Instant Genius - Robert Elliott Smith: Are algorithms inherently biased?

Episode Date: October 9, 2019

In this week’s podcast, we speak Robert Elliott Smith, an expert in evolutionary algorithms and researcher of artificial intelligence. His latest book, Rage Inside the Machine: The Prejudice of Algorithms, and How to Stop the Internet Making Bigots of Us All (£20, Bloomsbury), explores how the harmful effects of bigotry, greed, segregation and mass coercion are finding their way into the AI that runs our lives, without us even realising it. He tells us how powerful algorithms have been manipulated to divide people, why algorithmic bias has a dark history in the field of eugenics, and what we can do to fight back against the insidious influences of social media. Subscribe to the Science Focus Podcast on these services: Acast, iTunes, Stitcher, RSS, Overcast Let us know what you think of the episode with a review or a comment wherever you listen to your podcasts. Listen to more episodes of the Science Focus Podcast: What's the deal with algorithms? – Hannah Fry Does data discriminate against women? – Caroline Criado Perez Is racism creeping into science? – Angela Saini What happens when maths goes horribly, horribly wrong? – Matt Parker How technology is changing politics – Jamie Susskind There's no such thing as Blue Monday – Sir David Spiegelhalter Follow Science Focus on Twitter, Facebook, Instagram and Flipboard Hosted on Acast. See acast.com/privacy for more information. Learn more about your ad choices. Visit podcastchoices.com/adchoices

You said this place was steps from the water. We just haven't found the steps yet. How much did we save? Enough. Enough to get lost. Or you could book a stay with Hilton. Welcome to your oceanfront room. Just steps from the water.

The Hilton sale is on now. Book on Hilton.com or the Hilton app and save up to 20% to get the stay you expected. When you want savings, not surprises. It matters where you stay. Hilton, for the stay. Study And play.

Come together on a Windows 11 PC. And for a limited time, college students get the best of both worlds. Get the Unreal College deal, everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 premium

and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller. Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m.S. College PC. Ambition comes in all shapes and sizes. At First Citizens Bank, we roll with your goals because we're built for what you're building.

Fit for your ambition for Citizens Bank. This podcast is sponsored by name, audio and focal. Streaming has made music more accessible than ever, but true listening is about more than ease. It's about quality. British audio experts name audio, alongside French acoustic specialist focal, Combine handcrafted tradition with cutting-edge innovation and high-end materials, delivering digital precision with analog warmth. So you can experience exceptional sound at home.

Music just as the artist intended. Visit name audio.com to learn more. These algorithms of how we select our elected officials, coupled with this informational algorithm, could be manipulated to divide people. And that's really what's going on. And the thing people need to realize going back to it again is algorithms are really simple-minded. People who are able to see that simple-mindedness can exploit it. You're listening to the Science Focus podcast from the BBC Science Focus magazine team.

With the UK's best-selling science and technology monthly, available in print and in several digital formats throughout the world. Find out more at ScienceFocus.com or look out for us in your app store. Hello, I'm Amy Barrett, editorial assistant at BBC Science Focus magazine. In this week's podcast, we speak to Robert Elliott Smith, an expert in evolutionary algorithms and researcher of artificial intelligence. His latest book, Rage Inside the Machine, explores how the harmful effects of bigotry, greed, segregation and mass coercion are finding the way into the AI that runs our lives, without us even realizing it. He speaks to online editor Alexander McNamara about how powerful algorithms

have been manipulated to divide people, why algorithmic bias has a dark history in the field of eugenics, and what we can do to fight back against the insidious influences of social media. Rob Smith, I have worked in artificial intelligence for 30 years. Most recently, I've written a book called Rage Inside the Machine, The Prejudice of Algorithms and How to Stop the Internet, Making Bigots of us all. I'm also a senior research fellow of computer science at University of College London, and I am CTO for a technology company called Boxar. So is it safe to say that you're quite an authority when it comes to things such as artificial intelligence

and computer science in general? I like to think so. So on that note, would you be able to tell us what are, you know, your book is about the prejudice of algorithms and it's got a lot about artificial intelligence in there, would you be able to explain what are algorithms and AI and how we've got to this position that we are, just how important they are to our lives at the moment.

