 I'm delighted to introduce this panel and we have John Elder joining us again from Elder Research with Ken Finnegan from the IDI Ireland, Barrios Sullivan from Insight and UCC and Patricia Scanlon from Soapbox Labs. So just to remind you all there is a drinks and canopies reception coming up very soon so this is going to be a little bit light hearted so we can enjoy a few drinks later on and discuss the outcomes of this very important and very serious issue that we're all facing. So of course this technological singularity or singularity is the idea that technological progress particularly in artificial intelligence will reach a tipping point to where machines are exponentially smarter than humans and as far back as the 1960s Alan Turing warned about the ultra-intelligent machine which he defined as a machine that can far surpass all the intellectual activities of any man, however clever. Since the design of machines is one of these intellectual activities an ultra-intelligent machine could design even better machines, now that's a scary thought, that would then unquestionably be, there would then unquestionably be an intelligence explosion and the intelligence of mankind would be left far behind. So Elon Musk is worried about this, Stephen Hawking is worried about this. Elon Musk created open AI, a billion dollar non-profit dedicated to ensuring the development of artificial general intelligence which is another term for human level intelligence that to make sure that it's beneficial to humanity. A few reasons why the singularity may be close, well first of all Moore's law right it's going to come to an end soon, people are concerned about Moore's law coming to an end because the number of chips on a transistor cannot be continuously shrunk because you'll end up in physical limitations of the electronics there and you'll have quantum effects and all sorts of problems, but Ray Kurzweil disagrees with that and he thinks that's an arbitrary measure of computational power and a more interesting measure is to suggest that the number of calculations per thousand dollars should be the ultimate measure of computational progress and under his reckoning that's actually going faster than Moore's law and can continue for decades to come. So imagine that the power of computers are going to continue at a rate faster than Moore's law for decades to come under this new measure of calculations per thousand dollars and I think that's what matters because money talks in computational power and industry. So robots are doing human jobs, GM assembly line in 1962, if you see Tesla's assembly line today you'll probably see some fairly smart robots doing a lot of jobs, welding, you have your Roomba vacuum cleaners came out in 2002, a lot of people are loving those. Rethink robots has made robots like Baxter and Sawyer who can work safely around humans and learn new tasks in minutes. So as the power of technology grows and continues to grow, robots are going to continue to take on even more tasks and that's clear. And even in the creative pursuits, one book written by a machine was recently accepted as a submission for the prestigious Hoshi Shinichi Literary Award in Japan. So they are becoming creative too. So with all that in mind, I think it's time to talk about the singularity. So, John, in your abstract for your talk, you wrote that AI has fallen laughably short of its potential. And we've been promised this AI revolution and singularity for a long time, but it always seems to be 10 years away. And you said it's fallen laughably short of its potential. Until all of a sudden, it hasn't. Is AI finally about to deliver on this hype surrounding it over the previous decades? And if so, why not? So over to you, John. Yeah, as I mentioned in my talk, some of the jokes around AI one was in terms of language translation, famous AI researcher was asked, how close is it? It was about a decade away in his estimation. So a decade later, he was asked, what do you what do you think of your prediction? He says, I see no need to update it. In other words, still a decade away. You know, and so that was kind of the that it was always under delivering. And then in AI's favor, a lot of times something new would be developed and it would be spun out of AI. And so AI was always left with the hard research problems and always sunk back into the trough of disillusionment in the Gartner formulation of the hype cycle. But AI always was hyped and always was under delivering. And it was kind of painful to listen to because they were getting money, they were extracting money from gullible people, you know, and it wasn't going to the deserving stuff that you were working on, right? So, but, but all of a sudden, AI is delivering, you know, AI's got the driverless cars, and it's winning at poker, and it's winning at Go, and it's crushed chess a long time ago, and so forth. And, and so it's really impressive. And then those aren't dummies that are throwing that are a little worried about the singularity about general purpose. So there there's more of a reason to believe in some of these possibilities than ever before. And we'll explore some of the limits and so forth. Just to lay my cards on the table. I don't believe the singularity will occur. I don't believe that that human intelligence is just on a continuum of computing power. But there are a lot of tasks that will be taken over. And you don't have to match human ability to still be useful. The standard that is being used in some of our presentations are can you beat the world champion at, but you don't have to beat the world champion at getting an airline reservation to be useful at a task of, you know, helping someone get the airline reservation or make a translation of a document to understand the gist of what's going on to be helpful in your daily life. And so, and having the immediate free ability to do that is extraordinarily powerful. So the special purpose AI is getting is is having a heyday right now where special purpose tasks are being taken over by AI trained on that purpose. So it's not general purpose intelligence, but special purpose intelligence and and niche after niche is is being sort of conquered right now with existing technology and it's very powerful. So it's it's gonna it's gonna be a great heyday I think for that in the near future and it'd be interesting to see how many tasks can be carved away and and gained for the AI world. So you don't believe it's ever gonna happen the singularity? I don't believe it'll ever happen. Never gonna happen. But a lot of useful things will come from it. So over to you, Barry. Do you think the singularity or the computational power achieving this general intelligence is is inevitable or also are you in the camp of it'll never happen? So I put my cards in the table. I don't think the singularity is ever gonna happen for a whole bunch of reasons, but I suppose that's not to say it's not a useful taut experiment which it is. But I think you know the I've been on debates about the singularity