Skip to main content
Loading Explore guides…
Ethical AI & AGI · Acedra
Home
Explore
Computer Science
Ethical AI & AGI
Sign in to save
Chat
If YouTube asks you to sign in, open the video on YouTube.
Watch on YouTube
Reload player
Transcript
105 segments
Copy
Hide video
0:00
catch this video ad free and uncensored only on nebula is it going to destroy the world that's the question on a lot of people's lips AI is coming they say it's inevitable they say we'd better teach it to be good and hope it listens can we make ethical AI the future may [Music]
0:37
[Music] depend the AI people are most worried about is artificial general intelligence or AGI a digital Consciousness that surpasses humanity and might decide to do away with us a lot of money effort and anxiety is right now being poured into making sure AGI aligns with human values but I'd like to put the alignment problem to one side for the moment we will come back to it later AGI does not currently exist what we do have are a lot of very different tools that are all
1:26
called AI we have algorithms that judge people on their parole hearings job applications and credit scores we have security systems that can recognize faces and voices and we have text and image generators like darly and chat GPT we are going to be talking about all of that because before we get anywhere close to Skynet there are ethical dilemmas that arise for regular AI we know that AI can produce unethical outcomes for example in 2018 Amazon experimented with using it to review job
2:00
applications and rate them from 1 to 5 they had to scrap it when it started downgrading applications from women so we know that things can go wrong is there anything we can do about it there are two broad approaches the first is build it better just build better AI for example data scientist Fabian B gang says we need to recognize algorithmic decision making has two steps in Step One the model makes a prediction like this resume is a five out of five in step two it allocates something on the
2:39
basis of that like therefore you should give this person a job these are really two separate tasks and therefore there are two different ways things can go wrong in Step One prediction the model might be inaccurate like Amazon's AI it looked at all the resumés submitted to the company in the previous 10 years and learned the patents which might sound good but Tech is a very male-dominated field so the system concluded from its limited training data that men must be
3:10
better applicants because they get offered more jobs and that's not true so if the problem is accuracy then more data and more diverse data could help in step two allocation the model might be unfair like Amazon's AI offered jobs mainly to men so we just constrain what the can do for example tell it that if 30% of applications come from women 30% of jobs need to be offered to women too if the problem is fairness just tell the computer to be fairer and we've done it
3:42
we've made ethical Ai and all it took was one bit of code if evil then [Applause] [Music] [Applause] don't but maybe it's not that easy here's an ethical dilemma suppose we are building an AI to help a university decide which students should be offered scholarships based on their resumés the university wants to save money by accurately predicting which students will achieve a certain grade level that represents a good use of funds and suppose at this particular University
4:22
that students of color drop out more than white students and are therefore more likely to represent a loss of scholarship ship money the AI we build will learn that pattern however suppose that students of college drop out more because there's problems with faculty racism and lack of affordable housing and harassment by campus police they do drop out at higher rates and race is a predicting Factor but it's not their fault so should we alter the model to ignore race that seems fair
4:59
but it would make step one prediction less accurate and undermine the purpose for which the university wants to use the AI as for step two allocation we could tell the model that if 30% of applications come from students of color then 30% of offers need to go there too if racism then don't but that would mean some scholarships get offered to students who really weren't the most likely to succeed which again undermines the point of building it and you might
5:28
say well that doesn't matter it's fairer that way and personally I'd agree with you but now we're not talking about technology we're talking about political philosophy because it turns out there's a tradeoff between accuracy and the kind of society we should live in and the trouble with that is not everyone agrees what kind of society we should live in this dilemma can come up a lot another example might be if you search Google images for CEO should it show you
5:58
pictures that accurately reflect reflect most CEOs which is to say pictures of men and therefore risk reinforcing sexist biases or should it show you images of the kind of world we want to live in where CEOs of all genders are guillotined for their crimes so okay how to make ethical AI we've got build it better and there are some ideas there but also some tradeoffs so maybe we also need police it better AI can never be perfect so when it goes wrong we need to
6:31
make sure that people have options in the novel The Trial by France Kafka the protagonist Joseph Kay is arrested and as he moves through the criminal justice system he's never told what crime he's accused of the story is horrifying because Joseph is rendered powerless by the bureaucracy he can't take action and he can't even find out what action he could take until eventually he is executed and never told why philosopher Kate venberg says that when AI goes wrong it's like a Kafka novel you get