Algorithms are well-stated procedures, usually in computers today. It's interesting that the word comes from the name of a Arab mathematician, whose work was popularized in Europe by Fibonacci, and then the word was people used the procedures, mathematical procedures in that book, were known as algorithmist. and from this Arab name, a kind of Latinization of that name, and then Jeffrey Chaucer anglicized it into algorithm. So that's where it comes from.

It means a well-stated procedure. And what is AI, I would say AI is algorithms that we happen to label as being something like human intelligence. But usually they aren't a lot like human intelligence. There's very few algorithms that in any way even try to be, like human intelligence except in very vaguely metaphorical ways. And I actually prefer a term introduced by the science fiction author, Neil Stevenson, which is pseudo-intelligence, which I think makes it a little clearer that what algorithms are doing today is really something that's a sort

of imitation of human intelligence. And where are algorithms today? Well, they're everywhere in our lives. They are the things that pick out what news you get in your Facebook or Twitter feed. They are the things that pick out what matches are suggested to you on Tinder. They are the things that determine what jobs are assigned to Deliveroo or Uber drivers and many other people's minute-by-minute work is assigned by algorithms. And they're even in things like criminal justice now because there are algorithms that are suggesting areas that police should patrol and suggesting to judges whether they should give parole sentences to individuals who are seeking parole.

So effectively, algorithms are literally everywhere in our lives today. So that sounds to me like they're, in the modern day, it's very much anything that's in a sort of digital world. Have algorithms sort of ruled, not ruled, but been a significant. part of our life, maybe one day, who knows, up until now, was there a point in history where suddenly they've taken on a much more prominent position? Well, it's an interesting question. I think that an algorithmic approach to looking at how humans interact and affecting

how humans interact really begins quite a long time ago and really made its way into our lives largely through economics, which is a way of looking at economics. a political science that was about how do we shape politics to affect a good society. And in time, it's become a way of a sort of quantified social science of human beings. And it's existed that way for quite a long time. I'd say the entry point is through quantitative social science and then ultimately economics. But recently, I'd say over the past 30 years, the years that I've been in AI, it's become a part of our life's infrastructure, the way electricity and water and roads are through the fact that most of us are involved with an algorithm

most of our day. We're working online, doing online searches, looking at our Facebook feeds, looking at where to go to dinner, looking at what job we might do next. So algorithms in the last 30 years have become an essential infrastructure of at least half the people on Earth and a growing number as time goes fine. I guess we don't really see these algorithms happening, but they are everywhere. Yeah, and one thing I think that's really necessary is for people to realize that they are there, that, you know, the first area that I think people need to realize is the information that's curated to them that's delivered through searches and through news feeds is mediated

and arranged by algorithms. And those algorithms are, of course, using your personal data, the data of your, of your who your friends are and lots of other factors to accomplish commercial and power-seeking aims. And we're constantly, your Google searches are shaped based on you. You don't get the same Google searches as other people. Your view of the news is shaped to you. You don't get the same news as other people. And ultimately, the commercial products that are suggested to you probably more constantly than you realize

are a version of that feed that's direct. directed at you. So this feed that's directed at me, obviously, it knows my data, presumably, which is how it can direct it at me. But who are the people that are creating these algorithms? Well, it's a variety of people. Certainly the people creating them, most of algorithms are lives. A lot of them come from the Big Five technology companies, Google, Amazon, Facebook, Apple,

and I'll get the last one if I thought about it for a moment. effectively a lot of the algorithms our lives are suggested by those, but then there are lots and lots of other people in Silicon Valley who are writing those algorithms. And then there are people who are trying to influence those algorithms, particularly in social media. And we all know about the Cambridge Analytica scandal and the Russian troll farms that are trying to influence things on social media. They're sort of people who are playing with the algorithms that exist and the algorithm that reality we live in, they're trying to manipulate. So the algorithms are being written in Silicon Valley. The algorithms are being pushed this way and that by many, many actors all over the place today. So when you mentioned what happened with the troll farms,

