Towards the future

Key Takeaways

welcome to deep mind the podcast where we're exploring the world of artificial intelligence we're assessing what we know and what we don't know we're looking at what we're trying to know we're calculating what we will know mapping out where we're going and working out how we will know when we get there I'm Hanna Frey and I'm an associate professor in mathematics and I spent the last 12 months at deepmind and in this episode we are going to start looking forwards to the future the self schooled variety of AI known as artificial general intelligence but first you know what I don't feel like we really given you the Grand Tour of this place deep mind headquarters in Kings Cross in London so follow me in my squeakiest of sneakers closed doors coming through so the language they do machine learning understanding language with me is the extremely likeable Coria Caracciolo one of the right firsts deepmind is joining way back in 2012 is actually it's a group of people who are either focused on agents better memory more planning how can we get him get those into the agents all of your rooms are named after famous mathematicians Gauss room Hedy Lamarr Oh actually I know her I know her she's actually a very famous actress you know I liked her best in I take this woman with Spencer Tracy but she was actually also scientists right the deep learning group is more interested in just like coming up with algorithms and architectures on any data domain some seriously delicious equations on the whiteboard I've learned yes I mean blackboards and whiteboards if you have an idea in the corridor it's important use it this area is the machine learning so one of the main projects going on here with this group is the imitation just just demo so Sabres in the office this is the area for the neuroscience group they do a lot of thinking about what are the important problems what are the challenges that the discipline is really hard yeah table football seven of my course a regress ah the reinforcement learning team this is one area that has been core to what we try to achieve with a GI if you're interested in agents it has to be active they've got some proper weight eat expose that's not not exactly bedtime raging is a molecular electronic structure theory I know what all of those words mean but just not in that particular order sweeping funds sleeping I didn't know that they've been coming in for a year no one told me about these there anyone in there look in my defense she didn't have a Do Not Disturb sign on the door okay enough of that already back to the simple stuff like solving the Enigma of intelligence just one tiny snag before we get there first we're gonna need a proper definition of what intelligence is because intelligence you see it's quite a slippery beast to pin down [Music] we've got a bit of a head start when it comes to human intelligence although it might not be everyone's favorite metric IQ is one of the most stable psychological tests we have and it does a pretty good job of measuring limited intelligence markers like reasoning and logic but IQ still doesn't get us any closer to a definition of intelligence if we're going to get anywhere with this we need to define it properly we need some way to capture what we mean by intelligence that works just as well for humans and dogs as rabbits and machines and there have been a few suggestions for what intelligence is over the years in 1921 the psychologist V Henman said intelligence was the capacity for knowledge and knowledge possessed which sounds quite good on the surface until you realize that it also applies to libraries libraries can possess knowledge do libraries count as intelligent probably not in 1985 Marvin Minsky the cognitive scientist said that he thought intelligence was the ability to solve hard problems and that seems a bit more like it it also captures what a eye has already proved it can do here's Riya Hansel senior research scientist at deep mind in the last few years there's a huge number of different narrow specific things that programs can do as well as a human they can interpret your voice as well as a human they can maybe translate from English to French and back again almost as well as a human they can recognize things and images almost as well as a human these sort of narrow specific things that word narrow is not to downplay the transformative power of this sort of thing just in this series we've looked at energy conservation medical diagnosis and protein folding all of which certainly show the machines ability to solve hard problems and all of which are examples of narrow AI but real intelligence general intelligence that needs something else something more some scientists have described intelligence as the capacity to learn or to profit by experience others think it's about adapting and thriving in the environment you find yourself in but whoever you are smart disagree that's somewhere along the line intelligence is something about your ability to interact with an external environment and being able to adapt has to be part of it too so you can't be fully familiar with the environment you've got to be able to deal with unanticipated challenges that get thrown at you if you're intelligent in 2007 after going through hundreds of competing arguments Shayne leg one of the three co-founders of deep mind wrote an influential paper in which he and his co-author tried to pin down precisely what was meant by intelligence and here is the definition that they came up with intelligence measures an agent's ability to achieve goals in