you have warned me not to get too close, it does look a bit hazardous! this roadblock dog has not been programmed to war, it is learning by itself, it is called reinforcement learning. what we have done here is given at this task to try to stand up and move forward.— at this task to try to stand up and move forward. when it does well, and move forward. when it does well. one _ and move forward. when it does well. one of— and move forward. when it does well, one of the _ and move forward. when it does well, one of the computer- well, one of the computer scientist gives it the digital equivalent of a pat on the head. it's doing better and better. is strong, i tried to put that in its way and it's stronger i am. have you trained it to make a beeline for me? i feel like i'm battling the ai here and the aia is winning! in just a few hours it's moving around by itself, even coping with the unexpected. i'm really feeling and instincts to help it but that of course is not good. i5 it but that of course is not aood. , ., it but that of course is not aood. , . ., , good. is learning, and maybe looks mean — good. is learning, and maybe looks mean sometimes - good. is learning, and maybe looks mean sometimes when| good. is learning, and maybe i looks mean sometimes when we put it around but we're just trying to see what can handle and help with behaviours. this robot dog _ and help with behaviours. this robot dog is — and help with behaviours. this robot dog is the _ and help with behaviours. this robot dog is the brainchild of professor peter opl. what's about reinforced learning is the bot collect its own data, we give it a score, we say good robot, bad robot, and over time it starts seeing when i guess the good robot signal and start behaving more like that. it sounds like bringing up a child. �* , ., sounds like bringing up a child. 2 ., ., ~ ., , child. it's a lot like raising a child or _ child. it's a lot like raising a child or a _ child. it's a lot like raising a child or a dog, - child. it's a lot like raising a child or a dog, very - child. it's a lot like raising i a child or a dog, very similar in that you don't get to just push data in, like with other learning methods, you have to let it generate its own data, and then you tell it, this was good, this was bad. it’s and then you tell it, this was good, this was bad. it's called generative _ good, this was bad. it's called generative ai, _ good, this was bad. it's called generative ai, the _ good, this was bad. it's called generative ai, the same - generative ai, the same technology behind a breakthrough in the way machines use human language. gptlr is machines use human language. gptli is the latest system from openai. gptlo is the latest system from 0 enai. , �* gptlo is the latest system from oenai. , ~ .,, gptlo is the latest system from oenai. , ~ ., openai. openai was formed eight ears auo openai. openai was formed eight years ago and _ openai. openai was formed eight years ago and it — openai. openai was formed eight years ago and it has _ years ago and it has revolutionised language based intelligence. it is revolutionised language based intelligence.— intelligence. it is a dream where tax _ intelligence. it is a dream where tax can _ intelligence. it is a dream where tax can flourish - intelligence. it is a dream where tax can flourish in l intelligence. it is a dream - where tax can flourish in front of you. where tax can flourish in front of ou. ., , where tax can flourish in front of ou. . , , . where tax can flourish in front of ou. , �* of you. last year openai released _ of you. last year openai released an _ of you. last year openai released an update - of you. last year openai released an update to i of you. last year 0penai| released an update to its technology called chatgpt, it can answer complex questions, tell you a joke or write a complex as a poem. it's not just what it is capable of that has changed the game, openai created a free version that anyone can use.— created a free version that an one can use. ., ,_ , anyone can use. for the systems we are developing, _ anyone can use. for the systems we are developing, to _ anyone can use. for the systems we are developing, to have - anyone can use. for the systems we are developing, to have a - we are developing, to have a big impact, we have to figure out how to make them accessible.— out how to make them accessible. . , , ~ accessible. peter advise openai durin: the accessible. peter advise openai during the company's _ accessible. peter advise openai during the company's early - during the company's early days. during the company's early da s. �* ., during the company's early da s. ~ ., ii'j ~ during the company's early da s. �* ., ::'j~ 11" , ~ days. around 2018, 2019, openai made about _ days. around 2018, 2019, openai made about saying _ days. around 2018, 2019, openai made about saying we _ days. around 2018, 2019, openai made about saying we think- days. around 2018, 2019, openai made about saying we think a - made about saying we think a lot of the key ideas are in place, maybe not all of them but a lot of them, but we're not doing a large—scale enough, we need more computers, more data, and things will emerge are more capable.