is there a problem with our algorithms in the fact that nefarious people can change them so that manipulate what we're seeing and our feeds and everything? Well, the thing that I want people to realize more than anything else is that algorithms, while massive and complex and manipulating extremely large amounts of data and somewhat intractable to human beings are actually terribly simple-minded in many ways. Algorithms basically have a very statistical and quantitative view of humanity. And because of that, they can be manipulated. Effectively, I think, you know, if you watch the Netflix documentary, The Great Hack,

which is very good, by the way, and everybody should watch. And I do mean everybody should watch it. If you watch that, it's a wonderful expose of what happened during the Brexit referendum and during the election of Donald Trump through Cambridge Analytica's attempt to manipulate people. Now, the one impression that could be left from the Great Hack that I think is something people should be wary of is the impression that they had psychometric models and big data analysis. an AI that was so powerful that it made people putty in their hands. That's really not true. Algorithms are really very simple-minded in their understanding of people. And most people know this

because they get recommendations of books on Amazon, and they know that those recommendations are a big crap, really. But here's the thing. Because we've ceded our social lives and our news feeds to algorithms that we don't really understand, that has created a situation. where people who do understand those algorithms can come in and basically segregate us, right? Effectively what Cambridge Analytica did is they basically used a form of informational segregation. They basically were able to say, okay, here are ways we can split up the social network based on simplifying features of people, where they live, their age, their gender, small political beliefs they have. And we can separate those people. And if we separate them finally enough, we can effectively do a sort of informational gerrymandering.

That's what they were really doing. What they were doing is delivering information to small groups in such way that they split votes in places, that they were able to split off votes such that they got the political outcome they wanted. So they were simultaneously manipulating this highly simplified curator of our news and the fact that we have electoral system. that do things like in America, the Electoral College, in this country first past the post, so that those algorithms of how we select our elected officials, coupled with this informational algorithm, could be manipulated to divide people. And that's really what's going on. And the thing people need to realize, going back to it again, is algorithms are really simple-minded.

People who are able to see that simple-mindedness can exploit it. And also something that my title of my book suggests is that simple-mindedness can be expected to divide people along traditionally prejudiced lines because effectively when you simplify people down to simple features, unsurprisingly, you're going to have divisions based on things like color of skin or a religious checkbox or gender. And so that simple, those simple algorithms are, you know, because they're so simple in the sense, what they can do, that's causing a bigger problem with just, you know, people and how people see things. Absolutely. And the thing is, is that I, being a guy who's worked in artificial intelligence my entire adult life, I think it's growing. I think there's great things that I should use the term pseudo-intelligence myself. There are great things pseudo-intelligence can do. However, when it's vaunted in the public media is basically saying that it's smart or smarter than people,

that's enforcing a view of people that's highly simplified, you know, and a view of thinking that's highly simplified. And when we vaunt that, we make ourselves vulnerable. So really what I'm about is people's intelligence is a highly complex phenomena. It's a multidimensional, highly complex phenomena that's already simplified by things like our exam oriented culture, by economics, by our voting systems. When we then add to that the fact that algorithms are delivering a huge part of the infrastructure of our lives, and those algorithms are explicitly based on these historical, simplified ways of looking at people, then we've amped up the level of a possible simple-minded division.

of people. And in my mind, I believe it's one of the reasons that we see the polarized and divided world that we see today. Does that mean that there's an inherent problem with algorithms, which is the fact that they are essentially, from the very outset, they're created to not be dumb. They're very clever things, but just to minimize things in a way that has damaging effects for us as people. Well, algorithms are geared to take complicated stuff, simplify it, and generalize about it, right? And that's fine when you're dealing with nuts and bolts. It's fine when you're dealing with things that don't impact a lot of people's lives too much. But when you direct

that at human beings, simplification and generalization, driven by a pursuit of value or power, becomes a problem. And it's always been a problem. The thing is, if you look at the early quantitative social science, you see that it simplifies the generalized about people and was very useful for the advancement of our understanding of human beings. However, it was always the next-door neighbor of racism. Social science that quantifies people has always been a neighbor of intolerant, divisive racist and misogynist and many other sorts of divisive politics. It's always been there, and there's always been this fight. What's changed, First of all, I think it's important people to realize that history, and I go on about the history.

