it's got to be a collaborative Collaboratively supported place. So it's always funded it exists forever Like unlike Wikipedia Wikipedia run servers and then they ask us every year to send some dollars to pay for those servers It turns out they get a lot more than they need So they actually have hundreds of millions of dollars extra which they don't mention but anyway But that model is kind of relying on donations and it's a little sketchy, right? But I think the cloud Into infrastructure issues something we have to solve separately and I have some ideas about how to Make it safe in the cloud and then yeah, then then we work on the OERs in there and the trust and we can do a lot of things, but yeah, that's Making sense to not pile it all into the local device because if the if the local AI is Minimal it'll run on more devices, which helps the equity issue too because it can run on watches and You know phone all sorts of phones and probably fridges and things Okay I'm gonna jump to another topic before the end Sorry, oh there is Okay, I Yeah, yeah good. No, I love I think we'd love to hear from you Fred Yeah, let's do it now Let's do it now come a show of something. Do you want to use your laptop plug it in? Yeah, great great. I think Martin deserves a bit of recognition for Four days and probably five hours of sleep You know still passionately pushing forward this vision of how open educational technologies can just make the world a better place And I think you guys probably picked up that AI is both an enabler and an opportunity for us to Get further along this vision and if we don't come up with ways that we can use AI responsibly For the purpose of lifelong learning others are gonna do it and they'll probably be more corporate and probably less responsible So just a rant for a second when you talked about open ed tech and you know all the things we want to do There's a part of here like there's a brand of open ed tech and I think of like the red or what a project red or whatever Like maybe we get to a place where we're not having to build everything But we we vet other things and people trust this brand they trust the sources and they trust the people behind it and Organizations may want that trust from us and if they do well They have to meet these criteria and then it just sort of grows over time So there's value in what we're doing, but there's also maybe value in the brand that we're building as well Okay, so in the context of a personal assistant, I'm gonna put this down for a sec All right, I'm gonna do the one-handed thing as well here now. So You guys know me big blue button been working on this for 14 years really passionate about Building like the world's most effective virtual classroom for learning outcomes and one thing that we've been looking at is AI as well. So in the context of like helping the teacher How could AI help a teacher more effectively engaged with students? So imagine you're in a class and you want help you want help to just save you time So you can spend more time interacting with students the human part of it, right? Which is what a virtual class can provide the human interaction So one thing we know that is if you as teacher uploads slides, you're gonna talk about content So I've got something here I found in the internet about Rome and at some point you want to say our students getting this Are they following what I'm doing in a very powerful way is to do a poll. So I could Start up a poll and start like okay typing one-handed into it and settle can I do a poll? But the text is already there and we read it into memory with screen readers. So for screen readers So we basically used open AI's API Which could be like Lama to or can run the model locally. I mean I think this is where we're getting to and I could say well, I Want to generate a poll question, you know based on the content of this slide So it grabs the content sends it off to open AI with a prompt It says you're a teacher and it comes back with a slide with a poll question now one of these should be correct in fact the one that's bolding is correct and So if I don't like that one I can regenerate it and the idea would be is instead of me typing something up and trying to Remember it. I bet what it came back with and then at some point. I Will just basically just do the poll. So let's say I accept the poll And I get to change a little bit here and if I start it then on my screen here. It's a very big poll I could just press one of them and it should come back where there's one choice and Then at the end I won't try to publish this one because I think it would dominate the screen But that's an example of AI assisting the teacher to more effectively engage students online Thanks Fred That's good. I'm so glad to have you a voice up here too today. Look, I think we're really at the end anyway What No, no, I think I'd Hard to even get into another discussion only a couple of minutes from the end and we should you know geez they're really They're really strict with the meals here. So I want to make sure you everyone gets the meals I really apologize about that. It's not our fault. It's the way the hotel catering works. They're very like So we we just have to work with that and keep to time But maybe some you know general Feedback or topics, I mean is this been useful for you help you think about these issues as I'm seeing some nodding. I'm glad for that These are the you know our role as education Technologists which we all are in some way is To plan the future of education technology Not just use it as is today where the world is looking to moodle To open ed tech To all of us, you know, you know, you're all experts in your own Organizations, they're looking to you to help tell them and guide them So we're all here to guide each other and this is the the association The where we can work that out for ourselves so we can help the others so I'm glad we're able to have this in person today But remember there are a lot of things happening online The chat can be either very quiet or very chaotic because it's a chat It's maybe not always easy to follow As we get a little bit more time to have more administration going on. I really want to explore how we can do this these sort of working groups thinking and planning brainstorming designing Synchronously we use big blue button in our meetings and also asynchronously through the documents and Chats and and then also just keeping in touch with developments through mastodon You know, make sure you get into mastodon Follow open ed tech follow other people if you want if you want to count on our server just Sign up if you haven't already and and drop us a note and we'll create an account for you It's not actually no you can go to open ed tech dot social and apply for an account But we're only accepting accounts from people we know You know, I wish I wish I could just give it to everyone, but they that server already costs us quite a bit So great. Yes, and Dan. Did you want to say something? Yeah, thank you for facilitating I think it's been really fascinating and I love this approach instead of passive listening, which is great But this is great as well. I think it's worth mentioning. There are already a couple of plugins on the Moodle plugin database which are AI Focused I'm just looking at them now AI text to questions generator AI chat block as an AI connector I think there's even a Text to image Text to image generator as well. Indeed. So these things are out there and being developed and you know Run one on a test site see how it works and get using it. Well, totally I mean that but I'm not wearing my moodle shirt today. I'm wearing my open ed tix. No, I appreciate that But you know the moodle community is developing most of the AI stuff around moodle at the moment We we are there is that survey I pointed out yesterday where we want to find out more thoughts about moodle specifically But yeah, this is more about the the longer range vision for how moodle is part of a much bigger thing and So thank you for being at the open ed tech workshop We are finishing exactly on time and it will return you to your regularly scheduled moodle moot out the door Thank you