7:05
denied something and you don't know why and if you don't know you can't improve if your resume gets rejected by the recruitment AI that doesn't give you any feedback you can't write a better resum next time not only does that make you powerless she says it'll probably undermine people's trust in institutions that use AI which could be a big problem if we're talking about say a court so she proposes the right to an explanation if an AI makes a decision that affects your life you should be
7:35
entitled to know why it did that in some places like the EU this is already a real legal right but here's another dilemma if I own the recruitment software that rejects your resume and you say explain this what kind of explanation do you want me to give do you want me to tell you how the model Works generally like do you want me to show you the code or tell you what it did specifically in your case there might be some practical trade-offs to be made here if I'm a big company then I
8:12
might not have the time to give you individual feedback the most helpful kind of explanation would probably be what's called a counterfactual one that's where we say if you had done this differently the AI would have given you what you want like if you hadn't spelled the company name wrong on your application it would have given you the job that kind of explanation is useful because you can do better next time obviously though if we're giving counterfactual explanations of
8:39
algorithmic decisions those counterfactuals need to be true it needs to be true that if you had spelled the company name right the AI would have given you the job otherwise that's not actually a useful explanation and that can present a problem because a lot of AIS are what's called blackbox models they use multiple layers of nonlinear programming and they're so complicated that their outputs are a mystery even to their creators we could guess that if you spelled the name right it'll give
9:08
you the job but how can we prove that to you one solution to this is quite fun have you ever heard the nursery rhyme about the old lady who swallowed a fly so she swallowed a spider to catch the fly it turns out you can kind of do that with AI if we have a black box model whose Behavior we want to explain we can build a second surrogate model to approximate a simpler version of what the first one did and get counterfactual explanations from that it's kind of like
9:36
when police hire actors to do crime scene reenactments and show what probably happened like it's not bulletproof but it is an option however here's another dilemma suppose we build an AI whose outputs we know are unfair like we deliberately build the racism machine and we put it in charge of our company's recruitment it's a black box model so when a candidate of color gets rejected and requests their right to an explanation we turn to our surrogate model which tells
10:09
lies we build the racism machine and then we build the Uncle Tom machine to tell everybody that the racism machine just has some legitimate concerns and by calling it The racism machine you're actually silencing artificial voices which is not very tolerant of you so you use your right to an explanation but the the explanation you get is and you probably don't have the time or expertise to prove it this process is called Fair washing and it was first proposed by a group of scientists in a
10:43
paper published in 2019 not content with establishing the theoretical possibility they did it they built an AI they knew was biased they deliberately built the racism machine and then they built a second s model that they called laundry ml to launder the outputs of the first one turns out you can make the results seem fair even when you know they're not weirdly after that paper was published laund ml got a job writing for the [Music] telegraph okay how to make ethical AI
11:24
we've tried build it better and police it better and there are options there but also limitations technical limitations yes but also there seems to be a deeper problem I'd like to talk about the penis detection
11:48
machine Fair washing isn't just a technical snag the reason it's an issue is ordinary people wouldn't have the ability to ch challenge institutions who misuse AI so not a tech problem a problem of unequal power and indeed a whole Suite of problems present themselves when we remember that Ai and all technology is used in an already imperfect Society a fantastic illustration of this is the penis detection machine if you've flown on a plane in the last few years then you've
12:25
probably seen one it's the thing that you step into at airport security and you put your hands up and it goes its proper name is the full body scanner but we in the transfem community know it as the dick detector the spotter the Wang radar or the penis detection machine if you're cisgender which is to say not transgender then you might not know this but just before you step into that scanner someone on the other side looks at you and then presses one of two buttons blue or pink the
12:59
scanner looks for lumps that might indicate something hidden under your clothing which if you are a transgender woman with a penis and breasts presents a dilemma if they press the blue button for boys then the machine thinks that your titties are the bomb if they press the pink button for girls then it's unexpected item in shagging area you might have to answer some humiliating questions possibly out yourself as trans and maybe get groped by airport security and we do laugh