with course file and I think that there's a flaw in his argument which is that it's you know so whether so there's a couple things or on the computational power thing yes absolutely so there's massive growth in the power of computers that's true. But if you think about the brain as a computational system the amount of power it takes is tiny right it's absolutely minuscule you wouldn't light a you wouldn't light a bulb from the amount of power that your brain takes. So if you're waiting for if you're waiting for more computational power then you're on the wrong road basically because that is not how it works right so there's there's something fundamentally different about these things but at the same time I suppose artificial if you think of artificial flight we didn't build we didn't build an artificial bird right so so I think something that is that demonstrates AI doesn't have to look like a human mind. So that's the second point and I suppose the third point is that all of the successes that you know so John mentioned many of the successes if you think about them the components of them are just our you know tools our AI systems very very narrow AI systems that can do very narrow tasks very very well and even things as complex as driving these are engineered combinations of these they're not general intelligences right and in fact there is no scientific evidence at all suggests that a singularity is even possible right so you know Stephen Hawking who's a you know a bright individual in Elon Musk who is a very successful individual you know just because they say just because they're worried about this thought experiment doesn't mean it's true right so I don't think the singularity will ever occur but certainly what's going to happen is that's what we call AI is going to have massive impact as John says but I suppose one thing that I will say and this might be controversial is that the successes that we've had with AI they exhibit absolutely none of the capabilities none of the characteristics of intelligence at all actually so the problem with AI is that it has a bad label because we don't build artificial intelligence systems we build smart tools so for example deep learning tells you nothing at all about human intelligence and in fact the reason it's so successful is that we can just function fists functions that have billions of parameters and you know for most useful things that we want to use that for perception like vision machine translation such functions are just enough to do that so they seem intelligent but they're not there's nothing at all intelligent about them and that's coming from someone who's you know entire professional career is in the field of AI. Patricia help me out here we need a believer you divine smart devices that can understand human voice write down to kids who have no tolerance for machines that have that won't interact in a natural environment these machines these voice connected devices are connected to a super powerful internet but insane amounts of computational power is it not just inevitable that the this singularity is going to happen and I look at this I look at this very differently I think when you say is it going to happen when right so you know in the next by 2040 no totally agree with you right 100 years 200 years like are you honestly thinking that in 300 years we're still talking about silicon we're not you know there's more things in heaven and earth Horatio that have been dreamt of in your philosophy and just because we don't know how it's going to happen and we can't imagine it on the physical devices or in the we we're talking processing power here we're talking about cloud we didn't even have cloud like you know we didn't have so when I think about it I think we don't know what's going to exist in 40 years how could you possibly say it's not going to happen and as I said earlier on you can't stop progress and you know I think honestly I think Elon muskin and Hawking and all that have a point I don't think everybody needs to push the panic buttons right now but if you don't think about it and you don't talk about it um you you can put rules and laws in place to deal with this but the people you got to worry about are you know the king jump owns of the place who aren't going to play by the rules like so you know I think it's a very interesting thing that we can say oh we must control it we must regulate it really because the people you got to worry about aren't going to follow you know follow that and it's the same with nuclear power right it's it's it's an amazing clean energy source but it can also destroy the planet so I think we are I think it has to be a longer term conversation and for the next generation and the generation after that to be cognizant and have real conversations but yet maybe not in my lifetime maybe but maybe not in my lifetime but I think I'm in the same camp as you Patricia I believe you know progress is is marching forward and who are we to predict where we'll be in 10 20 30 40 years and I think Carzail himself has put a 10 year prediction on it but I would be a little bit further out of it like yourself let's let's move on to you know quickly I suppose um well you know let's imagine it's inevitable like I know you guys are totally opposed to this over to you Ken Europe to some extent has been very focused on GDPR regulation now I don't want to go into GDPR because we could talk about it all day but has it missed the risk that is AI and how does that relate to the regular as as we said regulation and rules that that people are looking for to try to corral this thing and prevent it from becoming that catastrophe that it may become in what where it's 40 years or 100 years time yeah it's it's it's I guess from from our perspective actually GDPR has been a very important conversation from an idea perspective for obvious reasons and but in the meantime we're very much focusing on AI as well and a lot of people on the stage here know that and probably in the audience as well and we had an event last May and to discuss actually all the topics are in and around AI and how to make Ireland's kind of like a as we termed it AI Ireland so what's what's our we wanted a benchmark where we are now in terms of capability actually the last speakers were talking about understanding a team so selecting 41 people and understanding their capabilities and that's kind of what we were doing as well from an from an economic development perspective so we're looking at team Ireland if you want understanding what the capability that exists here in industry and from an academic perspective and and from a national perspective it was a really interesting event that day actually we had the design led workshop where we had industry Irish companies academia we also had the unions there we had the military in the room as well we had social innovators just to have that conversation about actually what does AI mean to Ireland and what does the future look like from an AI perspective the sound just yeah I know