a wide range of environments and that is what they are aiming for in this building here's a reminder of what a senior research scientist Maurice Shanahan told us in Episode four the holy grail of AI research is to build artificial general intelligence so to build AI that is as good at doing an enormous variety of tasks as we humans are so we are not specialists in that kind of way you know a young adult human can learn to do a huge number of things and can indeed do an enormous number of things and can adapt to a huge number of different challenges you can learn to make food you can learn to make a company you can learn to build things to fix things you can do so many things to have conversations to rear children so although those things and we really want to be able to build AI that has the same level of generality as that and that's really still an open challenge we don't really know quite how to get there but if anyone has an idea of what it will take its dem assess our best the CEO and co-founder of deepmind will be talking to him in the next episode of our Adi but for now here is a little glimpse into his thinking I'm waiting to see a lot of key moments for example I think a really big moment will be when an AI system comes up with a new scientific discovery that's of Nobel prize-winning level that to me would be a big watershed moment so you know capable some kind of true creativity in some sense I think other big points will be when it can use language and converse with us in a naturalistic way it's capable of learning abstract concepts these are all things that I think a high-level cognitive abilities that we're nowhere near yet and I think will be big so I'm posting on the way now it's reasonable to be asking yourself how do you even approach such a colossal task where do you even start do you totally and completely believe that AGI as possible yes back to my tour guide core I cover to Luke he's the director of research at deep mind I believe that they will come where we will be at that stage right now we are not right now all we can do is go back from that and then have a hypotheses about the important problems the important algorithms that we need to do the important solutions like those key things we need to have and then start building more and more and more so let's take an example then let's say the first day that someone at deepmind decided they wanted to look at the problem of navigation so building an agent that you can drop into an environment yes where it's going do you just have like a big brainstorm on a whiteboard of all of the different possible aspects that might contribute to being able to build that agent yeah it starts with that because if someone must work on something like that then probably there will be a good number of people here who would be interested in the same thing we will start discussing okay what is the goal like when we say navigation you gave a particularly good example right like going from here to a given location how are we going to specify it what kind of environment this is what kind of control space does the agent have all these start affecting what kind of algorithms we should use and are we going to do this purely from vision are we going to do this in a simple environment in a grid world or we're going to start from a grid world and then we need to think about the path towards going to a 3d environment are we thinking about actually also putting this on a robot like with real vision like all that discussion starts happening there's massive then I mean just that one form is enormous it is but that's why it's also research a big part of it is trying to constrain the problem space like being able to write down and specify what what you want to do and then making sure that it is actually a challenging problem and there are good metrics that we can quantify or we're saying you're successful right exactly and that itself as I said is quite an iterative process and getting that critique getting that view from other researchers starts at that point in time because like when the research is at the idea space at the initial stages it's actually quite important to formulate the right problems and and sort of the right context Adi is not going to happen overnight it's the reason why as you heard on the tour there are so many different research groups in this building you're not going to crack Adi by attacking on only one front a formulating the right problems means going beyond individual skills like navigation and drilling down into the building blocks of intelligence and - from the beginning of deepmind if the goal is about a GI then it has to invoke control it has to involve an active algorithm and that is why you need to do reinforcement learning that is why you need to work on agents we need agents that can interact with their environment that can learn through trial and error what actions to take what karai calls its policy it's a fundamental thing that you would expect to find in any intelligent being humans dogs or agents and so if the scientists can get it right in one application the lessons learned should apply elsewhere the gist of training an agent is you can start completely from scratch and then slowly the agent builds that knowledge that strategy of how to achieve a certain task and it creates it builds it comes up with its own strategies it comes up with its own policy so then of course looking at it and trying to understand does that pose make sense does that buy is it doing that policy sometimes it's so surprising because it's something that we haven't thought