13:29
about it but going through that in front of my parents was one of the most humiliating experiences of my life the penis detection machine sees the presence of a penis on a woman as strange it does that because it was built by CIS people if I'd built it I'd make it so that it called every woman strange if she didn't have a penis which would be extremely funny and also makes the point that technology encodes ways of seeing the real world is not simple and black and white and doesn't
14:06
always correspond to Binary gender but the machine expects that moreover by being part of a system of surveillance and occasionally humiliation and assault it helps enforce binary gender when all you have is a hammer all your problems look like nails we often assume that machines like that detect information that's already there but they also do the opposite project information onto the body in a process that scholar Simone Brown calls digital epidermalization the machine tells you what certain
14:45
features of your body signify like that's strange you the person whose body it is might be like this isn't strange for me this is normal and I like it but people assume that machines are neutral so they listen to The Machine more than they listen to you and that can land you in trouble the term epidermalization comes from philosopher France Fannon who spent some time in France in the 40s and 50s and found that people called him strange unwelcome and dangerous because he was
15:20
black they weren't just detecting the color of his skin they were projecting meaning onto it and this could happen anywhere on the bus at work he was never secure because at any moment he could have the meaning of his being assigned to him by a white person and he said that's part of what it means to be in a racial minority you don't get to stand up and say this is me this is my deal white people tell you what your deal is digital epidermalization is the exercise
15:51
of power cast by the disembodied gaze of certain surveillance Technologies for example identity card and E passport verification machines that can be employed to do the work of alienating the subject by producing a truth about the body and one's identity or identities despite the subject's claims what does this have to do with AI well consider something like facial recognition Ai and in particular automatic gender recognition in theory that's a useful piece of tech if we're
16:21
looking for someone in a crowd using facial recognition and we know that we're looking for a man then by detecting gender we could immediately rule out all the women and get a result faster however you can't always tell someone's gender just by looking at their face automatic gender recognition might work well enough most of the time which is to say the meanings it assigns might be useful most of the time but like the penis detection machine it encodes a particular way of seeing and
16:53
it's always going to make mistakes particularly with trans people like the penis detection machine that could could result in us being humiliated or even hurt and you might say well that's unfortunate but trans people are a very small minority two things firstly a mistake that keeps happening to a specific group of people isn't just a mistake that's systemic discrimination secondly who told you were a small minority being left-handed used to be thought rare but then we stopped
17:27
punishing children for being left handed and we found out there's actually a lot more of them than anyone thought aren't we right-handers lucky that left-handed people never sought what would surely have been their Justified Revenge so how to make ethical AI not just facial recognition but any AI that categorizes people recruitment software University applications anything we could try build it better more data more diverse data more buttons s on the penis detection machine and we could try
18:02
police it better give people options if they are miscategorized but that still might not be enough Tech researcher Oz Keys says gender recognition Ai No matter how inclusive we try to make it is always premised on the idea that someone else can look at you and tell you what your deal is and that's always going to be fundamentally incorrect because one of the fun and beautiful things about human gender is deciding your deal for yourself automated gender recognition
18:38
purports to recognize gender but my analysis of the way in which it operates shows that this is only true if one denies the role that self- knowledge plays in gender and consequently denies the existence of trans people using a more inclusive view of gender one that recognizes the Primacy of the self we can see that the systems are not recognizing gender which would be impossible possible to reliably do by inference but instead imposing their own views of gender on unwitting users and
19:05
research subjects 30 years of research has not produced a single system that can genuinely recognize gender it has only produced systems that try to assign it to summarize AI is used in an already imperfect society and might exacerbate existing INE equality the solutions may not always be technological so when building it we need to pay careful attention and broaden the scope of our inquiries and critically interrogate our assumptions I've read a lot of papers about this issue and in the final
19:42
