Latest / Ethical Machines / How AI Robs Us of Meaning

Transcript

How AI Robs Us of Meaning · Ethical Machines

• 0:01
  [SPEAKER_03]: One of the major concerns about AI is that it's going to replace our jobs and therefore we won't have the money that we need in order to sustain ourselves and our family, et cetera.
• 0:09
  [SPEAKER_03]: There's another kind of concern that's related, though, which is that engaging in a work or engaging in lots of different activities is something that confers meaningfulness in our lives.
• 0:18
  [SPEAKER_03]: It allows us to exercise our skills, achieve levels of excellence, and that is no small part of what makes it the case that our lives are meaningful, we feel fulfilled, so on and so forth.
• 0:29
  [SPEAKER_03]: So how worry should we be about that?
• 0:32
  [SPEAKER_03]: My guest today, Sven Nehom, is quite concerned.
• 0:36
  [SPEAKER_03]: Sven is the professor of ethics of AI at IMU Munich and the principal investigator of AI ethics at the Munich Center for Machine Learning.
• 0:44
  [SPEAKER_03]: And he thinks that while there are certainly some things that we want AI to do, some sort of low-level mundane things, there's a danger that is going to take over too much.
• 0:54
  [SPEAKER_03]: One that he talks about is what he calls the laziness trap that there'll be things that we could do that would provide meaning, but because we get lazy we let AI do it and then we don't get the meaning that we should have in our lives.
• 1:07
  [SPEAKER_03]: He also says that
• 1:09
  [SPEAKER_03]: That means we need to reach a kind of AI sweet spot.
• 1:12
  [SPEAKER_03]: That is to say, we use AI for the things that don't confirm meaning, but we don't do it for the things that do confirm meanings.
• 1:18
  [SPEAKER_03]: We have to figure out where exactly does it make sense to use AI to replace our activities and where is it really crucial that we engage in the activities ourselves.
• 1:26
  [SPEAKER_03]: You'll hear me push back on a lot of the things here.
• 1:28
  [SPEAKER_03]: One major pushback that I offer Sven is, I'm not so sure how large the threat is.
• 1:34
  [SPEAKER_03]: So how large is the threat?
• 1:36
  [SPEAKER_03]: Who's threat and that's one major area of, I don't want to say, hey, I suppose it's disagreement.
• 1:41
  [SPEAKER_03]: We disagree a bit.
• 1:42
  [SPEAKER_03]: He's more concerned than I am.
• 1:44
  [SPEAKER_03]: Another point of not disagreeing, but at least, place where I have questions is,
• 1:49
  [SPEAKER_03]: who's responsible if that's the case?
• 1:51
  [SPEAKER_03]: The go-to thing is to blame AI or the AI companies for doing this and oh my god the AI companies robbed us of the meeting in our lives.
• 1:58
  [SPEAKER_03]: But I also wonder how much responsibility should we take?
• 2:02
  [SPEAKER_03]: If we as a people choose to outsource certain kinds of activities to AI
• 2:08
  [SPEAKER_03]: to what extent is it our fault for being lazy?
• 2:11
  [SPEAKER_03]: Is it to what extent is it the culture's fault for outsourcing things to AI?
• 2:15
  [SPEAKER_03]: When we don't need to, I don't have a sort of subtle view on this, but these are places where I push back on Span and so what makes it, I think, an interesting conversation.
• 2:22
  [SPEAKER_03]: So that's where this story or this episode is about.
• 2:25
  [SPEAKER_03]: For those of you who don't know, I have a new book coming out in April called The Ethical Nightmare Challenge.
• 2:30
  [SPEAKER_03]: That's about April 21st, is the official one-state.
• 2:33
  [SPEAKER_03]: If you want more information about when it launches, where it launches, how to get it, what it's about, et cetera, you could go to readblapinutsubstack.com and there's some information there.
• 2:41
  [SPEAKER_03]: And if you have any questions, comments, concerns about the podcast, et cetera, suggestions on who should be on top, so I should cover.
• 2:47
  [SPEAKER_03]: Please email me at ematreblapin.com.
• 2:50
  [SPEAKER_03]: Okay, that's it for me.
• 2:51
  [SPEAKER_03]: Let's go talk to Sven.
• 3:07
  [SPEAKER_03]: So here's your bold claim, AI robes us of our successes, and it's impact on a meaningful life.
• 3:14
  [SPEAKER_03]: Is that your view?
• 3:14
  [SPEAKER_03]: What does that amount do?
• 3:15
  [SPEAKER_03]: How does it rob us of our successes of life?
• 3:17
  [SPEAKER_01]: Let's unpack that a little bit.
• 3:19
  [SPEAKER_01]: I think that's a version of the claim that I want to make.
• 3:21
  [SPEAKER_01]: So first of all, yeah, we do.
• 3:23
  [SPEAKER_01]: I think we have to start by just saying something about what we mean by AI.
• 3:27
  [SPEAKER_01]: And obviously, as a topic that people are discussing, they're different.
• 3:32
  [SPEAKER_01]: I think one of your other guests that was on the recent Luchon of Florida, for example, has one suggestion about how to define AI.
• 3:37
  [SPEAKER_01]: Other people have made other suggestions, et cetera.
• 3:40
  [SPEAKER_01]: But I start with a very general idea, just that AI or technology that can take over tasks from us.
• 3:46
  [SPEAKER_01]: that we can outsource task to AI systems, where these are tasks that we would typically use.
• 3:51
  [SPEAKER_01]: What regardless are intelligence to perform, so like whether it's writing texts, producing art, making recommendations, making decisions, driving cars, etc.
• 4:03
  [SPEAKER_01]: etc.
• 4:03
  [SPEAKER_01]: These are all things that we use are theoretical and practical intelligence to do.
• 4:07
  [SPEAKER_01]: And now we have chat bolts that can help us
• 4:12
  [SPEAKER_01]: We have self-driving cars that can drive us around, we have recommender systems that can help us make recommendations, et cetera, et cetera.
• 4:18
  [SPEAKER_01]: So we have technologies that have the property of being able to take over work that we would otherwise do.
• 4:24
  [SPEAKER_01]: Yeah, and obviously this can be nice and convenient.
• 4:27
  [SPEAKER_01]: It could be that the work we're talking about is boring, meaningless, pointless work that we wouldn't want to do, but we don't want to do.
• 4:34
  [SPEAKER_01]: And so it's nice and convenient to hand it over to AI.
• 4:37
  [SPEAKER_01]: But some of these things might be things that are meaningful to us.
• 4:41
  [SPEAKER_01]: Again, go back to producing art.
• 4:43
  [SPEAKER_01]: That's a nice example.
• 4:45
  [SPEAKER_01]: They can music or I mean writing on the author a few books.
• 4:49
  [SPEAKER_01]: I won coming out soon and so I wrote those myself I didn't use generative AI these days.
• 4:55
  [SPEAKER_01]: However, we do have the option of instead of doing all of the writing related work To herself who can out so or more or less of it to AI and the more you outsource of course the less there is for you to still be doing and so there are these two options
• 5:13
  [SPEAKER_01]: Again, going back to what I just said in a moment ago, if what we give over to AI are boring, meaningless tasks, then it's not going to rob us of our successes, or make our lives less meaningful.
• 5:24
  [SPEAKER_01]: But if we hand over or outsource the tasks that we regard as meaningful, or where maybe we have an opportunity to serve Excel, to show our talents, to put in hard work,
• 5:36
  [SPEAKER_01]: to do things that maybe no one else could do because we are the one person who has to skill if all of that is going to be outsourced to AI then all of a sudden we do have a problem because maybe I don't have to put in the work anymore I don't have to make the effort I don't have to show my talent and so that's the worry.