— are more capable. that computing _ are more capable. that computing power - are more capable. thatl computing power arrived are more capable. that - computing power arrived in are more capable. that computing power arrived in 2019 when microsoft invested $1 billion in openai. we when microsoft invested $1 billion in 0penai._ billion in openai. we saw 0 enai billion in openai. we saw openai four— billion in openai. we saw openai four years - billion in openai. we saw openai four years ago - billion in openai. we saw openai four years ago as| billion in openai. we saw - openai four years ago as this extraordinary group of people in a research lab that we felt we could say they are on the right path. we could say they are on the right path-— right path. crucially, microsoft _ right path. crucially, microsoft provided l right path. crucially,j microsoft provided it right path. crucially, - microsoft provided it with one of the most computer networks in the world.— in the world. that partnership has really _ in the world. that partnership has really blossomed - in the world. that partnership has really blossomed in - in the world. that partnership has really blossomed in all. in the world. that partnership | has really blossomed in all the ways we had hoped, maybe even a little faster than we might have imagined.— little faster than we might have imagined. little faster than we might have imauined. ., , ., ., , have imagined. chatgpt now has more than _ have imagined. chatgpt now has more than 100 _ have imagined. chatgpt now has more than 100 million _ more than 100 million registered users but the latest version is not free, it's only available to subscribers and will be built into future microsoft products. it will be built into future microsoft products. if you are t in: to microsoft products. if you are trying to write _ microsoft products. if you are trying to write a _ microsoft products. if you are trying to write a document, i trying to write a document, create a powerpoint slide, go through your e—mail, this is now a tool that will help you, you can ask it to compose a first draft, you can ask you to take a word document and turn it into powerpoint slides and as i always say, make sure you read them before you present them, you are still in charge! today's ai revolution has been driven by the work of three computer scientists. jeffrey, jansen and joshua developed artificial neural networks to help computers recognise patterns and mimic the human brain. , , ., , brain. there is still a very larae brain. there is still a very large gap _ brain. there is still a very large gap that _ brain. there is still a very large gap that we - brain. there is still a very i large gap that we observed between the capability of ai between the capability of a! system and what we observe in animals and humans, and basically the quest has been to bridge this gap, to figure out what we are missing, what type of intelligence to humans and animals exhibit, and the good news is we are making a lot of progress. news is we are making a lot of progress-— progress. they have become known as — progress. they have become known as the _ progress. they have become known as the godfathers - progress. they have become known as the godfathers of i progress. they have become l known as the godfathers of ai. aia is about building machines that are intelligent. what does it mean to be intelligent? people may disagree, but the way i and many others like to define it is the ability to learn, to understand and make decisions. learn, to understand and make decisions-— decisions. the godfathers laid the foundations _ decisions. the godfathers laid the foundations for _ decisions. the godfathers laid the foundations for machines | the foundations for machines that could learn. what would power the next step was data, but on a scale never possible before. , �* but on a scale never possible before. , ~ _ , before. there is the ai system like tat gtp, _ before. there is the ai system like tat gtp, those _ before. there is the ai system like tat gtp, those are - before. there is the ai system | like tat gtp, those are trained on perfect data but for those systems to be useful, they have to be trained on something like 1000 billion, i to be trained on something like 1000 billion, 1 trillion words, something of that order of magnitude, something that would take on the order of 20,000 years for a human to read, so they require an enormous amount of data, pretty much a third of the internet.