paragraph they always pull back and use a vague sentence like that because academic philosophers don't like telling people what to do their go-to solution is usually uh let's try doing some more philosophy practical steps tech companies can hire sociologists and philosophers to work on every step of AI development and pay particular attention to the ways in which the AI might exacerbate existing inequalities when in doubt hire someone with a Humanities degree wo we have covered a lot of
20:14
ground we've got technical problems and societal problems but I'm sure that if we host some diversity equality and inclusion seminars we can solve most of that so we are ready to make ethical AI
20:31
right I mean we must have covered everything by now it's not like there's an even bigger and even more fundamental problem a while ago someone used AI to make non-consensual pornography of me I got a bad feeling about this this person took screenshots from a previous episode of philosophy tube do you remember the one we did about effective altruism when I was dressed as the latex nun and they fed those images to an AI image generator which produced a series of pictures of me all of which were
21:07
quite bad in similar costumes and also some where I was partially nuded and then they posted those images publicly okay uh I'd like to tell you a little bit about how that look came together in the show notes I called it the Techno nun I came up with the idea after researching effective altruism for a long time I I wrote about 53,000 words worth of notes for that episode and I wanted a look that combined the idea of altruism helping other people with science hence techno and none and when I
21:47
had a design I was happy with I worked with Brian who is the philosophy tube stylist to refine it and then we went to a lovely lady named Nan who runs a latex company called Dead Lotus coutu and we worked a little bit more with her and then her and her assistants made that outfit fun fact lady gagar gets her latex from the same place as me Steph you're such a philosophy tube riper I'm kidding I'm sorry I loved House of Gucci we filmed that episode in a studio
22:20
in North London which I hired Brian was my dresser on the day and Nikki the philosophy tube makeup artist did all of the hair and the makeup Nikki worked looks so hard on this show like she often comes up with these little ideas at the last minute which just like Elevate things like on on the previous episode we did when I was the demon lawyer uh using victory rolls in my hair to look like horns it was Nikki's idea uh and then Mr X who is the philosophy chbe camera operator who prefers to
22:47
remain anonymous uh set up the camera and the lights and operated the camera and color graded all of the footage uh and then I edited the footage I do all the editing on philosophy tube myself it usually takes me about a week of 12-hour days and then I uploaded the final video to YouTube and to nebula I did the thumbnails the titles the subtitles the metadata and like all the stuff that takes a video file and turns it into an episode of philosophy tube I could go on
23:15
I haven't even mentioned the people who built the cameras and drove us to set and made the coffee if you added up all the time and money and work it took to produce those screenshots it would be staggering there's a reason that philosophy tube has a patreon page where people who like the show chip in what they can and it's because the people who make it work hard and this person took those screenshots and flattened all of that work in order to produce some shoddy images of me with my tits
23:57
out they evidently didn't consider the ethics of doing that they appropriated all that labor from all those people without permission and presented it as their own and I don't think they realized how insulting that is they assumed that what we had labored to make was just data ju just something they could take for themselves and I personally don't see it that way the fact that the end result was non-consensual pornography was kind of just the icing on the turd
24:32
interestingly a few months prior to that somebody made handrawn non-consensual porn of me and that felt more violating weirdly than the AI stuff I I think because the intent was quite different the person who made the AI porn of me I I think genuinely didn't know that the AI was going to do that and they didn't look closely and they posted it in a spirit of excitement and as soon as I said oh my God that's disgusting please don't they apologized and deleted it whereas the person who made the
25:02
handdrawn version obviously had spent a long time thinking about me in a very obsessive and lustful way and by posting it publicly they were deliberately trying to make me feel sexually violated which yeah I mean that sucks if I'm if I'm honest with you uh it's hard to deal with that first thing in the morning and then just go to work I was insulted my real naked body looks way better than that I tell you this story to introduce the data flattening problem that process is how
25:40
AI like chat GPT and darly mini work they scrape the internet for data no consent is taken no care or thought is given the people who make these models assume that they can do that because it's just data just abstract and immaterial and that's not true training data is made by and from people you might be aware that in the United States the writer Guild and the Screen Actors Guild have been on strike at time of recording the wga might have reached a provisional agreement with the studios
26:12