• 5:54
  [SPEAKER_03]: Okay, but let me push back and so I can get deeper into this.
• 5:57
  [SPEAKER_03]: Because I don't immediately understand the threat or let's say I don't understand it.
• 6:02
  [SPEAKER_03]: Maybe it doesn't seem as threatening as it does to you.
• 6:05
  [SPEAKER_03]: You could still do all those things, right?
• 6:06
  [SPEAKER_03]: The fact that it can be done some other way.
• 6:08
  [SPEAKER_03]: Doesn't matter.
• 6:09
  [SPEAKER_03]: We engage in lots of activities.
• 6:10
  [SPEAKER_03]: That can be done more efficiently.
• 6:12
  [SPEAKER_03]: But in some cases, it's engaging in the activity itself that confers the value or the meaning or the fulfillment.
• 6:18
  [SPEAKER_03]: And so the fact that something else couldn't do that thing.
• 6:21
  [SPEAKER_03]: It doesn't really matter.
• 6:22
  [SPEAKER_03]: You know, I like to think about things like sports, for instance, or you mentioned writing.
• 6:26
  [SPEAKER_03]: I rock climb.
• 6:28
  [SPEAKER_03]: I want to get to the top.
• 6:29
  [SPEAKER_03]: I engage the activity of climbing the mountain or climbing the rock face to get to the top.
• 6:33
  [SPEAKER_03]: Other people, other things, other animals can do it way better than I can.
• 6:36
  [SPEAKER_03]: Mountain goats that they're remarkable at climbing as we're climbing rock faces.
• 6:41
  [SPEAKER_03]: But that doesn't undo the value or the meaning of fulfillment that I have.
• 6:44
  [SPEAKER_03]: I also have a book coming out.
• 6:45
  [SPEAKER_03]: It doesn't mean though, and there's, I didn't use gender value to write it.
• 6:48
  [SPEAKER_03]: All the sentences are written by me,
• 6:51
  [SPEAKER_03]: If, if I had written the book with you, I would have said if I had written that book.
• 6:56
  [SPEAKER_03]: Okay, I mean, that would have been a point to something like it doesn't stop me from writing the book myself.
• 7:00
  [SPEAKER_03]: So what's the threat here?
• 7:01
  [SPEAKER_01]: I agree with you.
• 7:02
  [SPEAKER_01]: You can say the same thing about chess playing computers, people are doing things like playing chess.
• 7:07
  [SPEAKER_03]: History of manufacturing.
• 7:08
  [SPEAKER_03]: There's lots of things I wish to do by hand that people got replaced by various kinds of machines.
• 7:12
  [SPEAKER_01]: Absolutely.
• 7:13
  [SPEAKER_01]: Well, actually, that's a good example, also.
• 7:14
  [SPEAKER_01]: So maybe now, if you are,
• 7:16
  [SPEAKER_01]: someone who's very good at making furniture by hand.
• 7:19
  [SPEAKER_01]: They might be less of a market for your skills.
• 7:22
  [SPEAKER_01]: So, this week, of course, in your spare time, I have friends and colleagues who do this.
• 7:26
  [SPEAKER_01]: They make furniture in this spare time, and I'm sorry, they post their furniture.
• 7:29
  [SPEAKER_01]: That's lovely sometimes on Facebook or other social media, whatever it is, and people admire it, et cetera.
• 7:35
  [SPEAKER_01]: And that's maybe satisfying to that person, and it's nice to get recognition that you get from your group or friends and colleagues and peers or whatever online.
• 7:42
  [SPEAKER_01]: But maybe this is not something that you can now do for living and more and more of course a lot of the companies this is what they're telling us they were trying to develop tools so that companies need to hire fewer people And they can increase productivity to bring down costs and then they will be no need to be engaged in with all this humans So they want salaries, they want all sorts of benefits and
• 8:04
  [SPEAKER_01]: Sure, sure, et cetera.
• 8:05
  [SPEAKER_01]: So this is one of the ways in which it might be a little bit of a problem because yes, you can spend only a time writing books, creating arts, making decisions, making recommendations, but maybe no one wants to pay you to do it.
• 8:15
  [SPEAKER_01]: And so you would have to do it in your spare time.
• 8:18
  [SPEAKER_01]: Now, of course, people might say, actually, I just think of my job as a job and what I do in my spare time, that's why I really live my life, so to speak.
• 8:27
  [SPEAKER_01]: Yeah, yeah.
• 8:27
  [SPEAKER_01]: Other people might say, no, if this is not something that sort of is society's recognizing where I can't make a living, where it has to be demoted to a hobby, so to speak, then suddenly writing books and you're making whatever medical recommendations or whatever work it is I'm doing, might seem to lose some of its value, at least it's socially recognized value.
• 8:48
  [SPEAKER_01]: So that's one way of approaching the issue why would this be something to be concerned about?
• 8:53
  [SPEAKER_03]: Yeah.
• 8:54
  [SPEAKER_03]: Okay.
• 8:55
  [UNKNOWN]: So
• 8:55
  [SPEAKER_03]: There's sort of two tracks going on in this conversation that feels like one is what is about the value and meaning of something like creative, relatively idiosyncratic work by people who are for lack of a better word, passionate about that work or find that work intrinsically engaging blah, blah, blah, blah.
• 9:12
  [SPEAKER_03]: That's sort of one path that you're worried about AI undermining that.
• 9:15
  [SPEAKER_03]: And then the other side is, I think you think this, tell me if I'm getting you wrong,
• 9:21
  [SPEAKER_03]: whatever the work is, even if you're not passionate about construction management or whatever it is, you nonetheless find meaning and fulfillment in it.
• 9:30
  [SPEAKER_03]: So it might be, it might not be the case that but you're not driven to the work in a way that ideally any way an artist is driven to his or her work, but rather that there's something about the doing of work that is meaning making it creates value in one's life.
• 9:45
  [SPEAKER_03]: And that's another, it's the mundane work is nonetheless
• 9:50
  [SPEAKER_01]: Yeah, indeed, I'd used to be part of an Italy's binary project with some works, psychologists, and they would go out, I would do the philosophizing in a post-doc, and then we would have a post-doc in the works psychology and a professor, and the empirical research send us in, they would go out and interview people working in warehouses and things like that, and we were surprised at least with the philosophers about what sort of things people would find fulfilling and what aspects of the job that they would actually, they didn't want to automate
• 10:20
  [SPEAKER_01]: Yeah, sometimes are operating with these sort of ideas about what people find meaningful and what they don't find meaningful there or it may be false.
• 10:27
  [SPEAKER_01]: Yeah, but nevertheless, yeah, there are many ways of approaching this I mentioned before that maybe there's less going to be less of a market and less interested in hiring someone to do a certain kind of task.
• 10:37
  [SPEAKER_01]: And another thing that I also worry about in terms of it is to just go back to the initial statement about whether AI, Rome, or our success, is that there's something that I like to think about as the laziness trap, AI invites handing over tasks to it to AI and then of course, most tasks where people can excel or things going to be, that way they're going to be doing a lot, they have to practice, they have to learn, have to do a lot of work, before you reach a level where anyone would want to recognize you as a particular skill, writer, or just do something like
• 11:06
  [SPEAKER_01]: And again, if you have to put in a lot of work in order to reach a level where it's also possible to outsource it to AI, I do worry that also more fewer people will actually put in the work and learn the skills since have that level where they will actually reach.
• 11:21
  [SPEAKER_01]: People will have a recent admire them to hire them to sing the lemma out and so to speak.
• 11:27
  [SPEAKER_01]: Yeah, that's enough part of the worry here.
• 11:30
  [SPEAKER_03]: Yeah, I don't know.