— the internet. those billions of words can _ the internet. those billions of words can include _ the internet. those billions of words can include material. the internet. those billions of. words can include material most people would rather avoid. mr; people would rather avoid. my concern people would rather avoid. ij�*i concern is people would rather avoid. m: concern is the people would rather avoid. m; concern is the current emphasis on scale and artificial intelligence, which is to say creating enormous training datasets of content, insufficient attention has been given to what is in them, in many cases it is just garbage in an garbage out. if you are harvesting the entire internet, you will have so many horrifying things, including in there, is a part of this really gets to you, the core production ethos of how ai is made, which is collect it all and figure it out later. iii and figure it out later. if machines are learning from what is on the internet, material thatis is on the internet, material that is harmful needs to be filtered out, and much of that work still needs the scrutiny and judgement of humans. and 'udgement of humans. there are and judgement of humans. there are certainly _ and judgement of humans. there are certainly countries _ and judgement of humans. there are certainly countries where - are certainly countries where this type of outsourced labour is most commonly located, china, malaysia, kenya, some examples but indeed it is a growing market for people to be really doing this extremely difficult, extremely traumatising work, so in a sense, it maps to a global north, globalsales sense, it maps to a global north, global sales dynamic of extraction and exploitation of labour for these systems. this. r , the place where we were engaged for at least four months, it doesn't bring any find — months, it doesn't bring any find memories in as far as the uroject — find memories in as far as the uroject is _ find memories in as far as the project is concerned,. find memories in as far as the project is concerned, .- project is concerned,. richard 0 erated project is concerned,. richard operated as — project is concerned,. richard operated as a _ project is concerned,. richard operated as a content - operated as a content moderator. the company had been hired by openai to sift through huge volumes of text and remove anything you did not want it artificial intelligence learning from.- artificial intelligence learning from. artificial intelligence learnin: from. ., learning from. the text of writin: learning from. the text of writing we _ learning from. the text of writing we encountered i learning from. the text of. writing we encountered was learning from. the text of- writing we encountered was very disturbing, very depressing things that caused anxiety and mental breakdown.— things that caused anxiety and mental breakdown. richard was aid mental breakdown. richard was paid less than _ mental breakdown. richard was paid less than £2 _ mental breakdown. richard was paid less than £2 per _ mental breakdown. richard was paid less than £2 per hour. - mental breakdown. richard was paid less than £2 per hour. he i paid less than £2 per hour. he and his colleagues were repeatedly exposed to descriptions of murder, best reality and abuse. i remember encountering _ reality and abuse. i remember encountering a _ reality and abuse. i remember encountering a situation - reality and abuse. i remember| encountering a situation where a human being is having intercourse with an animal and even as this happens, there is an inference was right over there watching as the whole thing is taking place, and it was very disturbing, it was content that you would not want to dwell in your mind, and when you are working on this exercise for a period of nine hours per day, with very minimal breaks.- hours per day, with very minimal breaks. . ., , minimal breaks. richard says he asked the company _ minimal breaks. richard says he asked the company for - minimal breaks. richard says he asked the company for support i asked the company for support but did not get any. the;r asked the company for support but did not get any.— but did not get any. they had very poor _ but did not get any. they had very poor facilitation - but did not get any. they had very poor facilitation in - but did not get any. they had very poor facilitation in as - very poor facilitation in as far as rendering that support. we never received counselling sessions. ., , sessions. the two companies arted sessions. the two companies parted company _ sessions. the two companies parted company in _ sessions. the two companies parted company in 2022. - sessions. the two companies - parted company in 2022. richard lost hisjob, along parted company in 2022. richard lost his job, along with three former colleagues, he is now petitioning the kenyan government to improve conditions for tech company workers. ., ~' conditions for tech company workers. ., ~ ., , ,., workers. the work we did was so im actful workers. the work we did was so impactful in _ workers. the work we did was so impactful in as — workers. the work we did was so impactful in as far _ workers. the work we did was so impactful in as far as _ workers. the work we did was so impactful in as far as creating - impactful in as far as creating a safe space for the people who are using chatgpt right now. what content moderators are going through is something that is not fairand going through is something that is not fair and it is unbelievable.