in the last like 3 hours but we don't know the content of it yet part of what they're striking about is AI because text and image generators obviously pose a threat to artists livelihoods Studios love the idea of AI because you can use it to create a lot of content cheap and fast and yeah most of it's going to be crap but you only have to generate House of the Dragon once and you've got the biggest show in the world actors and writers don't want our work to train Ai
26:40
and we don't want to work on stuff generated by Ai and part of the reason is that a bot would scrape a thousand scripts or performances crafted by a person and spit out a soulless ripoff it's an insulting misunderstanding of what we do when we make art so how can we make ethical AI is there a way to use data that prioritizes the consent and the labor of the people who work to produce it right now the main suggestion is to regulate tech companies through things like privacy laws antitrust laws
27:22
copyright laws beyond that there's a slightly stronger set of proposals called Data owning democracy the idea is that all the data you produce like your location data and your purchase histories and your browser data you'd get a copy of all of that and then you could decide what to do with it like sell it or donate it maybe beyond that there's an even stronger set of proposals called digital socialism where tech companies would be democratically controlled by their workers and the
27:50
infrastructure they use would be owned by Society so socialism those last two sets of proposals data owning democracy and digital socialism recommend redistributing property a little bit to increase the economic power of citizens and they do that because AI kind of messes with our concept of private property the English philosopher John Lock said you own your body and your self your person and that includes your ability to work and if you take the raw materials that nature provides and apply
28:30
your ability to work you can transform them like you take a tree and you carve it into a table the things that you make are like an extension of you you put yourself your labor power into them and so just as you own yourself you now own those things they are your property and if you want to you can take the things that you own and you can swap them for money and you can take the money and you can swap it for body stockings and bondage gear and then you own those things too I think a
29:01
lot of people's idea of private property is something like that kind of if we stopped to really think about it which let's face it we often don't because it's such a fundamental part of our society that it often goes unquestioned The Economist Friedrich Hayek actually said that private property is the Cornerstone of civilization once you start breaking those rules the whole thing is going to come down trouble is AI does kind of break those rules if a black box AI scrapes 10,000 movie
29:31
scripts and spits out a new one how can we ever tell whose labor went into the final product and in what proportions and therefore how much they should be paid probably we can't and if we try using a surrogate model we're going to encounter the fairw washing problem from earlier which means the people who produced that training data are almost certainly going to get ripped off in this scenario the AI basically steals the property from writers and slaps a new label on it so they can't do
30:00
anything which is another reason why screenwriters don't like it annoyingly it's so easy that just because it breaks the rules doesn't mean corporations won't use it and so some corporations think we need new rules introducing Sam Alman CEO of open AI makers of chat GPT Sam thinks AI is about to revolutionize the economy it's going to generate a lot of content quickly and cheaply which means it's going to generate a lot of wealth but because it's also going to
30:37
put a lot of people like writers out of a job most of that wealth is going to go to private corporations governments will struggle to raise income taxes since most people won't have income anymore so if we're not careful this could be the AI jobs apocalypse Alman Compares himself to Robert Oppenheimer and he takes himself very very ser seriously so something fun that I like to do is every time he mentions computer models pretend he's talking about fashion models
31:06
regulatory intervention by governments will be critical to mitigate the risks of increasingly powerful models I think we're at the end of the era where it's going to be these like giant models the model will confidently state things as if they were facts that are entirely made up ultman thinks he has a solution to this jobs apocalypse problem he's a georgist a philosophy named after 19th century Economist Henry George georgists believe that all taxes should
31:36
be eliminated and replaced with a tax on land because land only gets its value from the work that other people do around it for example property in central London is really expensive because you're close to all the theaters and restaurants and tourist attractions that people want to go to but landlords didn't make them those things Society did so the idea is they should share the value of their land with Society through a tax georgists believe that capitalism
32:08
would work great if landlords just got out of the way and let the free market be free it used to be a fairly popular idea partly because it sounds radical but you don't actually have to redistribute anybody's property just tax it a little bit differently and partly I think it into that relatable feeling you get when you pay your rent like oh my God I am getting ripped off Alman says the USA should pass a constitutional amendment to eliminate all taxes except land tax and