• 11:31
  [SPEAKER_03]: I guess I'm not persuaded because
• 11:34
  [SPEAKER_03]: I think when people are intrinsically interested in something they do stuff maybe will there'll be so much pressure and do that you do to the laziness trap, but I don't know I think a lot of people still want to do it the other thing is that
• 11:45
  [SPEAKER_03]: I take it that, especially for more creative endeavors.
• 11:48
  [SPEAKER_03]: You have people who are in somewhere or are they infusing their personality into the thing that they make, or to put it as less poetically this, their personality shapes the thing that they make.
• 11:58
  [SPEAKER_03]: And while it's true that an AI can make an X whatever X is, whether it's a work of written art or visual art, or maybe we have AI robots that create woodwork, woodwork to tables, whatever, I take it that,
• 12:12
  [SPEAKER_03]: They're still going to be room for lots of different personalities to shape this thing, right?
• 12:17
  [SPEAKER_03]: So take the work, I'm going to take something really my day to maybe this is an instrument, but I'll do it anyway, because I don't know, you got me talking about what we're working now.
• 12:24
  [SPEAKER_03]: There are so many different woodworkers out there.
• 12:26
  [SPEAKER_03]: There's so many different ways to make a table.
• 12:28
  [SPEAKER_03]: There's so many different tastes, styles, something along those lines of aesthetics, and I don't even know how many hundreds, thousands.
• 12:35
  [SPEAKER_03]: So I think, okay, yes.
• 12:37
  [SPEAKER_03]: will have maybe fewer generic things.
• 12:40
  [SPEAKER_03]: I don't know that AI is going to get to the point, maybe it will, but I don't see it yet.
• 12:45
  [SPEAKER_03]: Where AI gets to the point where it manifests the set of all personalities, instantiated by humanity, and then creates, would work tables in that, in a line with that, with that personality, and then just do the same thing for the
• 13:01
  [SPEAKER_03]: So it seems to be like, you might find something.
• 13:04
  [SPEAKER_03]: But like, yeah, I really like this piece of art.
• 13:06
  [SPEAKER_03]: It was AI generator, whatever, manifest this kind of personality's quote unquote.
• 13:11
  [SPEAKER_03]: And other people say, yeah, I don't love the personality of this AI or that I actually like the personality of Sven and the work that he did.
• 13:17
  [SPEAKER_03]: So I'm going to buy that one.
• 13:19
  [SPEAKER_03]: I don't know.
• 13:19
  [SPEAKER_03]: I think just like what it comes to human tastes and aesthetic preferences and what I'll stop there.
• 13:27
  [SPEAKER_03]: Let a thousand flowers bloom, so to with humans plus AI.
• 13:31
  [SPEAKER_03]: Let a thousand flowers bloom and we'll have AI created stuff and human created stuff and some will prefer one, some will prefer the other, et cetera, et cetera.
• 13:38
  [SPEAKER_01]: Yeah, I should maybe mention that I'm not one of those people who are against AI, I have a possible way and I don't agree with my colleague in philosophy, Peter Kyrnig, so another professor in Germany who says that there's a little bit of a negativity bias in AI ethics research and one of the things I like to say when I introduce the topic is that you're
• 13:58
  [SPEAKER_01]: There is of course the question of what harms and risks and the problems are there related AI but also one of the opportunities for other possible benefits and how can we make this in something good and then did when we start out I did say that okay if there are meeting less boring and so on task that you can have outsource to AI then go for it of course
• 14:16
  [SPEAKER_01]: And as I'm saying that I'm also reminded that a lot of the things that where we do learn a skill and learn to excel there's going to be some boring and seemingly meaningless aspects of it that you have to go through.
• 14:27
  [SPEAKER_01]: And I do worry that the whole idea that AI can do it for you.
• 14:33
  [SPEAKER_01]: And what I call the laziness traffic, it is a real problem because I've been seeing recently people posting on academic social media about tools that help them to create literature reviews.
• 14:43
  [SPEAKER_01]: And so instead of going and reading some of the perhaps boring philosophical literature on topic, we would ask the AI to write a summary for you.
• 14:50
  [SPEAKER_01]: Yeah.
• 14:50
  [SPEAKER_01]: And that's the sort of thing that I think could actually bring down a lot of the quality of a lot of the work that a lot of people do because actually when you read work and you're looking for one thing maybe the thing that you find is another that's the thing that you want to go and run with sort of speak and especially if you do like work in the history of philosophy things like that.
• 15:08
  [SPEAKER_01]: Yeah.
• 15:08
  [SPEAKER_01]: and part of the point of reading the literature is to try to see something that other people haven't found.
• 15:13
  [SPEAKER_01]: And so, if more and more people start using these tools and it's so convenient, you don't have to believe the boring stuff.
• 15:19
  [SPEAKER_01]: Actually, the boring stuff can actually help to make actually being bored is one good way of boosting creativity.
• 15:27
  [SPEAKER_01]: If they're more entertained, you're the less reason you have to create something new to create something interesting.
• 15:31
  [SPEAKER_01]: You can just consume and be passive whereas if you're bored, you have a reason to be creative to do something new.
• 15:37
  [SPEAKER_01]: So on the one hand, I do agree with this idea that we shouldn't just be out for seeing problems and risks and negativity here.
• 15:44
  [SPEAKER_01]: There are also issues that we should worry about.
• 15:46
  [SPEAKER_01]: And this whole topic about, you know, is AI gonna make...
• 15:49
  [SPEAKER_01]: a lot of people lose some of their capacity.
• 15:53
  [SPEAKER_01]: Is it, how did they become too lazy?
• 15:55
  [SPEAKER_01]: Are they going to be fewer people who excel like whether it's, woodworking is probably a bad example, but because actually, okay, good, we can create furniture with machines, but it's going to be a different kind of thing.
• 16:05
  [SPEAKER_01]: And that's usually not what we talk about.
• 16:07
  [SPEAKER_01]: But we talk about AI, we should talk about outsourcing, sort of intellectual labor.
• 16:11
  [SPEAKER_01]: Now, I didn't want to go back to one of the things you said.
• 16:14
  [SPEAKER_01]: So let's say that you like high-person health and you want something written by me rather than AI.
• 16:19
  [SPEAKER_01]: Meanwhile, there has been this development where you can find tune AI, chatbot, so that they sound like particular people.
• 16:27
  [SPEAKER_01]: Sure.
• 16:28
  [SPEAKER_01]: And actually, this can now be done to a point where it's hard to tell the difference between the actual person and a chatbot imitation of them.
• 16:35
  [SPEAKER_01]: The famous philosopher Daniel Denner before he died, for example.
• 16:38
  [SPEAKER_01]: He participated in an experiment where some researchers took all of his writings and he wrote a lot.
• 16:44
  [SPEAKER_01]: Since the 60s up until recently,
• 16:46
  [SPEAKER_01]: And they fed that into, I think, GPT-3, or something like that.
• 16:50
  [SPEAKER_01]: And then asked it to only produce outputs that would sound similar to those last layers of training data.
• 16:56
  [SPEAKER_01]: And indeed, I'll even experts on that as well as we had trouble telling new snippets of text produced by this chatbot, DJ Dan, apart from...
• 17:04
  [SPEAKER_01]: You could get texts that have the sort of the person to have a particular person produced by AI.
• 17:12
  [SPEAKER_01]: And that might be cheaper also, like if Taylor Swift's record company doesn't want to pay her to produce new music, they might have a Taylor Swift bot that would produce an extra album.
• 17:23
  [SPEAKER_01]: She might have a reason to worry, and maybe you and I, she owns her IP.
• 17:26
  [SPEAKER_03]: She's who they'll add to them.
• 17:28
  [SPEAKER_03]: Okay, so I hear the point.