- is not fair and it is unbelievable. . ., , is not fair and it is unbelievable. , ., unbelievable. the company told anorama unbelievable. the company told panorama that _ unbelievable. the company told panorama that employees - unbelievable. the company told panorama that employees were | panorama that employees were provided counselling by professional mental health therapist. the work with openai was a pilot project that ended early as soon as the team reported concerns. openai said its mission is to build safe and beneficial artificial intelligence, and that it believes the challenging work of content moderator needs to be done humanely and willingly, so it has established its own ethical and wellness standards for data anna —— annotated. machines are sifting through and analysing fast quantities as the accelerating pace of ai development, already leading to computers acquiring the first human skill, language. we are at a critical _ human skill, language. we are at a criticaljuncture _ human skill, language. we are at a criticaljuncture in - human skill, language. we are at a criticaljuncture in the - at a criticaljuncture in the sense that we now have machines that understand language, can manipulate language, even past four humans for a while with most people, and that was a milestone for aia.- milestone for aia. having mastered _ milestone for aia. having mastered language, - milestone for aia. having mastered language, ai i milestone for aia. having mastered language, ai is| milestone for aia. having - mastered language, ai is now mastered language, a! is now beginning to master other human skills like reading emotions. this is the basis of diplomacy, we can analyse everything that is happening in yourface, where you are looking. this team at university - where you are looking. this team at university has - team at university has developed a system that analyses tiny movements in your face to analyse how you are feeling. face to analyse how you are feelina. ., , , .,, feeling. your eyes opening, there is a — feeling. your eyes opening, there is a muscle _ feeling. your eyes opening, there is a muscle that - feeling. your eyes opening, l there is a muscle that comes from the back of your skull and attaches to your eyelids. what attaches to your eyelids. what does that mean _ attaches to your eyelids. what does that mean you _ attaches to your eyelids. what does that mean you are - attaches to your eyelids. what does that mean you are feeling? that is quite often when people want to show that they are very excited, that they are interested, but it can also be a sign of fear,. i interested, but it can also be a sign of fear,.— a sign of fear,. i was interested, - a sign of fear,. i was interested, i- a sign of fear,. i was interested, i wasn'tl a sign of fear,. i was - interested, i wasn't scared. a sign of fear,. i was _ interested, i wasn't scared. we start by looking at the different facial muscle actions, that's a good exercise, where they are looking, so your gaze on selection and the tone of voice as well as what you are saying. from those basic ingredients of your behaviour, we then determine, for example different local conditions you might have or emotion you might be displaying. this might have or emotion you might be displaying-— be displaying. this tack is currently _ be displaying. this tack is currently being _ be displaying. this tack is currently being trialled i be displaying. this tack is i currently being trialled with nhs trust in nottinghamshire to help assess the mental health of pregnant women.— of pregnant women. they are ve bad of pregnant women. they are very bad at — of pregnant women. they are very bad at picking _ of pregnant women. they are very bad at picking up - of pregnant women. they are | very bad at picking up gradual changes, this technology is objective and repeatable and it will improve as you start doing the treatment.— will improve as you start doing the treatment. you think people are ready for _ the treatment. you think people are ready for al _ the treatment. you think people are ready for al to _ the treatment. you think people are ready for al to assess - the treatment. you think people are ready for al to assess your l are ready for al to assess your mental health? i are ready for al to assess your mental health?— mental health? i don't think everybody — mental health? i don't think everybody will _ mental health? i don't think everybody will want - mental health? i don't think everybody will want to - mental health? i don't think everybody will want to use i mental health? i don't think. everybody will want to use it, but that's ok.— but that's ok. artificial intelligence _ but that's ok. artificial intelligence is - but that's ok. artificial intelligence is now - but that's ok. artificial. intelligence is now finding but that's ok. artificial - intelligence is now finding its way into every aspect of our lives. in the workplace, it's starting to take on new roles. artificial one of the key things we have picked up on is the increasing use of us to pick up on al the increasing use of us to pick up on a! at work. decisions about how people do they work, who to do it with, how quickly they have to do it and then also even decisions significant as whether or not they use or keep a job. —— might lose or keep theirjob. in 2018, taxi driver alexander joined uber. like most uber drivers he had never actually met his boss.— met his boss. you work a lot and the only _ met his boss. you work a lot and the only source - met his boss. you work a lot and the only source of- and the only source of information could be your own experience and maybe chatting with other drivers.