32:42
corporation tax this tax money should then be used to buy shares in companies which get given to Citizens as universal basic income so they don't need jobs and they have a stake in the economy doing well so it's like capitalism but we you made it work for everybody hang on a minute did you catch that the AI billionaire is saying don't regulate me don't redistribute property to empower citizens and don't tax me now tax me later when my company has made trillions of dollars which it
33:21
definitely will because AI is definitely not a bubble even though it costs my company a huge amount and our main product seems to be hype tax me later when I've got all that money and uh do it by passing a piece of legislation that's uh very very difficult to pass and in the meantime don't change any of the data flattening that lets me get rich cool cool man smart cards on the table I don't think ultman and people like him are facing the real challenges posed by AI at
34:01
all I think they might be doing something else Alman owns a company called worldcoin they scan people's irises and pay them for that data in cryptocurrency I've seen it advertised on Tik Tok and that might sound like data owning democracy CU like we're paying people for their data right but then you think about it for more than 5 minutes and you realize okay so Alman thinks that AI is going to wipe out people's jobs so people need Universal basic income and his plan for that is to give them
34:36
crypto so can you actually spend that on anything like rent or groceries No in fact I know from a lot of my friends that crypto is often used to pay only fans girls so the men using it don't get found out by their wives if you're married to a crypto bro check his browser history but other than paying for porn how is this actually helping turning worldcoin into Global Universal basic income would take a huge amount of legislation and infrastructure that Alman has no plan no Authority and
35:10
no ability to deliver which might explain why within a month of its launch worldcoin fell by 44% and its offices were raided by police nice job dude it seems like alman's real plan is to tell a scary story about AI Dazzle some investors Harvest people's data and pump the value of a crypto scam this isn't making ethical AI this isn't engaging with any of the real questions it throws up about private property or anything else and a lot of AI chat is like this it's not ethics it's
35:53
marketing hi I'm Kelly Slaughter founder and CEO of Ali the world's first feminine AI company AI is Transforming Our World unlocking quadrillions of dollars of value but we need to share the benefits with everybody that's why I am proud to introduce Ally artificial intelligence by women for women Our Story begins a few years ago when I was at a business conference in Switzerland about lethal autonomous weapons everybody was talking about the military applications of a eye how it could
36:28
enhance our Battlefield capabilities and increase our power to inflict violence and I looked around the room and I realized oh my God I'm the only woman here women are so underrepresented in stem I had to work so much harder than my male colleagues to become a Founder when I left Harvard I had nothing but a dream and my dad's software company I sacrificed so much Family Health Sleep friendships my marriage and it's all been worth it because I followed that dream and now
37:10
I'm here I believe every woman should have the opportunity to follow her dream by working for me and that's when I had the idea for Ally just like the Journey of a th000 miles begins with a single step the most sophisticated artificial minds are assembly LED from millions of tiny computer-based tasks we're giving women in refugee camps and prisons the chance at a global family by empowering them to become micro entrepreneurs teaching our algorithms to recognize text faces and more one click
37:47
at a time the tasks are flexible easy and fun every one of our micro entrepreneurs are guaranteed 100% female and they get on the job child care when they bring their daughters to work alongside them the data they provide go toward building alley our artificially intelligent but genuinely feminine AI system maybe she'll be part of a self-driving vehicle facial recognition software a security system the possibilities are endless we need smart business to win the AI war with China
38:23
and I believe diversity is a strategic priority when we make the next generation of lethal autonomous weapons half those weapons should be made by women the technology we're building at Ali has the potential to reshape the world it's not intelligence it's artificial intelligence
38:50
[Music] [Applause] at the risk of sounding very obvious if you want to make AI you need electricity cables power plants lithium lithium is used in rechargeable batteries it's in laptops and cameras and electric cars I own a small battery powered device which almost never leaves my hand which needs lithium to function and as well as my vibrator I also own a smartphone lithium has to be mined out of the ground and then taken to a Refinery on trucks and ships it has to
39:34
be turned into components those also have to be transported every step uses up non-renewable resources and emits CO2 this is all to say that AI is not a free floating mind it is a physical material thing not just material but dirty mining in particular causes a lot of damage to people and the environment damage that can be expressed in terms of financial costs like x billion dollars worth of damage the cost of digging stuff out of the ground is almost always more than