• 17:29
  [SPEAKER_03]: One caveat, of course, is that dang that head of it and so much, so they had to train on.
• 17:34
  [SPEAKER_03]: If Sven is up in coming and doesn't, hasn't published a lot, we don't really know what his right early voice is yet, because it hasn't been created yet, so he can't rip it off yet.
• 17:42
  [SPEAKER_03]: Okay, I do want to say that I see the point about spending time doing the boring stuff is important.
• 17:48
  [SPEAKER_03]: There's already this discussion about how we're going to fill the jobs of the future.
• 17:51
  [SPEAKER_03]: If the sort of more junior people are replaced by AI and the senior people are senior precisely because they went through that path of being a junior learning about your stuff, learning the nitty gritty, doing some tedious stuff.
• 18:04
  [SPEAKER_03]: And that's what allowed them to get better at the craft or get better at the job or have a deeper understanding of how the job works, et cetera, et cetera, how are you going to, if we, how are you going to get new senior people, if once these senior people die out, then the junior people weren't didn't exist, so how are they going to be the can't be senior, so I get that worry and I also see the point about.
• 18:23
  [SPEAKER_03]: especially for the kind of intellectual labor you were talking about the research that when you read an article right there's a lot of people will read articles especially in AI and something like wired magazine or MIT tech review or something like that and that's not so superficial usually those things when they're good not all the time but often based on real research you know that they've done in the lab so to speak and
• 18:49
  [SPEAKER_03]: When you go and read the original research, it's so much richer.
• 18:52
  [SPEAKER_03]: There's so much more stuff going on.
• 18:54
  [SPEAKER_03]: There's so much more subtlety and context.
• 18:55
  [SPEAKER_03]: That's not in that, say, MIT Tech Review article.
• 18:59
  [SPEAKER_03]: And so reading the research seems really important.
• 19:02
  [SPEAKER_03]: I also really like your point that sometimes you're reading the thing for reason X and then you stumble upon Y.
• 19:07
  [SPEAKER_03]: and that opens up door for you, but wouldn't have opened otherwise it seems, and be couldn't lose out on that.
• 19:13
  [SPEAKER_03]: And then to your point about the laziness trap, that all seems right, because you just think, I just want to know what these seven articles think, so I couldn't write my thing like missing anything based on what I've said, yes, no, okay, I haven't missed anything off we go, but perhaps you go to read each of those seven articles, maybe you just wanted them, you would have found some, but it's the expression, golden nugget, is that it's an expression?
• 19:32
  [SPEAKER_03]: I don't know, you would have found
• 19:33
  [SPEAKER_03]: You would have found something really valuable that you could use in further develop your research in a more interesting way.
• 19:39
  [SPEAKER_01]: Absolutely, and also one shouldn't forget that when experts write about a topic, whatever ends up in their articles is just a, it's not like the whole of their expertise.
• 19:48
  [SPEAKER_01]: The more you know about a subject, the more you have to select whatever you want to put in your research output, and then so,
• 19:54
  [SPEAKER_01]: when you read something written by an expert, you're already getting a kind of a summary, so to speak, of what they think is important by a topic.
• 20:01
  [SPEAKER_01]: So if you then have a summary of that, you're two steps removed from this rich body of expertise that's with human experts, and if you have fewer and fewer people who really develop deep expertise on different topics, and we're all just reading sort of summaries of summaries.
• 20:16
  [SPEAKER_01]: And I'm also one of those people who don't really like to talk about AI's summaries.
• 20:20
  [SPEAKER_01]: I mean, it's strictest speaking.
• 20:22
  [SPEAKER_01]: Some people say it's better to talk about ways of shortening texts.
• 20:25
  [SPEAKER_01]: Like when a human summarizes something, they judge what is important.
• 20:30
  [SPEAKER_03]: Right, they sit to size it.
• 20:31
  [SPEAKER_01]: And the synthesize, et cetera, they add, there may be interpretation, whereas an AI does something different.
• 20:36
  [SPEAKER_01]: And AI technology, like a large language model.
• 20:38
  [SPEAKER_01]: And it's not just that we're reading maybe summaries of summaries.
• 20:42
  [SPEAKER_01]: We're reading a short and diverse summaries.
• 20:44
  [SPEAKER_01]: Yeah, there are lots of reasons because concerned here if people rely too much on these tools.
• 20:50
  [SPEAKER_03]: Yeah, okay, let me now push back again.
• 20:52
  [SPEAKER_03]: I feel like we're getting to a point in this conversation where we're talking about a very tiny subset of the general population.
• 20:59
  [SPEAKER_03]: We're talking about
• 21:01
  [SPEAKER_03]: People, essentially we're talking about people with PhDs during research, something like that, and maybe even a subset of those, maybe it's more of the humanities that it is the natural sciences.
• 21:10
  [SPEAKER_03]: And I can be wrong about that.
• 21:12
  [SPEAKER_03]: But then again, there's lots of people with PhDs who are already lazy, who don't really do the reading, who don't really do the research, they trade on their big name, they're just good talkers.
• 21:21
  [SPEAKER_03]: We both know that there's people in philosophy who,
• 21:24
  [SPEAKER_03]: They're pretty lazy and they don't actually know what they're talking about, but they're good at speaking.
• 21:27
  [SPEAKER_03]: They're going to talking in persuading people that they do know what they're talking about.
• 21:30
  [SPEAKER_03]: Anyway, that's my extra credit.
• 21:31
  [SPEAKER_03]: But here's the upshadow of all this.
• 21:34
  [SPEAKER_03]: Yeah, there's already lazy people.
• 21:36
  [SPEAKER_03]: Maybe it's going to enable some more laziness.
• 21:38
  [SPEAKER_03]: But how much is it going to move the needle among those that already small subset of people?
• 21:45
  [SPEAKER_03]: And even if it does move the needle among that small set of people, what are the if any effects outside of those of that small group of people?
• 21:53
  [SPEAKER_01]: No, I think that's a very valid question and a good point.
• 21:56
  [SPEAKER_01]: And now the way I've approached it is actually some ideas by a couple of philosophers who had a really interesting paper about humanities papers in a university liberal programs.
• 22:07
  [SPEAKER_01]: So I think Timothy Ailsworth and Clinton Castro.
• 22:10
  [SPEAKER_01]: They have a paper about whether you should let, as if you were a student at doing a BA, or should you be writing your essays using chatty PT or some other large-liners tool?
• 22:20
  [SPEAKER_01]: And they argue that actually know this is, you might go on to work in a completely different field.
• 22:24
  [SPEAKER_01]: You're not going to be a professional philosopher or professor, or a professor, or whatever it might be.
• 22:29
  [SPEAKER_01]: But this is your one chance to engage with the big books and the great ideas of very good.
• 22:34
  [SPEAKER_01]: the past and present greatest thinkers and so are you going to waste this opportunity by reading summaries created by in a mindless AI program or are you going to take this opportunity during your 304 whatever years of college to actually engage with this great idea some of the sort of I have this opportunity to think about your values and your learn about culture so
• 22:55
  [SPEAKER_01]: that would make things a little broader about the class of people who don't college or doesn't have to be college level education can be before that in your high school you might do you can think about do you want do I want to read some of the whether you know the classics of your country whether you're traditional what might be what do you want to read the sort of
• 23:12
  [SPEAKER_01]: Of course, there was always the cliff notes and things like that before.
• 23:14
  [SPEAKER_00]: Sure.
• 23:14
  [SPEAKER_01]: Certain sense, this is nothing new.
• 23:16
  [SPEAKER_01]: What's maybe new is that people are also so extremely addicted to their phones.