about before sometimes you look at it and you see that need that doesn't make sense yes it it achieves something but it's clearly not optimal can you remember when you first realized or believed that AGI was possible so we tried these agents on Atari games and they didn't do much and we couldn't we couldn't make them work right like there was a there was a team of like like like several people there and we couldn't make them work and then slowly we started simplifying the problem and simplifying the problems and find the problem and we ended up with a very tiny simple trivial problem like really just like a five pixel by 10 pixel image and one pixel moving in that image and the agent trying to control that and once we reduced the problem to that which I call the emne stuff reinforcement learning really we could find the working solution we started having this deep reinforcement learning agents working right because it's a simple problem of course you can write a program to solve that but the idea was like try to do deep reinforcement learning and try to solve it that way try to solve it from pixels try to come up with a system that we think can generalize to more to two different problems more problems and once we saw that actually it was a matter of weeks we had ten or fifteen Atari games like being sold like from that tiny thing in the matter of weeks you go to Atari that that was a big moment that's what we keep in going that like we select more and more diverse set of problems that we think are important at the end for AGI that is a key point here intelligence is an agents ability to achieve goals in a wide range of environments so to get to AGI we need agents to solve harder and more diverse problems you're listening to deepmind the podcast a window on AI research the closer we get to AGI the more powerful and sophisticated this technology gets and the more we rely on it in our everyday lives the more dramatic the consequences could be of us misunderstanding the limitations of the algorithms and that is why in parallel to pushing the science forward researchers are also working to ensure right now the agents are reliable adaptable and crucially not corruptible [Music] you show the image of a bus to a neural network and it will say this is a bus right there's a bus in this image well it turns out that you can take the same image modify it a little bit which is almost invisible to a human eye but the neural network will say that it's actually an ostrich this adversarial attacks are things some most of the time that are invisible to the human eye that doesn't change the actual content of the image too much we don't perceive it but because these are algorithms and bastes they're they are sensitive to even very small fluctuations in the input data then it changes it changes the output why does that matter why do you want to stop that from happening in the real world ai well for two reasons one of them as I said is robustness because when we trained these algorithms we want them to be useful in real world and we train them on datasets that we captured from real world but like you cannot know exactly what's going to happen of course what kind of data is going to be at the end of today this algorithm is going to consume so we want to make sure that the algorithms are robust to these kinds of potential like noise like things that you want your algorithm to be robust to that right and from another point of view it's about safety being able to say that like someone maybe with an adversarial intent won't be able to change the output of this algorithm output of this neural network just by making very small adjustments to the inputs we have a we have a whole research group actually on that on working on rigorous and more robust artificial intelligence because yes these are real things in the end we are training these algorithms and as I said it's not just like looking at them but we're trying to do more quantifiable research on understanding why they are doing what they are doing and trying to interpret that and part of it is also understanding how robust they are [Music] we have already seen real examples of just how fragile intelligence can be researchers have shown that you can add a little bit of black tape to a stop sign that will trick a driverless car into speeding up the tape is so subtle that it would look innocuous to a human driver but it's just enough to make the algorithms inside the car miss read it as a 45 mile an hour speed limit sign instead of an instruction to stop other scientists have worked out how to fool facial recognition algorithms into thinking someone is mere Yakov it just by making them wear a specially designed pair of tortoiseshell glasses and the images that are used for medical diagnosis can end up giving wildly inaccurate results just if a slightly different brand of scanner was used to take them you want to reach a state where you can actually guarantee that things like that won't happen that's the main idea behind doing this research so not sort of try to defend against particular adversarial attacks or examples but actually come up with systems that are going to be robust no matter what from the other point of view of course if you are thinking about agents there's the whole safety issue what do you mean when you guys talk about safety what are you actually talking about if we have an intelligent algorithm making decisions for itself you want to have some sort of guarantees that I mean it's acting with its own policy it is aligned with what you intended it to do if you think that these