— with other drivers. everything is done through _ with other drivers. everything is done through his _ with other drivers. everything is done through his uber - with other drivers. everything is done through his uber app. thenin is done through his uber app. then in the summer of 2021 he got an e—mail. it accused him of fraud. got an e-mail. it accused him of fraud. ., got an e-mail. it accused him of fraud-— got an e-mail. it accused him of fraud. you are guilty or not iuil , of fraud. you are guilty or not guilty. there _ of fraud. you are guilty or not guilty, there is _ of fraud. you are guilty or not guilty, there is no _ of fraud. you are guilty or not guilty, there is no grey - of fraud. you are guilty or not guilty, there is no grey area i guilty, there is no grey area in this. i received a warning from the board. we detected fraudulent activity associated with your account. please stop, this is your first warning. the automated — this is your first warning. the automated message that he might have manipulated trip details or falsely claimed money he wasn't entitled to. but it could be specific. i contacted uber and _ could be specific. i contacted uber and the _ could be specific. i contacted uber and the answer - could be specific. i contacted uber and the answer on - could be specific. i contacted uber and the answer on the l uber and the answer on the other end of the line was that, well, we don't know exactly what you have done wrong. it was something detected by the artificial intelligence. however, don't worry about it, just stop whatever you are doing wrong. just stop whatever you are doing wrong-— just stop whatever you are doini wroni. �* ., ., doing wrong. another warning followed. he _ doing wrong. another warning followed. he was _ doing wrong. another warning followed. he was on - doing wrong. another warning followed. he was on the - doing wrong. another warning | followed. he was on the verge of being fired. i followed. he was on the verge of being fired.— of being fired. i don't know what i've — of being fired. i don't know what i've done _ of being fired. i don't know what i've done wrong. - of being fired. i don't know what i've done wrong. i - of being fired. i don't know. what i've done wrong. i didn't change anything in my work. how can i stop something if i don't know. can i stop something if! don't know. ., ., ., know. he again complained to uben know. he again complained to uber- he _ know. he again complained to uber. he received _ know. he again complained to uber. he received another - uber. he received another message more than a year later. this time it was apology. it was written by uber, black and white. they recognised that the artificial intelligence made a mistake in our case. 50 it is a single mistake madejust single mistake made just against single mistake madejust against alexandru, or maybe, maybe the artificial intelligence failed to make the judgement correctly in his case, maybe failed in a few other cases. it case, maybe failed in a few other cases.— other cases. it turns out alexandru _ other cases. it turns out alexandru wasn't - other cases. it turns out alexandru wasn't the . other cases. it turns out i alexandru wasn't the only other cases. it turns out - alexandru wasn't the only one. in april, uber and another taxi company were taken to court in the netherlands stop the court said drivers had a right to know how decisions about them were being made. and criticised uberfor were being made. and criticised uber for using automated uberfor using automated decision—making uber for using automated decision—making without human impact. —— input. this decision-making without human impact- -- input-— impact. -- input. this is a cautionary _ impact. -- input. this is a cautionary tale _ impact. -- input. this is a cautionary tale when - impact. -- input. this is a cautionary tale when too i impact. -- input. this is a - cautionary tale when too much responsibility is given to geac ai responsibility is given to geac alto make hugely significant decisions about people. this was a robo firing case where basically the ai was deciding that some drivers had spent too long waiting for products that they were going to deliver and there was some fraud implicated in that, that they were ripping off the company.— off the company. uber says it understands _ off the company. uber says it understands deactivating - off the company. uber says it understands deactivating a i understands deactivating a driver's account is a very serious decision and the specially trained teams involved in the process to ensure its approach is transparent and fair. injune, the government announced an ai the government announced an a! task force with £100 million in funding. it says it will bring together experts to research a! safety and help the uk be a global ai superpower. at! safety and help the uk be a global ai superpower. globalai superpower. ai can brini globalai superpower. ai can bring huge — globalai superpower. ai can bring huge benefits - globalai superpower. ai can bring huge benefits for- globalai superpower. ai can