• 23:20
  [SPEAKER_01]: That the first thought that a lot of people have when they have to do something or anything really is that to reach for their phone and think about, can I sometimes somehow use one of my apps?
• 23:29
  [SPEAKER_01]: One of the tools on this little Swiss army knife that I have in my pocket.
• 23:33
  [SPEAKER_01]: And yeah, this is of course this general issue that, okay, so good.
• 23:38
  [SPEAKER_01]: Not everyone is going to go on to be a researcher in a certain sense.
• 23:42
  [SPEAKER_01]: You don't have to learn the skills because that's what we're going to do in your work life later on.
• 23:45
  [SPEAKER_01]: But still, it might be this wonderful opportunity.
• 23:47
  [SPEAKER_01]: And a lot of people look back at it time and when they were students, whether in high school in college and think about how great and may have been to engage with this great ideas and that inspiring.
• 23:57
  [SPEAKER_01]: That would make it a little bit wider, but you're right.
• 24:00
  [SPEAKER_01]: Still, of course, we've been talking about research and thinking about ideas and things like that.
• 24:06
  [SPEAKER_01]: There are, of course, skills in all types of professions and lines of work, et cetera, where you have the same thing.
• 24:12
  [SPEAKER_01]: You can hand over the tasks to AI tools, or you can do them yourselves.
• 24:17
  [SPEAKER_01]: and if you want to get good at them you have to practice and you have to start or down as you mentioned and work your way up or you can just go directly from beginner to senior precision without having to do trading or are you going to then have the relevant skill set and yeah I get that point I don't know I think
• 24:38
  [SPEAKER_03]: We, there's lots of, there's lots of work to be done that has to do with this is actually a point that came up when I was talking to Luciano 3 where you mentioned just early on a conversation, which is that
• 24:49
  [SPEAKER_03]: For instance, artists don't necessarily create the physical work of art.
• 24:54
  [SPEAKER_03]: They might direct others, hey, here's what I want you to do.
• 24:56
  [SPEAKER_03]: Here's how I want you to create it.
• 24:57
  [SPEAKER_03]: Architects don't physically build the building.
• 24:59
  [SPEAKER_03]: They direct others had a build it and all that, but they don't actually physically build it.
• 25:04
  [SPEAKER_03]: Still, they're the artist responsible for the beauty of the building, no, whatever it is.
• 25:08
  [SPEAKER_03]: And so,
• 25:09
  [SPEAKER_03]: to the extent that we interact with these things.
• 25:11
  [SPEAKER_03]: And maybe we're not getting in the nitty gritty, but we're still directing it in various ways.
• 25:15
  [SPEAKER_03]: We're still exercising a kind of expertise or judgment or sensitivity to nuance.
• 25:22
  [SPEAKER_03]: So part of me thinks, are we just getting to,
• 25:27
  [SPEAKER_03]: I don't know.
• 25:28
  [SPEAKER_03]: Something like anxious about people being able to play the role of something like creative director.
• 25:34
  [SPEAKER_01]: I think that's a nice example and a nice question.
• 25:36
  [SPEAKER_01]: And again, here, let's say that someone like Lucia and a Florida to take him as an example, just because you brought him up again.
• 25:43
  [SPEAKER_01]: Let's say that he uses generative AI to produce his next philosophy paper.
• 25:47
  [SPEAKER_01]: He has so much experience with writing papers and doing great philosophy works, that he's going to be able to judge what counts as good outputs from the generative AI tool.
• 25:56
  [SPEAKER_01]: and might be able to use it in a creative and new kind of way.
• 25:59
  [SPEAKER_01]: But if you give this tool to someone as a beginner, and who is, this is, I saw something I actually always tell my students, and you have don't have a lot of experience yet with judging what counts as good or poor philosophy.
• 26:10
  [SPEAKER_03]: Yeah, for sure.
• 26:11
  [SPEAKER_01]: And plus there's the laziness trap that we mentioned on the other hand, you have a lot working against you.
• 26:16
  [SPEAKER_01]: And so in order to make sure that you learn how to judge what is good and what isn't good, art, architecture, philosophy, whatever it might be.
• 26:23
  [SPEAKER_01]: You might still first need to train without the tool in order to be able to optimize your use of the tool.
• 26:29
  [SPEAKER_01]: And after that's one idea, I thought I have about that.
• 26:32
  [SPEAKER_01]: Another is that I agree with the idea that we do sometimes admire people who play this more somewhat removed role.
• 26:39
  [SPEAKER_01]: They have a vision and then someone else has to put it in the work.
• 26:42
  [SPEAKER_01]: Sometimes, though, I think we don't give enough credit to whoever has to do that work.
• 26:47
  [SPEAKER_01]: That's maybe mentioned, fine, pretty much so.
• 26:50
  [SPEAKER_01]: Well, if you think, for example, about your to the art museum, you see some of this old master paintings, and then actually a lot of the canvas was painted by one of the people in the apprentices in the studio and maybe it's mentioned that it was done in the studio of this and that great master no name, so then you're those people who painted the sky at the landscape is going to be mentioned and so they might not get the credit that they deserve to get.
• 27:13
  [SPEAKER_01]: Of course, it's an interesting situation if you replace those human apprentices with a generative AI tool and then you make the same point, should then give credit to the generative AI tool.
• 27:25
  [SPEAKER_01]: Here there might be an interesting difference though, because the person who's trying to become the next master painter, they might want to get the credit because that's a way for them to get the credentials and to say, okay, maybe I can replace the great master painter when they die or retire whatever.
• 27:39
  [SPEAKER_01]: where as the janitor AI tool they don't it doesn't care about of course you can turn it into a chat but they will say that it cares et cetera but there's a point of where you might think that you should give more credit to the people building or producing the art that's just going to be quite different to this one way
• 27:56
  [SPEAKER_03]: I actually think I'm going to I'm going to put your thesis slightly differently and I actually think maybe there's one which our conversation about expertise and gaining doing this sort of nitty gritty stuff doing the boring stuff in some cases in order to rise to that level expertise and all this talk about who's responsible for it given that it wasn't created by someone's okay.
• 28:16
  [SPEAKER_03]: I think that what your thesis really is something like, we need to do these kinds of things, we need to exercise some degree of, I'm gonna put it boldly right now, we need to exercise some degree of excellence because that's what constitutes, I'm telling it, really was strong, strongly.
• 28:33
  [SPEAKER_03]: That's what constitutes human flourishing.
• 28:34
  [SPEAKER_03]: So I think that you're not concerned primarily about unemployment, we talked a little bit about that as a sub, but that is a concern perhaps with us, that's not what really,
• 28:43
  [SPEAKER_03]: The unemployment there would be the main problem aside from the camp buy stuff is for unemployed, we can't engage in the work, whether we think it's whether we fall off for us, let's say it's Monday or not.
• 28:55
  [SPEAKER_03]: We can't engage in everyday work that is consistent to human flourishing.
• 28:59
  [SPEAKER_03]: If we don't do the lower level stuff, we can't reach a level of expertise that constitutes human flourishing.
• 29:05
  [SPEAKER_03]: So in some words, I just want to recast your whole thing in terms of you care a lot about human flourishing.
• 29:10
  [SPEAKER_03]: That's a good thing.
• 29:10
  [SPEAKER_03]: And you're in really you think that
• 29:14
  [SPEAKER_03]: It's only to undermine the conditions under which humans can flourish.
• 29:18
  [SPEAKER_01]: Yeah, so the way that you're originally introduced to a whole idea, you also mentioned that they're living a meaningful life.
• 29:24
  [SPEAKER_01]: Yep.
• 29:24
  [SPEAKER_01]: And indeed, I'm a bit of a pluralist when it comes to the idea of meaning a life.