But what's changed is now those ideas have been turned into algorithms. It's not that the programmers are racist. It's that the idea of simplifying and generalizing about people has a tendency towards this kind of thing. Now that that tendency is automated and ubiquitous, it's a much bigger problem. I said, you know, for me, just looking, when I see a piece of, think of an algorithm, I don't think that, you know, it's next to this insidious part of our human history. Yeah, yeah, but it really is. Interestingly, I'm a senior research fellow at UCL.

UCL is where eugenics was invented. The word eugenics was invented by Francis Galton. He was Charles Darwin's first cousin. and he, you know, effectively endowed a eugenics chair and a eugenics department at UCL when UCL became the first nonconformist university in the UK. So this was a part of a progressive social agenda. And, you know, the people who belong to the eugenics society that was associated with that early department at UCL were people like Neville Chamberl and John Maynard Keynes and Margaret Sanger, This is a progressive social agenda. Unfortunately, then it migrated to America and became involved in really horrible eugenic policies there,

and then ultimately to the Nazis and Nazi Germany. But alongside at UCL, what the scientists were doing in the eugenics department was largely work on statistics, statistics about people. That was how they were trying to prop up the idea of eugenics. many of the algorithms and ideas in algorithms today were developed in that department, right? You know, it's fundamental statistical algorithms were developed to view people through a simplifying lens, and those have now become a part of our algorithmic infrastructure. Those ideas were never very far from very dangerous social phenomena,

but it needs to be said the scientists working at UCL in those early days were trying to do good. and with the appropriate human intervention, algorithms can do good. But unfortunately, they have to be very tightly controlled because it's all too easy for them to turn into tools of real evil stuff, as we saw in America before the Second World War and in many ways after, and certainly in Nazi Germany. Would you be able to just quickly explain what eugenics is just for anyone who doesn't quite know? Oh, yeah. Eugenics is an idea that we can change humanity for the better through selective breeding of human beings, which includes there's what they're so-called positive eugenics, which is encouraging people who we think are better to have more kids.

And then there's negative eugenics, which basically is preventing people who we think are less good from having children or from even. surviving. And it needs to be said that there were quasi-ugenic policies about sterilization of the mentally infirm well into the 1960s and 70s in the Western world. The term eugenics was a very acceptable, progressive term, up until World War II, because the Nazis made it so bad. The Nazis policies were explicitly eugenic, the elimination of Jews, the elimination of of Romani people. Those ideas were ubiquitously accepted in a certain progressive community across the Western world before War II.

After War II, eugenics became a dirty word. No one used that term anymore because it was so poorly thought of after the war. However, many of the philosophies survived and certainly the idea of the quantification of people, it went on and survives to this day. And there is a close relationship between our algorithmic infrastructure emphasizing those ideas and doing big data analysis of people and the rise of racist science that we're seeing today. There's a new, what's so-called racial realism that is coming in to science today, which

those things aren't unrelated. And the flaw in both of those things is people are highly complex phenomena. People aren't simple. And algorithms are, in fact, quite simple. And so these things, these social sciences, these sort of things that they're created that sort of, you know, were decided, you know, we said that these aren't good ways of measuring people. Is that sort of seeping back into things such as, you know, social media when we give

lots of details and information about ourselves to our social networks and then they are working out what we are and using those for their own own goals that it were. Absolutely. And you see lots of instances of this sort of thing going on. It's sort of inevitable when you simplify people and generalize about them that you will come out with results that are offensive. And you see some of those online, the unprofessional hair controversy at Google where a few years ago, someone discovered that if you typed unprofessional hair into a Google image search, you got pictures of black women, exclusively black women. So why? You know, people have asked the

question, why? Why did that happen? I believe the reason is that the shape, image recognition algorithms are based on very simple things. They're based on things like color, size, and shape, effectively, implemented through a lot of statistics and a lot of mathematical functions. Well, think about color, size, and shape of human hair. And I think anyone will realize that the natural color size and shape of hair is particularly recognizable for certain groups of people, particularly black women, not for black men so much given modern hairstyles. But for black women, large, voluminous, curly hair is very commonplace. And therefore, the algorithm was able to find that and focus on it.