bring huge benefits for our l bring huge benefits for our society and public services, but of course, it needs to be developed safely, securely and fairly. developed safely, securely and fairl . ., ., developed safely, securely and fairl. ., ., , ., , ., fairly. there are no plans for new laws — fairly. there are no plans for new laws to _ fairly. there are no plans for new laws to control - fairly. there are no plans for new laws to control ai - fairly. there are no plans for new laws to control ai at - fairly. there are no plans forj new laws to control ai at the new laws to control a! at the moment but the government's published a white paper saying existing regulators will be responsible for policing it. intelligence is kind of the most desirable commodity that we are lacking. so this will make people more productive, more creative and my view is this may bring a new kind of renaissance, a new era of enlightenment for humanity. but it has to be done right. ai development is mostly focused on mastering existing human skill, but could the technology take machines beyond human? hello, good to meet you. thank you for having us. thanks very much. neuroscientist alexander has spent a decade trying to understand how the rain works. iiii trying to understand how the rain works-— trying to understand how the rain works. if we want to build intelligent _ rain works. if we want to build intelligent machines, - rain works. if we want to build intelligent machines, maybe i rain works. if we want to build i intelligent machines, maybe we want to make things that act more like human brains. that is kind of what the into neuroscience actually. this ear his neuroscience actually. this year his team _ neuroscience actually. this year his team had - neuroscience actually. this year his team had a - year his team had a breakthrough. using the same a! take that can understand language, they built a computer that can read minds. hide language, they built a computer that can read minds.— that can read minds. we scan --eole's that can read minds. we scan people's brains _ that can read minds. we scan people's brains while - that can read minds. we scan people's brains while they - that can read minds. we scan| people's brains while theyjust listen to stories so we track how their brains respond while they are listening to hours and hours of stories. the they are listening to hours and hours of stories.— hours of stories. the team in texas has — hours of stories. the team in texas has trained _ hours of stories. the team in texas has trained the - hours of stories. the team in texas has trained the ai - hours of stories. the team in texas has trained the al on i texas has trained the al on their own brains. as they listen to stories inside the scanner, the computer watches what happens. what sort of brain activity are you looking for? ~ ., ., , ., for? we are looking for brain activity that _ for? we are looking for brain activity that is _ for? we are looking for brain activity that is related - for? we are looking for brain activity that is related to - activity that is related to specific ideas or specific words that appear in the stories so for example, whenever you hear somebody talk about parking a car there are certain patterns of activity in the brain that will be present that reliable —— reliably correspond to that kind of idea and we're trying to up that mapping from this very large data set. the computer looks for patterns in the vast amounts of data from the scanner. with enough training it can translate brain activity into words. to it can translate brain activity into words-— it can translate brain activity into words. to see how well it works, into words. to see how well it works. we _ into words. to see how well it works, we picked _ into words. to see how well it works, we picked a _ into words. to see how well it works, we picked a story - into words. to see how well it works, we picked a story for i works, we picked a story for the lead scientist to listen to. ~ ., the lead scientist to listen to. . ., ., . ., , to. the war of the worlds... headphones _ to. the war of the worlds... headphones on, _ to. the war of the worlds... headphones on, he - to. the war of the worlds... headphones on, he went - to. the war of the worlds... | headphones on, he went into to. the war of the worlds... - headphones on, he went into the scanner. , , ., scanner. margins in the pit had turned the _ scanner. margins in the pit had turned the heat _ scanner. margins in the pit had turned the heat read _ scanner. margins in the pit had turned the heat read on - scanner. margins in the pit had turned the heat read on them. | turned the heat read on them. the deep _ turned the heat read on them. the deep throbbing sound, the silver— the deep throbbing sound, the silver pencil of light and... trees— silver pencil of light and... trees and _ silver pencil of light and... trees and houses stretching as far as the eye can see were scorched. far as the eye can see were scorched-— far as the eye can see were scorched. �* , , scorched. and this is the text ofthe scorched. and this is the text of the al _ scorched. and this is the text of the ai created _ scorched. and this is the text of the ai created from - scorched. and this is the text of the ai created from the . of the ai created from the brain scan.