• 29:29
  [SPEAKER_01]: I don't think that's, for example, flourishing or achievement is the one and only thing.
• 29:33
  [SPEAKER_01]: The be all in hand all of that.
• 29:35
  [SPEAKER_01]: But I think that's one aspect.
• 29:36
  [SPEAKER_01]: So,
• 29:37
  [SPEAKER_01]: if you could have life in which you know you have rich and rewarding relationships where you flourish in the sense of reaching excellence in a certain art or research or carpentry whatever it might be if you also do good for others so that you're being in the world makes the world a better place and there are some other ideas that I would also want to add you know especially if there's a bunch of ingredients that's right so yeah not if we're sure
• 30:02
  [SPEAKER_03]: That's probably none of them are necessary, but some joint joint diversification.
• 30:06
  [SPEAKER_01]: That's right.
• 30:06
  [SPEAKER_01]: And so that you can ask adding more and more AI to every aspect of your life.
• 30:12
  [SPEAKER_01]: Is that going to make it harder to achieve this different aspects of a meaningful life?
• 30:16
  [SPEAKER_01]: Or is it going to make it easier?
• 30:18
  [SPEAKER_01]: That's one question.
• 30:18
  [SPEAKER_01]: Another question is, are the companies that are trying to develop these tools?
• 30:22
  [SPEAKER_01]: Is this what they're trying to optimize the maximize so to speak?
• 30:25
  [SPEAKER_01]: Are they trying to make our lives more meaningful?
• 30:32
  [SPEAKER_01]: What they say some other time, but when they're trying to really sell their products, they're usually talking more about your boosting productivity and making it less expensive to run companies or doing this or that.
• 30:43
  [SPEAKER_01]: Yeah, so there might be a clash here between, and actually if you look at, as I was writing, my new book, I was looking at some positive statements from AI companies about what's good about their tools, and then you would, indeed, have these kinds of things.
• 30:58
  [SPEAKER_01]: It's productivities, lowering costs, speed, and things up.
• 31:02
  [SPEAKER_01]: Sam Altman from Open AI talked about just entertainment.
• 31:06
  [SPEAKER_01]: more in AI in the world means more memes and things like that and the Microsoft they have of what they call Microsoft for good or something like that.
• 31:14
  [SPEAKER_01]: They have a lab that's developed and how can we make a societal impact and then what they say is okay we're going to have better image classification, more data, better data analysis, etc.
• 31:25
  [SPEAKER_01]: That's
• 31:26
  [SPEAKER_01]: all sounds great but none of it seems to be yeah we're gonna boost those things that make life more meaningful whereas you might indeed worry that the sort of key ingredients of meaning life may be possibly be areas where you don't really need AI the way just flourish and those when it comes to good relationships or learning to excel in some domain for those things you may not need AI and actually AI might be hindrance rather than help.
• 31:54
  [SPEAKER_03]: I'm going to say another thesis.
• 31:55
  [SPEAKER_03]: I don't know that it's yours.
• 31:57
  [SPEAKER_03]: I just want to hear you react to it.
• 31:59
  [SPEAKER_03]: Suppose I said something like, OK, so Sven, here's the issue by your lights.
• 32:04
  [SPEAKER_03]: It's that the AI companies or LLMs or whatever.
• 32:08
  [SPEAKER_03]: AI generally is good at speeding up time to get into your goal, whatever the goal is, getting the thing that you want, getting the object of your desire, whatever it is.
• 32:18
  [SPEAKER_03]: And you think that human flourishing is not mainly about getting the thing, but about doing the things.
• 32:26
  [SPEAKER_03]: So it's more to put it on the side of the mug, it's about the journey, not about the destination.
• 32:32
  [SPEAKER_03]: And by robbing us of the journey and getting us to the destination too quickly, they're robbing us of the condition on which we've learned.
• 32:40
  [SPEAKER_01]: No, yeah.
• 32:41
  [SPEAKER_01]: I agree.
• 32:41
  [SPEAKER_01]: You can put it in terms of the journey or the reaching your final destination or something like that.
• 32:46
  [SPEAKER_01]: Or you can be a little bit more dry and theoretical and use some terminology.
• 32:49
  [SPEAKER_01]: Sure.
• 32:50
  [SPEAKER_01]: I think it's Rob Goodman, political theorist if I hope I'm not getting his name wrong.
• 32:55
  [SPEAKER_01]: It makes a distinction between what it calls process goods and outcome goods.
• 32:59
  [SPEAKER_01]: And what process goods are things such as the creativity, the ingenuity, the, I don't know, using a good taste.
• 33:08
  [SPEAKER_01]: producing what whether it's art, good solutions, or if there's a problem.
• 33:12
  [SPEAKER_03]: I have no good relationship.
• 33:13
  [SPEAKER_03]: I have a good relationship with colleagues, peers, etc.
• 33:17
  [SPEAKER_03]: All of the things are the process.
• 33:18
  [SPEAKER_01]: That's right.
• 33:19
  [SPEAKER_01]: Those are things that part of the process.
• 33:20
  [SPEAKER_01]: And then at the end, there might be a report, a book, a product, a service, or something like that.
• 33:26
  [SPEAKER_01]: And indeed, it does seem that a lot of the AI,
• 33:29
  [SPEAKER_01]: high press around as you say get into the outcome as quickly as possible with as little effort as possible but a lot of other things that we value for example among work colleagues and what we recognize not obviously we do recognize people for the outputs that they produce but in terms so who makes you know for a good colleague with whom is it nice to go to work who would you want to hire
• 33:51
  [SPEAKER_01]: asked it to be a new colleague.
• 33:52
  [SPEAKER_01]: We don't only care about the output they produce.
• 33:54
  [SPEAKER_01]: That's obviously one thing that we care about, but also what is it like to be a member of a team with this person and it's going to be... And when you look back at maybe you're having work to the particular company organization and you move down and what are the things you miss?
• 34:09
  [SPEAKER_01]: Is it going to be the outputs that you produced?
• 34:11
  [SPEAKER_01]: Or is it going to be the process of working towards it?
• 34:13
  [SPEAKER_01]: So that's a different view of important part of life.
• 34:18
  [SPEAKER_01]: replace a lot of the process with AI tools that does those things for us, then indeed that might just take away some of the richness of the value of certain kinds of activities.
• 34:28
  [SPEAKER_01]: Again, I should emphasize that of course there are going to be aspects of work that maybe don't help you to learn your skills better, that are not particularly valuable, et cetera, et cetera.
• 34:38
  [SPEAKER_01]: So, such as these are things that would be nice to replace with AI.
• 34:42
  [SPEAKER_01]: And so there's something I like to use that expression, the AI sweet sport,
• 34:46
  [SPEAKER_01]: AI taking over just the right amount of tasks and more stuff, but not taking over the good stuff and not taking over so much that people don't learn the proper skills.
• 34:55
  [SPEAKER_01]: Of course, finding that AI's sweet spot is going to be a very hard and a sure.
• 34:57
  [SPEAKER_01]: So let me ask you a different kind of question.
• 34:58
  [SPEAKER_03]: I'm just going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going to, I'm going
• 35:08
  [SPEAKER_03]: Okay, look, I agree with you.
• 35:09
  [SPEAKER_03]: That all sounds right.
• 35:11
  [SPEAKER_03]: Judicious for who I'm flourishing and out of just about achieving certain goals or ends or something along those lines.
• 35:15
  [SPEAKER_03]: I don't desire satisfaction, but it's about engaging in the process of building good relationships, being creative, blah, blah, blah, blah.
• 35:24
  [SPEAKER_03]: Okay, so I'm on board.
• 35:25
  [SPEAKER_03]: I'm on board at the importance of process goods.