40:10
the value of what you dig the only way mining works as a business is because mining corporations don't pay that cost the cost to the miner's health is born by the miners the cost to the environment is born by everyone it's worth worth remembering that because the idea that the tech sector including but not limited to AI is clean is a very profitable lie the term artificial intelligence May invoke ideas of algorithms data and computer architectures but none of that can
40:46
function without the minerals and resources that build computing's core components rechargeable lithium ion batteries are essential for mobile devices and laptops in home assistance and data center backup power they underg guard the internet and every Commerce platform that runs on it from banking to retail to stock market trades the cloud is the backbone of the artificial intelligence industry and it's made of rocks and lithium Brine and crude oil the other thing you need to make AI is
41:18
people somebody has to go down the lithium mine and dig it up somebody has to work in the refinery Drive the trucks and the ships like we said before somebody has to produce all that training data someone also has to label that data to tell the algorithm that's a human face that's an object that's hate speech that's free speech that's a car that's a Ballard it actually takes a lot of people to do that job if you add it up everyone who does it they would make up one of the largest workforces on
41:47
Earth in the global South that job is done by people in countries like the Philippines India and Uganda sometimes people in refugee camps like in Kenya and Lebanon in the global nor North it's done by prisoners old people who can't afford to retire and people who can't get other jobs because for example they have a disability behind the search engines apps and smart devices stand workers often those banished to the margins of our global system who for lack of other options are forced to
42:16
clean data and oversee algorithms for little more than a few cents the feeds of Facebook and Twitter may seem to wipe away violent content with automated precision but decisions about what constitutes pornography or hate speech are not made by algorithms a facial recognition camera seems of its own valtion to spot a face in the crowd an autonomous truck to drive without human involvement but in reality the magic of machine learning is the grind of data labeling and as Jobs go data labeling is
42:50
pretty one of the reasons I love being an actress is that my boss the director is always a person if I want to try something different I can ask if they want me to try something different they just ask I don't get like an alert on my phone saying you were 4% less Melancholy in the previous take please adjust performance hiring decisions are made by casting directors not by computers that's also why I like working on nebula on YouTube there are content moderation
43:16
algorithms and Advertiser guidelines that are all enforced by computers but on nebula there's none of that if I have a problem I text Dave the CEO and he replies straight away as a worker I am more free when I work for humans data labeling is not like that it's called microw work because you get a tiny task like label these pictures and a fraction of a dollar for every task that you complete you might do tasks for 20 different companies in a day all coordinated through an online platform
43:47
like Amazon's Mechanical Turk it's very poorly paid has almost no workers rights you're overseen and assessed by algorithms so if something goes wrong the software fires you for no reason and you're locked out there's no one you can complain to and because it's spread all over the world it's very difficult to unionize in his book work without the worker scholar Phil Jones says the real AI jobs apocalypse is this not jobs getting taken away but becoming deskilled atomized more surveillance
44:18
less worker power and generally these jobs pay so little and are so precarious that the main difference between them and unemployed is that when you're unemployed you at least have more control over your time so Jones coins a new term to describe them sub employment sub employment describes work that is highly temporary casual and contingent work that involves large amounts of unpaid labor significant underemployment or high levels of inw work poverty or work that
44:49
more often than not no longer guarantees a life any better than the most abject forms of unemployment workers do a few few hours here and there not enough to qualify them for employment rights but enough to prevent them from claiming benefits and appearing in unemployment statistics a situation so uncertain that workers are compelled to accept ours whenever offered in the global North a lot of the jobs created since the 2008 financial crash are sub employment that shift was achieved through mass
45:21
incarceration and gutting welfare so people were moved from benefits to either jail or working poverty in the global South slums and refugee camps are kept full by Wars and Border policies that Northern countries often have a hand in just like mining AI only works as a business because the true costs are borne by people who are presumed because of their race or their national origin or their age or whatever to simply not matter you see it's all well and good to say that private property is the
45:55