— of the ai created from the brain scan. , ., brain scan. the decoded version which misses _ brain scan. the decoded version which misses a _ brain scan. the decoded version which misses a lot _ brain scan. the decoded version which misses a lot of _ brain scan. the decoded version which misses a lot of things - which misses a lot of things but got some of the good stuff, fired the bullets and then i had a huge crack is a large piece of concrete hit my forehead and a giant ball of fire. ., ., ., , fire. the general idea of big noise got — fire. the general idea of big noise got hurt _ fire. the general idea of big noise got hurt is _ fire. the general idea of big noise got hurt is there - fire. the general idea of big noise got hurt is there in i fire. the general idea of big i noise got hurt is there in both of them. and then also the idea that something was on fire. the scorched less _ that something was on fire. the scorched less as _ that something was on fire. the scorched less as a _ that something was on fire. tue: scorched less as a giant ball of fire, it captured all those things but clearly not the right words. 50 things but clearly not the right words.— things but clearly not the riiht words. ., ., , right words. so far it can only anal se right words. so far it can only analyse the — right words. so far it can only analyse the brains _ right words. so far it can only analyse the brains of - right words. so far it can only analyse the brains of a - right words. so far it can only | analyse the brains of a handful of volunteers but they hope it will unlock the secrets of how our minds work. tide will unlock the secrets of how our minds work.— will unlock the secrets of how our minds work. we are really -iushin our minds work. we are really pushing on — our minds work. we are really pushing on a _ our minds work. we are really pushing on a lot _ our minds work. we are really pushing on a lot of _ our minds work. we are really pushing on a lot of our - our minds work. we are really pushing on a lot of our effort i pushing on a lot of our effort in the lab and towards using this to actually understand the brain better. that is our scientific goal. we want to know how do our brains work twitter how do our brains process language, how do we think? , ., , ., think? sounds great but what ha--ens think? sounds great but what happens if — think? sounds great but what happens if it _ think? sounds great but what happens if it ends _ think? sounds great but what happens if it ends up - think? sounds great but what happens if it ends up in - think? sounds great but what happens if it ends up in the l happens if it ends up in the wrong hands?— happens if it ends up in the wrong hands? happens if it ends up in the wroni hands? , ., wrong hands? some people are scared or think the _ wrong hands? some people are scared or think the thought - scared or think the thought police is coming. i think it is a fair reaction to this to say this is scary, i don't want this is scary, i don't want this to happen. the first thing we thought when we got this working was this is fantastic, it is working! and then oh, my god, this is working.— god, this is working. other thou . ht god, this is working. other thought coming? _ god, this is working. other thought coming? not - god, this is working. other thought coming? not yet. | god, this is working. other- thought coming? not yet. none ofthe thought coming? not yet. none of the thought _ thought coming? not yet. none of the thought technologies - thought coming? not yet. none of the thought technologies we | of the thought technologies we have would be effective at actually policing people's thoughts. actually policing people's thoughts-— actually policing people's thou . hts. ., thoughts. the mind reading ai thoughts. the mind reading al was trained — thoughts. the mind reading al was trained using _ thoughts. the mind reading al was trained using early - was trained using early versions of open a! technology but it is not as open as it was before microsoft invested. independent research is no longer have transparent access to its latest tech and the data that trained it. there is less transparency than there was and a lot of people are not able to use chatgpt to do the things they wanted to. i use chatgpt to do the things they wanted to.