• 35:27
  [SPEAKER_03]: But look, that stuff that's going to take human effort to do, yes, there's the laziness trapped, I'll to speak, but it's up to humans to choose not to be lazy and you're not going to say this, but I'm going to pretend that you said it and you seem to think that it's the fault of the same old ones of the world and the other AI companies, you know, that they're building these AI that are going to do this stuff and increase productivity blah blah blah.
• 35:49
  [SPEAKER_03]: And it's kind of just undermined human flourishing as though humans are these completely passive things to the whims of their technology masters when in fact if we can't figure out how to do the hard work or do the things that are actually fulfilling that's on us that's on our
• 36:08
  [SPEAKER_03]: bankrupt culture, not the presence of AI.
• 36:11
  [SPEAKER_03]: If AI undoes the conditions for human flourishing or human flourishing takes a noose dives, statistically speaking, that's not the fault of AI or not solely the fault of AI.
• 36:20
  [SPEAKER_03]: It's the fault of a bankrupt culture that couldn't figure out how to prioritize value things properly.
• 36:27
  [SPEAKER_01]: Okay, yeah, so you might say that okay, so let's say that I make my life meaningless but using too much AI and I'm bowing down to this tech masters, et cetera, then I might deserve to have a meaningless life and it's my own fault.
• 36:40
  [SPEAKER_01]: So to speak, yeah, I do have a little bit of sympathy for that.
• 36:45
  [SPEAKER_01]: We must say although of course a lot of the people who are affected might be young people who are too young to really have a good idea of these things and to really be responsible who needs guidance for example.
• 36:55
  [SPEAKER_01]: And then if they're the university
• 36:57
  [SPEAKER_01]: I had been people and so on are like pushing, shoving AI down their throats, saying it's a future.
• 37:04
  [SPEAKER_01]: Then in a certain sense, maybe this young people who then are throwing away opportunities by buying into this, it's maybe not their fault, well, it's maybe other people.
• 37:12
  [SPEAKER_03]: Yeah, I think that's, yeah.
• 37:13
  [SPEAKER_01]: But yeah, so I'll go.
• 37:14
  [SPEAKER_03]: Of course I say though, I hear that point.
• 37:16
  [SPEAKER_03]: I think that's a good point.
• 37:17
  [SPEAKER_03]: It's a great point.
• 37:17
  [SPEAKER_03]: Yes, like some people,
• 37:19
  [SPEAKER_03]: the young ones, particularly don't know what they're choosing.
• 37:21
  [SPEAKER_03]: They just, they don't know.
• 37:22
  [SPEAKER_03]: So that's why I'm trying to push something like it's the culture's fault or its humanities fault or it's our collective fault for fostering a culture of deference to technology, something like that.
• 37:35
  [SPEAKER_01]: It is interesting, at this particular moment in time, the amount of influence we as a society in culture need allow the tech world to have on our lives, the way more organized society, it's maybe getting a little bit out of hand and I do think that you are seeing more and more maybe not resistance is maybe too strong but like the discussion about do we really want this?
• 37:57
  [SPEAKER_01]: Is this really
• 37:58
  [SPEAKER_01]: whether your Trump supporter or not was very striking that when he was inaugurated for a second term like there were so many tech people that were there for inauguration who took a very prominent sort of positions in the room and they were highlighted just showing what important roles they have like as this in the US aside but not just in the US all over the world you have this and a lot of people see that then they feel that actually we voted for let's say Trump in this case or whoever they voted for we didn't vote for Mark Zuckerberg or whoever
• 38:26
  [SPEAKER_01]: And find if he soccer broke, put himself up for election, and he's voted into office, then let him rule us, so to speak.
• 38:35
  [SPEAKER_01]: But if it's, if these people are not elected representatives, but they're then invited to take over things, then you might think that even if you are a fan of the Mark Zuckerberg's, Sam Alpen, so the world, you might still want them to have that position as part of a political process as opposed to just
• 38:55
  [SPEAKER_01]: When it comes to something like democracy, that's also something where some people have a completely instrumental view of it.
• 39:01
  [SPEAKER_01]: So if it helps us that's fine and we don't care how the sources just made, whereas other people think that no, having the rights that you have in the liberal democracy, having the sort of the status as being a citizen with all sorts of basic rights, etc.
• 39:15
  [SPEAKER_01]: That's a valuable thing in and of itself.
• 39:17
  [SPEAKER_01]: And maybe we don't want to outsource that AI and the tech world, so to speak, we want to have a process that helps not just give us, but also constitutes our having these cancer rights.
• 39:28
  [SPEAKER_01]: And that's one kind of concern, which is again, it's compatible with thinking that yes, the tech world should have a lot of influence, but then we'd want them to be voted into that influence and not just having them by their way into it or a lobby their way into it, maybe I should say.
• 39:41
  [SPEAKER_03]: and sell their way into it.
• 39:43
  [SPEAKER_03]: They sell the stuff that we buy it because we think, oh, this is amazing.
• 39:46
  [SPEAKER_03]: This is so great.
• 39:47
  [SPEAKER_03]: This is so convenient.
• 39:48
  [SPEAKER_03]: I had to be lazy now.
• 39:49
  [SPEAKER_03]: So...
• 39:50
  [SPEAKER_03]: Okay, that's just me saying a part of it has to be our responsibility.
• 39:54
  [SPEAKER_01]: Absolutely, I agree.
• 39:56
  [SPEAKER_01]: And also, I think people are really being not just lazy in the sense that this convenience to use these tools.
• 40:02
  [SPEAKER_01]: They are not also a bit lazy sometimes with respect to what finding out what exactly does the tool do, how does it work, what is it good at, what isn't it good at?
• 40:10
  [SPEAKER_02]: Yeah.
• 40:11
  [SPEAKER_01]: And obviously it then falls partly on people such as myself, a work in education to think about how can we
• 40:19
  [SPEAKER_01]: Of course, they can listen to your podcast.
• 40:21
  [SPEAKER_01]: That's the one thing they can do.
• 40:23
  [SPEAKER_01]: And so And you can have different kinds.
• 40:25
  [SPEAKER_03]: That's the boring stuff for them.
• 40:26
  [SPEAKER_03]: They just want to get a somewhere out of the transcripts.
• 40:28
  [SPEAKER_01]: That's right.
• 40:30
  [SPEAKER_01]: That's What then one this has to produce content about how AI works and that AI can then some rise in a nice way So that people can learn about something.
• 40:37
  [SPEAKER_03]: I don't know.
• 40:38
  [SPEAKER_03]: I actually go back and forth on this.
• 40:40
  [SPEAKER_03]: Sometimes I think Yeah listen people you're in Facebook don't told you got to use Facebook.
• 40:44
  [SPEAKER_03]: You know like Facebook you think it's a cesspool don't use Facebook get off it
• 40:47
  [SPEAKER_03]: On the other hand, sometimes I think to your point, they don't know how things work.
• 40:50
  [SPEAKER_03]: And then those are questions about, are they responsible for knowing how things work?
• 40:54
  [SPEAKER_03]: And sometimes I think,
• 40:55
  [SPEAKER_03]: No.
• 40:56
  [SPEAKER_03]: So take like safety and aviation.
• 40:59
  [SPEAKER_03]: I don't know what the regulations are.
• 41:00
  [SPEAKER_03]: I don't know what constitutes safe aviation.
• 41:02
  [SPEAKER_03]: I don't want to know.
• 41:03
  [SPEAKER_03]: I don't want to do that.
• 41:04
  [SPEAKER_03]: I don't want to, hey, read, do your research on the 747.
• 41:06
  [SPEAKER_03]: I don't want to do my research.
• 41:08
  [SPEAKER_03]: That's, forget it.
• 41:09
  [SPEAKER_03]: It's too much work.