And it got tied to the idea of unprofessional because there were probably some sites out there that had some pictures of unprofessional hair. And those were the most recognizable pictures of the algorithm to pick up on. That's the reason. So what you see in that controversy is this idea of simplification and generalization about people connecting to these things that we think of as racism. And if you think about it, it's like this. What is racism except tying a simple visual feature of a person to a complex and meaningful human concept like intelligence or worth or professionalism? So, you know, I guess the positive side of that is this. It should, why should anybody be asking a question about unprofessional hair?

And, you know, this prompts the whole question of what does it mean for hair to be unprofessional? does that mean so does that mean that the algorithm itself was created in a way that could be perceived racist or it's just a you know something that's been overlooked and this is the thing that I'm really on about are there people yeah let me let me preface this to say are there people who are is there too too little diversity in the programmer community and therefore bias in that community absolutely are there people out there trying to manipulate the internet to have effectively racist results on our politics, absolutely. But the overlooked phenomena, the phenomenon that people really need to see, is that that algorithm was not programmed to be racist. That algorithm was programmed to simplify and generalize about image. That's what it was programmed to do. And that's a useful thing. But when you pointed at people, what do you get? You get results like that. That's the missing element that people really need to understand.

Algorithms simplify and generalize. When that's pointed at people, we get phenomena that we think, progressive people think, are very bad. So I can see how when an algorithm is pointed at something that has a very sort of human context, that it can have these results that, even though I guess to the algorithm are perfectly correct, they're just abhorrent to us. Does it have the same effect in things that aren't related to people? So, for instance, I'm thinking to sort of say something like a self-driving car.

Yeah. Yes. The bottom line is, I think now we're seeing the fall of the hype cycle about self-driving cars. And I think there was a recent New York Times report that said that the big three auto companies are beginning to admit that self-driving cars are further out in the future than in the future than they thought. And the reason is that detecting the unforeseen is really a lot harder than people thought it was. I was talking about this years ago. I expected this to come.

And the thing about the unforeseen is this people because of the complexity of their lives and the way they behave, occasionally do things you don't expect. If all the cars on the road were robots controlled by robots, then self-driving cars would be really easy. But the world that the cars drive through is a human world, and therefore it's a highly complex. So effectively, we have the same problem. The simplification of these complex human phenomena is harder than we think. And in self-driving cars, because the car companies don't want cars ever running people over, that's blocking the progress of that technology. And I think it's probably going to block it really quite severely for a very long time.

One of the things I say to people is this is, you know, if you've got 100 self-driving cars on the road. It's very sensible to say, okay, for safety and for efficiency, if they're going largely to the same place anyway, as lots of cars are, just hook them together so that they're hooked together physically so they can't cause any additional chaos. Well, then you start thinking, well, just turn the engine in one of them on and turn the engines and all the others off, because why run several engines because it's less efficient? And then what you have is a train. And so you start thinking, why are we working so hard on self-driving cars when we have failing train infrastructure?

It makes absolutely no sense whatsoever to me. So just on the point of the algorithms and the fact that these things are happening, what can we do to stop getting any worse now? There's a lot of things to do. One is there's a profound realization we've got to make about the first thing is understanding. And I've talked a little bit about trying to understand algorithms better. And it's not that hard to understand. Algorithms are, there's a lot of math there, there's a lot of programming there.