Cornerstone of civilization but some people's private property has always been a bit more Cornerstone than others when we remember things like slavery colonialism the subjugation of women as unpaid laborers in the home it starts to look like for the last few centuries at least the real Cornerstone of civilization has been private property for some people yes propped up by theft and exploitation of marginalized people for example John Lock said that when you put your lab
46:26
labor power into natural resources you make them your property well he also said that if you're not doing stationary agriculture on a piece of land you're not really laboring on it and therefore it's not really your property and that argument was then used as the explicit legal justification for stealing land from nomadic indigenous people in the United States a theft that John Lock himself directly oversaw and profited from this is a trend that we in the 21st are starting to recognize now and
46:59
hopefully address because if we really believe that all human beings are equal then we cannot also do all of that stuff but AI doesn't disrupt that Trend it continues it at the start of the video we noted how AI is quite a promiscuous term it's used to describe all different kinds of systems so in her book Atlas of AI researcher Kate Crawford uses a different term large scale Computing and I think that really brings home how these Technologies aren't a new type of thing we've had tools like this
47:38
for ages these ones are just bigger and more powerful which means they use up more resources emit more CO2 and exploit more people and yet even though these systems are very large and Powerful they're also vulnerable for one thing they're very vulnerable to climate change if there's for example a drought in Central America and ships can't get through the Panama Canal anymore and the flow of components that large scale Computing needs can slow down the more CO2 these systems
48:10
emit the more that's going to happen until eventually it will stop these systems are also vulnerable to workers if the people working in mining or shipping or data labeling are suddenly unable or unwilling to do their job then the system can stop again climate change is really going to accelerate this do you really think people in the Philippines are still going to be working for Silicon Valley when half their countes underwater they're going to be like Abigail I don't
48:37
care that your vibrator isn't charging I've got bigger problems deliberate action by workers can jam the system too when Google employees found out their company was working with the US Department of Defense they kicked up such a fuss that Google cancelled the contract Amazon employees staged walkouts over the company's carbon footprint when Act and writers discovered that Studios wanted to replace us with computers we went on strike the more physical and spread out
49:05
these systems are the more opportunity there is to take a hammer to them that's why striking is such a powerful tool and why governments who act on behalf of fossil fuel companies are so Keen to suppress worker power so can we make ethical AI
49:30
well who's
49:39
we now we can return to the alignment problem suppose that a hostile AGI came into being tomorrow we understand now that large scale Computing needs resources and and labor and if our AGI had an army of Terminators it could probably get those things on its own but it's going to have to build those Terminators first and the factories that build them so more likely it would need people recruiting them through trickery force or persuasion which means the power of a hostile AGI
50:18
would still depend on its relation to our existing political and social systems it is a mistake to think that a disembodied mind could destroy the world from inside the cloud even if we built badly aligned AGI its power would come not from technology but from the human systems in which that technology is embedded so when people talk about ethical AI maybe we shouldn't think about Skynet maybe we should think instead about working conditions climate change and how to make the economy serve humans
50:55
rather than the other way way around we shouldn't assume that AGI is inevitable and try and Patch ethics on as an afterthought but rather ask how we can make Justice and sustainability and fairness the goal of every piece of tech we build if we really want to make ethical AI then we might like to consider this perspective there is no ethical computation under capitalism [Music] it feels a bit weird to do the nebula ad after that emotional ending but like I said the people who make this show work
51:36
really hard and if you want to help us then you can nebula is a streaming service that's partially owned by the creators on it including me so we're trying not to be Uber capitalist a while ago the philosophy tube crew made a documentary about how we make this show it goes into my research and writing process and if you'd like to see it click the link in the description go.
52:25
nebula.com stepping up their production of original content you might remember last year I wrote an award-winning stage play called the prince a lot of you came to see it in London and the recording is available on nebula if you enjoyed that then you might be interested to know I'm writing something new see you on nebula
53:02
[Music]
53:12
[Music] o
53:37
[Music]
54:28
[Music] oh [Music]
54:51
[Music]
55:27
o [Music]
55:38
[Music]
55:56
a [Music]
56:05
[Music]
56:19
[Music] if you look at the progress from model to model even some of our biggest critics are like wow
Here to help you learn
Sign in to keep a private conversation with your course tutor.
Ask me anything about the material…
Sign in to chat