— they wanted to. i think fundamentally - they wanted to. i think fundamentally even i they wanted to. i think i fundamentally even when you think about where law and regulation will go, the world will need an approach that ensures that al systems are safe, secure and transparent. we will have to figure out exactly what transparency means. they will be a balance as there are in other areas because you want to foster competition and that doesn't mean show your competitor everything about what you're doing. as chatgpt was taking the world by storm, leading figures in the world of artificial intelligence signed a statement warning of an existential threat. comparable they said to pandemics or nuclear war. one of the signatories wasjoshua of the signatories was joshua ben teo'o. of the signatories was joshua iten teo'o-— of the signatories was joshua ben teo'o. , ., ,~ ben teo'o. the existential risk was what _ ben teo'o. the existential risk was what happens _ ben teo'o. the existential risk was what happens if _ ben teo'o. the existential risk was what happens if we i ben teo'o. the existential risk was what happens if we read i was what happens if we read kate machines that are smarter than us that are not aligned with our needs and could potentially harm us just like a new species on earth that would be smarter than us might do. but the godfathers of ai don't always agree. this professor did not sign the agreement. fight; did not sign the agreement. any technology _ did not sign the agreement. any technology can _ did not sign the agreement. in; technology can be used as good or bad. the existence of bad actors is not new but the technology that enables bad actors to do bad things is the same technology that protects them. the same technology that exists today. there are people who are trying to use social networks like facebook and instagram and others to deceive disinformation, to corrupt the political process and the democratic process. the best protection we have against all of those attempts today makes a massive use of ai.— massive use of ai. even those at the cutting _ massive use of ai. even those at the cutting of _ massive use of ai. even those at the cutting of ai _ massive use of ai. even those at the cutting of ai research . at the cutting of a! research have been at how quickly technology has evolved. i could not have expected _ technology has evolved. i could not have expected that - technology has evolved. i could not have expected that we i technology has evolved. i could. not have expected that we would progress as quickly for al not have expected that we would progress as quickly for a! as a few years ago. i wish i had been paying more attention to the questions of loss of control but now that i do, i feel a sense of responsibility to explain my understanding of what could go wrong. it is to explain my understanding of what could go wrong.— what could go wrong. it is no loner what could go wrong. it is no longer in _ what could go wrong. it is no longer in the _ what could go wrong. it is no longer in the hands - what could go wrong. it is no longer in the hands of i what could go wrong. it is no longer in the hands of the i longer in the hands of the godfathers. it comes to the existential crisis letter, a couple of colleagues of yours assigned it. what do you think of that letter? i assigned it. what do you think of that letter?— of that letter? i think it is iood of that letter? i think it is good to — of that letter? i think it is good to imagine - of that letter? i think it is good to imagine the i of that letter? i think it is| good to imagine the worst because that is the best way to avoid it. but! because that is the best way to avoid it. but i don't think that the worst possible thing is necessarily the most probable thing. i think the biggest danger is not what machines will do to people but what people will do with machines to other people. that is the story notjust of technology, that is the history of humanity. technology, that is the history of humanity-— of humanity. machines are iiettin of humanity. machines are getting smarter— of humanity. machines are getting smarter and i of humanity. machines are getting smarter and ai i of humanity. machines are getting smarter and ai is i of humanity. machines are i getting smarter and ai is now all around us, butjust because it can do something doesn't mean it should. and maybe the question is how will we live alongside the machines that could become smarter than us? live from washington, this is bbc news. billions of people around the world are expected to watch the highly—anticipated women's world cup final, celebrating soccer, while issues of pay and inclusion loom large. hello i'm caitriona perry. you're very welcome. it's been the remarkable sporting tournament which has inspired, broken records, and sparked a conversation about opportunity and equality. the women's world cup wraps up on sunday, with a tantalising grand final pitting england's lionesses against spain's la roja. the month—long competition has been played in australia and new zealand by 32 teams for the first time, drawing fans from all over the world to cheer, cry, and celebrate the game. ahead of the big match, there's growing excitment in the two countries which have booked a spot in the final. there were wild reactions when england made it through, as you can see here. now, ahead of sunday's match, some young fans of the lionesses have a message for the team. i would like to see you win for the first time ever. good luck, russo, to win the final. good luck. i hope the other team. doesn't score anything. good luck. hope you win the world cup. you should do good because you are probably the best women's world cup team there. i don't know what's brought you so far. but i'm very proud. you've done really well— throughout the whole tournament and you've done really well to get this far, i so i'd like to say good luck. good luck. you can win this.