But the stuff about simplifying people and about that history, if people just, you know, learn the history of how people are simplified mathematically and how place that's been to racism, it becomes. comes really rather obvious. So understanding is the first step. The second step is that we need to realize that in simplifying people, we've often simplified them down to the idea of driving value as a singular goal and the idea of things like survival of the fittest as being the way that systems evolve. That's not really an accurate portrayal of evolution, and value-seeking is not really an portrayal of economics is becoming increasingly obvious. And by a shift away from those kind of singular, unipolar optimizing sort of ways of looking at the world, we can shift towards other perspectives. And the perspective I outlined in the book is one that balances the idea of

doing better along some metric, some value seeking, with the idea of diversity and mixing of ideas and behaviors and things like that. And I mean, that not just as a hallmark card about diversity, but really I mean that as a technical phenomenon. And if you balance those two, there are certain kinds of algorithms I've worked with that show that you can reach a more productive evolutionary system. And I think we need to do the science to basically understand how do we encourage our algorithmic infrastructure to encourage a combination of diversity and optimization that causes effective social evolution. But I think both those elements have to be there.

And those elements then turn into things like how do we regulate, what kind of corporate ethics do we have around algorithms, and ultimately how do we behave as individuals in interacting with algorithms? So if we have that understanding and we begin to understand what algorithms are doing in our society, what should we do as individuals? And I have some suggestions about that too, particularly about how we, as social media persistence can change our behavior to try to make things a bit better. better. What sort of things should we do then? What would be your suggestions there? Well, the thing about preserving diversity, some of our work is basically shown that the dynamics of social media are very particular and they tend to lead towards polarization. The thing is is that people like to think that on the internet we're all broadcasters, but we're really not. We're narrowcasters to our connected friend group, really. And those narrowcasting channels

combined together across a massive network lead naturally to polarizations, particularly when they're mediated by algorithmic curators. So in order to overcome the influence of those algorithmic curators, what we need to do is be more human. And there's a few things you can do. One is you can friend more broadly and more tolerantly. I know it's difficult to do that in a very polarized time. But if you've got someone who said one thing that offended you, but you basically think it's

an okay guy, unblock that purpose. person, refriend that person, because the greater connectivity will make the digital segregation less possible. The other thing is the algorithms share based on very simplified criteria. They basically look at headlines and look at keywords and headlines and look for things that they think will cause a mode of click-through reactions. Don't do that, right? Be different from that. Don't click, don't share just on the headline. Don't like just on the headline. Read through, and I also say to people, familiarize yourself with the content creator, the human content creator. Like start liking some writers, start following particular authors, or particular organizations or particular new sources.

And there's a lot of results that say it's much better to share stuff that you think is good yourself than to simply batter back at people about stuff you dislike about their stuff. So try to be more positive in your sharing, try to be more human-driven in your sharing. I think it's good to add a comment and try to make that comment as insightful as you can because that humanity injected into the system helps retard the algorithmic effect. Now, ultimately, I think regulation and corporate ethics are going to be necessary because the level of influence involved. But we can, through better understanding and better behavior ourselves, have some effect on the dynamics today. So are we going to be the arbiters of change there?

Or is it going to have to be, you know, we can do that a lot, but is that going to be influenced by the tech giants or is it by government? Is someone going to have to step in and say, no, we need to put something down on this? Someone's going to have to step in. If you look at the evolution of broadcast media, you know, as radio grew up, it was rapidly realized that we had to be cautious about fairness and about fairness. manipulation in print media and then in broadcast media. And ultimately, that's going to have to happen in social media as well.

It's inevitable. You know, and in fact, we're going to have to move a little bit backwards because back in America, we used to have the fairness doctrine, which was effectively a way that broadcasters were licensed was they had to convince the Federal Communications Commission a bunch of human beings that they were being fair in their news coverage, that they were not manipulating people or presenting only one side of the news. That doctrine dissolved back during the Reagan administration and has not come back and has been fought against in Congress by one of the two parties, not to get too political

about it since then. And we see the artifacts of that in broadcast media, certainly in America and to some extent everywhere. But in the online media, it's even worse because there is effectively no regulation of any kind to no suggestion of fairness. You add into that the value-seeking potential algorithms that want you to click through because that click-through generates a micro-payment for somebody, then you have a real problem. And that's where we are today.