algorithms continued learning all the time then we want that process to be also producing an agent that is aligned with what we have intended for it to do fill in the gaps for me never give an example of what you're trying to avoid it becomes quite technical in the sense that it's not like you are trying to avoid the agent from making a mistake right because mistakes happen right like betraying policies and they are not always optimal data to the writing all the time it's not about that really it's more about you have a learning algorithm and you want to make sure that it sort of conforms to certain boundaries of behavior in essence if we can develop safe robust ethical AGI then the impact could be staggering but as demmas mentioned so too could the discoveries on the way to Adi has agent learned to solve increasingly hard problems Trevor back is a product manager for the deep mind science program he played a key role in the Moorfields Eye Hospital collaboration that we talked about in episode 5 but he also has a hand in deciding what the future holds for deep mind and what challenges and opportunities lay in store for them to tackle next so we're really at the stage of exploring what other areas we should work on but the possibilities are endless if you look at the way the alpha fold system works there's nothing really specific in there to protein folding it's around understanding the way that atoms interact the way that you can build material from base concepts so looking at material design is a really exciting stage you know could you design or imagine a high-temperature superconductor that's been sort of brought to life through an AI algorithm right is there more efficient ways of looking for those types of materials why do we care about high-temperature superconductors so this is the amazing opportunity of working in in science so our previous work in healthcare has been focused on very specific problems and it takes a lot of time and effort and energy to build an AI system that works for those specific problems if instead you can spend your time and energy solving a fundamental science question then perhaps you can you know instigate a whole new field of interest and potentially impact a much wider array of problems so why is superconductivity a problem you know if you could solve superconductivity not only could you you know solve a lot of the energy problems by having a larger genetic field around fusion but you could also create a new type of computing system you know there's lots of opportunities that come from just a single breakthrough in one of these areas I think it's important not to understate this because the idea that you could have some kind of an impact on nuclear fusion for example the implications for the earth and humanity are just enormous right exactly I think this is the reason I came to D mind was the opportunity to have the greatest impact I could in the world and I think AI is one of those revolutionary technologies that amazingly could you know impact everything we do but also everything we think about doing in the future and the sort of opportunity to explore and search the space of opportunities is really something that's well designed for AI to do it's a very efficient search algorithm and so if you're able to set up the problem in such a way that is searchable via a are you really could have 10 a hundredfold type of increase in the opportunity for finding novel materials or finding anomalies in astronomical data you know finding new types of stars finding more black holes you know all these types of opportunities are available simply via the application of AI and it's not stupid to say that actually lots of the biggest problems that face humanity at the moment are science problems right like you know access to food and water climate change healthcare all of these things is stuff that AI can make progress in I think that's right I think a lot of the the sort of physical problems that exist in the real world are certainly things that if you can make a difference to some of the the sort of foundational science aspects you know you could reduce the cost of energy essentially down to zero or you could make food more readily available across the world and so that would really help society progress in a number of ways it's a tantalizing prospect solving intelligence and creating a G I will take the full range of research explored in this podcast and more memory reasoning logic learning language embodied cognition and more so much more and we're going to explore some of those ideas in our next episode when I meet with deepmind co-founder demis hassabis he tells us how he created the world's leading AI research outfit reveals what keeps him up at night and opens up about his hopes for the future I'm just fascinated and also troubled by the things around us that we seemingly don't understand all the big questions you know the meaning of life had the universe start what is consciousness all these questions which I feel like a blaring sort of klaxon in my mind that I would like to understand and my attempt at doing that is to build AI first if you would like to find out more about artificial general intelligence or explore the world of AI research beyond deep mind you'll find plenty of useful links in the show notes for each episode and if there are stories or a sources that you think other listeners would find helpful then let us know you can message us on Twitter or email the team at pod cast that deep mind calm you can also use that address to send us your questions or feedback from the series you