Barry you have a point on Gina Fjord and on relations AI but just before we get to that can I think like what the IDA is focusing on is like how do we could become leaders in AI technology and AI capabilities but that's that's propelling us towards you know if that is propelling us towards it inevitable it's actually pushing us in that direction but we're I don't think we're doing anything on the other side to say well what are the regulations and rules we should be pushing into Europe for this are we well actually one of the points that came up so we wanted to kind of benchmark where we were as a country see what our team was or what their special capabilities what was a unique differentiator but in addition it was like we we took feedback from everybody in the audience that day to understand actually what are the blockers for us to progress in terms of making Ireland kind of like a global leader if you want in artificial intelligence and we have actually we've been addressing all those what are what are this brings us to a singularity not too sure and this is really fun by the way I don't get to ever have conversations about singularity when I'm sitting in front of clients of IDA but but it was really interesting listening to the challenges like for example one of the challenges was in around access to open data so we all know that that data is the is the fuel for for artificial intelligence and the perception out there was that okay Ireland's pretty bad turns out actually we're third in Europe from a from a government perspective and access to open data perspective and which is pretty impressive and we've I guess it's like finding out those issues that the perception of issues out there and being able to address them maybe we can speed up our progress towards the singularity one thing that I think that what I'm seeing on the ground at the moment is the whole idea of convergence so applied AI versus core AI and we're just starting to see kind of like organizations start to collaborate together and I think I actually asked a question over Twitter before of the last panel to understand where does Ireland rank in terms of tech and sport because it sounds like it's a perfect match in Ireland sport is very popular but also tech we have we've a fine pedigree in technology and are we up there in terms of that convergence are we up there in terms of on the league tables from a from a sports tech perspective because we can see like ag tech becoming more popular fintech has become more popular in the country so it's that weak AI I can see is progressing on when will we hit the kind of like the that's great always be selling Ireland Ken all right one of your thoughts on the GDPR regulation AI issue yes there is actually a direct link between AI and the GDPR so in so I want to ask people who reads this kind of legislation because I'm not kind of nerd but but there's a article 22 in the GDPR talks about algorithmic decision-making and it basically says that that's a an EU citizen has the right not to be subject to solely decisions are solely made by algorithmic means on other words you have you will have a right as an EU citizen not to be subject to an entirely a decision made by an AI system and suppose what that means is that the AI system will need to be able to explain itself to you or you'll be able to interrogate the decision and I suppose last year I did this interview with Silicon Republic where there's a sort of a glib title to it which is that AI that AI will become illegal in Europe in 2018 and what I meant was yes it's going to come that what I meant was that there are there are many AI techniques so deep learning for example which can't actually explain themselves so if you were a human being and you know you're interacting with a deep learning system that's deciding whether or not you should get a certain type of insurance premium and if you're not happy you say well why did you give me that insurance premium it's actually currently not technologically possible for a deep learning system to explain why right so so you know according to the GDPR is currently written and AI research as it currently stands there are techniques that are not that will not be compliant with the GDPR so I think that's something that's worth thinking about well just outlaw it there we go we're done and no no rest in the same priority unless it comes from North Korea or our Silicon Valley or somewhere else no but you know this is it's great if deep learning is outlawed only outlaws will use deep learning it'll go to the dark web right so yeah but I suppose but you know the flip side is that it's a that's a good thing for us as data scientists because of it sort of tells us where we should be doing research which is unexplainable artificial intelligence right so that's so we very very little time left so let's do this real quick and just for fun so if really short answers each person if the singularity does happen how will we know we won't we'll all be dead easy we will have solved so we will the singularity will have happened and at that moment in time when we when we also have solved the problem of immortality when I guess when we all live in this utopian society where the robots will do everything for us and we can actually pursue what really interests us okay John that is one of the one of the goals and hallmarks of the singularity crowd is to download your consciousness and have a type of immortality so I think when we feel a sense of loss of control that the the computers are pushing us around sometimes happens to you at the checkout yeah every time when you get to the matrix I feel like you're using it okay so we start to recognize that the singularity has happened what happens next very briefly sit back and let the robots do okay I don't know I think we enjoy our immortality because we think that we will solve the same problem Ken yeah what happens next it's an interesting question it's like that philosophical question what is intelligence and what is consciousness consciousness will the machines be able to tell us what that is because we can't answer that question we hope we're still amusing to our machine overlords yes exactly imagine imagine if there was a singularity you could come up with some fantastic jokes right we would just be laughing for all eternity I can't go to that be you can ask Alex at the moment tell me a joke Alex so uh show of hands please is a singularity inevitable yes in the audience please how many would say it's inevitable and first of all ever ever singular ever there you go okay we've got a good 60 70 percent how many people say do that again do that again six or seven people six or seven people how many people would say it can never happen even in two hundred three hundred years really okay very very confused so with that maybe we'll continue this conversation over a few drinks and and some food in the experience hall after I say a few thank yous and closing remarks so thanks very much to our panel it was very much fun very fun thank you