And I really think that won't be fixed through. I think our individual grassroots efforts would. I think that corporates can try. And I think there are people within corporations who are trying to change things. But ultimately, corporations have to deliver value for shareholders. That's the algorithm to which they are responding. And they're legally mandated to. So ultimately, change at a level of governance has to take place

or we won't really see real, effective society, evolutionary media again. It seems to me that the, when you say, like, you go online and you see it's unregulated. You can see that because, you know, you've got. Twitter streams and Facebooks and everything, it seemed very polarized one way and the other. Like, if you were to look at my Facebook feed, it would be very different to someone else who had a different opinion.

And it doesn't seem like that bridge is being crossed. Is there any, you know, is there a cause for optimism? Can we look forward to a future where these algorithms aren't causing such an issue? I tell you, the big cause for optimism for me is this. I think that from the advent of that eugenics movement I talked about and possibly even earlier, the idea of quantifying people, this idea that

penetrates into the social sciences, and particularly into economics and then into politics, the idea of quantifying people and driving simple value as a way to kind of make the world a better place. I think that now that is infrastructural, now that that's in our information superhighways, in effect, we can begin to see the flaws in it. You know, it's like now it's manifest. It's like, okay, our media is now entirely online and entirely, you know, algorithmically mediated. Well, look what it's doing to us.

You know, we can see that now. And I hope through what I'm trying to say in the book is people can realize that people are complex and that the quantification of them, although useful sometimes, always has to be considered in light of a more human perspective. And I think ultimately this could lead to a real deeper understanding of the nature of humanity fundamentally and its interaction with technology for the long term. So my optimism is that effectively things have now reached ahead and it will allow us to move into a new era of kind of understanding human complexity in a really deeper scientific way.

That was Robert Elliott Smith talking about the dangerous biases built into our algorithms. His book, Rage Inside the Machine, The Prejudice of Algorithms and How to Stop the Internet Making Bigots of us all, is available now. If you want to skirt around an algorithm and go for a good old recommendation from a friendly human, why not pick up a copy of BBC Science Focus magazine? In the October issue, we find out how gut-friendly probiotics and prebiotics could help treat anxiety and depression. There is of course much more inside, but if you can't wait until you pick up a copy, then why not treat yourself to another episode

of the Science Focus podcast? Might I suggest my recent interview with Richard Dawkins, where we discuss whether we can live in a world without religion? And don't forget to rate and review wherever you listen to your podcasts. Finally, if you'd like to know more about the history of science and eugenics and its impact on our modern world, head to BBC Eye Player and watch Eugenics' Sciences' greatest scandal. Look out for our interview with Adam Pearson, co-host,

of the two-part series in the November issue of BBC Science Focus magazine. Thank you for listening to the Science Focus podcast from the BBC Science Focus magazine team. With the UK's best-selling sites and technology monthly, available in print and in several digital formats throughout the world. Find out more at sciencefocus.com or look out for us in your app store. This podcast is sponsored by name, audio and focal. The texture and emotional depth of music can be lost through digital sources or poor signal. Name Audio believes you can have digital precision with analog warmth.

Alongside French acoustic specialist vocal, name creates high-end audio systems, combining innovation with craftsmanship, so you can listen to music, just as the artist intended. Discover more at name audio.com. Did you know if your windows are bare, indoor temperatures can go up 20 degrees? Get ahead of summer with custom window treatments like solar roller shades from blinds.com save up to 45% during the Memorial Day Early Access Sale. Whether you want to DIY it or have a pro handle everything, we've got you.

Free samples, real design experts, and zero pressure. Just help when you need it. Shop up to 45% off site wide right now during the early access Memorial Day sale at Blinds.com. Rules and restrictions apply. There's a moment when you start to wonder, what's the right next step? Not about changing who they are, just finding the right kind of support. At Kingsley Manor, life stays expressive, connected, and full of character, shaped by people who have lived interesting lives and aren't finished yet.

So it doesn't feel like a change. It feels like a continuation. Explore your options at canesley Manor.org, a non-profit month-to-month senior community within the Front Porch family.