• 41:10
  [SPEAKER_03]: I want the government to come in and say, hey, here are the safety standards that we have experts, these are the safety standards that you have to read, reach this.
• 41:17
  [SPEAKER_03]: And so we don't expect.
• 41:19
  [SPEAKER_03]: every person who gets on a plane, every person who drives a car to understand crumples Owens.
• 41:23
  [SPEAKER_03]: That can't, that can't be, that's too high of a bar.
• 41:26
  [SPEAKER_03]: We have government protection from predatory lending.
• 41:29
  [SPEAKER_03]: Great.
• 41:29
  [SPEAKER_03]: I don't want to have to learn about what constitutes predatory lending and then constantly be in the lookout.
• 41:34
  [SPEAKER_03]: Whenever I talk to a credit card company, someone goes lines.
• 41:37
  [SPEAKER_03]: So, part of me thinks, hey, listen, it's on you, do the work, make the decision, other part of me thinks it's too much, we can't handle all that information, we can't do all that research, we need a protector, so we do need government to come in and say, hey, we got this, you don't have to read all these things and learn all this stuff out, let's say, how large language models work, that's way too big of a lift for the average citizen, we got this.
• 42:02
  [SPEAKER_01]: Yeah, no, so I agree, and some of those examples that you mentioned, like aviation and maybe also the medical profession, those are good examples where in addition to having the government regulate this very heavily so to speak for the sake of safety, you also within those professions have also some boards that certify colleagues and work on boosting safety status and I think
• 42:24
  [SPEAKER_01]: actually the tech industry, the AI industry is interestingly ambivalent with respect to this.
• 42:28
  [SPEAKER_01]: Most of the major tech companies that work on AI, they also have people that are hired to work what they call AI safety and they have lots of people within the organization who argue that these tools are, again, they provide lots of benefits but they're also also risks.
• 42:43
  [SPEAKER_01]: But one interesting thing is that at this
• 42:45
  [SPEAKER_01]: They wobble back and forth a little bit sometimes, they go out and say, okay, we need the government to regulate or sometimes they go and say that no, we understand this better than the government, so we should regulate ourselves.
• 42:55
  [SPEAKER_01]: But also, given just the sums of money that are involved in the social influence, et cetera, there's something intoxicating about being in that world.
• 43:03
  [SPEAKER_01]: So you see a lot of AI companies and let's not just only pick an open AI, anthropic is an interesting company because they are,
• 43:11
  [SPEAKER_01]: At the same time, marketing themselves is the AI safety and philosophically and otherwise informed company.
• 43:18
  [SPEAKER_01]: However, at the same time, they also are pushing ahead with this products and they even say at the beginning of the preamble, I think, of their AI constitution that, yes, we recognize a kind of, almost like a paradox here.
• 43:30
  [SPEAKER_01]: We are saying that these tools are risky and problematic and might lead to also the problems and yet we are one of, they think of themselves as one of
• 43:40
  [SPEAKER_01]: And so we're pushing ahead and I call that in my book, the AI Wager.
• 43:44
  [SPEAKER_01]: We're betting on AI to solve our problems at the same time, we're gambling with the, well, at the extreme and with the future of humanity.
• 43:52
  [SPEAKER_01]: I mean, maybe at the less extreme and with the future of excellence in some of the things that we've talked about.
• 43:56
  [SPEAKER_01]: Sure.
• 43:57
  [SPEAKER_01]: Sure.
• 43:57
  [SPEAKER_01]: And the question is, is it responsible to do this?
• 44:00
  [SPEAKER_01]: Are we proceeding like some of the people in the aviation industry with all sorts of regulations, safety measures, and redundant systems, et cetera, to make sure that the plane gets from A to B in a safe way that it's going to land without a problem?
• 44:13
  [SPEAKER_01]: Or are we pushing ahead?
• 44:14
  [SPEAKER_01]: actually in space travel there used to be a enormous safety culture and then at some point during I guess the 80s or something like that I might be getting my decade wrong.
• 44:24
  [SPEAKER_01]: There was a more lakhs safety culture all of a sudden and then indeed space shuttle started crashing and there were all sorts of problems where that's the general level of a surveillance about safety just went down and I
• 44:37
  [SPEAKER_01]: I worry that AI development is more like the space industry and the eight is whatever decade it was where, yes, there are safety measures and people are working on this, but this idea of taking the next step as quickly as possible and out-compete all your competitors is just so intoxicating that even though there's that tendency to care about safety stuff, it's just
• 44:59
  [SPEAKER_01]: We're not proceeding in a kind of the manner of someone who is driving a car while buckled up like sober Following all the traffic rules I'm making sure to be the discharge one is due to care towards other fellow travelers and passengers on the road Whereas It's more like being like in a drunk driver That's what you want to quickly get home after a night out on the town So yeah, there is a safety culture But it seems to be a little bit loose and Lex and Yeah, not quite doing everything that you should be doing
• 45:28
  [SPEAKER_03]: So let me ask you one last question.
• 45:30
  [SPEAKER_03]: So what is the thing?
• 45:31
  [SPEAKER_03]: If you had to pick on a top thing or top three things, whatever you like, of what we should be doing better, whether it's people or government or the AI teams, what ought they to be doing in order to not make it the case, that they are undermining human flourishing, et cetera, et cetera.
• 45:48
  [SPEAKER_01]: I think there's a division of labor here and we've talked about some of the different groups that have different tasks in my opinion and we've talked about government.
• 45:55
  [SPEAKER_01]: We've talked about researchers, students.
• 45:58
  [SPEAKER_01]: We've talked about the tech industry, etc.
• 46:01
  [SPEAKER_01]: And a lot of us everywhere now in society have this idea that for whatever is a problem, the solution is can be summarized in
• 46:11
  [SPEAKER_01]: In different ways, problematize and investigate this idea and ask, is that going to be right all the time everywhere for all problems or are they going to be problems where actually we don't need AI we might need.
• 46:25
  [SPEAKER_01]: Here's a good example.
• 46:26
  [SPEAKER_01]: We spoke about governments and politics.
• 46:28
  [SPEAKER_01]: There are more more people everywhere.
• 46:30
  [SPEAKER_01]: and populations are bigger and bigger and if you think of democracy as something where you want to have dialogues and people talking and deliberating together, the more people you have that the harder it is to do this.
• 46:41
  [SPEAKER_01]: And so what's the solution?
• 46:42
  [SPEAKER_01]: Some people say AI.
• 46:43
  [SPEAKER_01]: Yeah, if Google has something called the Habermas machine that's supposed to take inputs from a lot of people in creative sort of a sort of a sort of a sort of middle path between all these views.
• 46:52
  [SPEAKER_01]: They call it the Habermas machine.
• 46:53
  [SPEAKER_01]: They call it the Habermas machine, yes.
• 46:55
  [SPEAKER_01]: And maybe I didn't know that.
• 46:56
  [SPEAKER_01]: Yeah, no, and they have an interesting research paper about this, and actually there was an interesting and responsible philosopher said that actually however mass always has emphasized the idea of dialogue direct dialogue people actually talk he wears the harbour mass machine takes written statements by people and translates that into a sort of some sort of middle right path, but there's no rational dialogue.
• 47:18
  [SPEAKER_01]: No, and then people are asked to rate the output of the large language ball that this album must be shown.
• 47:23
  [SPEAKER_01]: So you take away that idea of having people talk with one another.
• 47:26
  [SPEAKER_03]: But they are the problem with the people who made the album must be shown.
• 47:30
  [SPEAKER_03]: They just use an AI to somewhere it's hover mouse.
• 47:31
  [SPEAKER_03]: And they got the wrong idea.
• 47:32
  [SPEAKER_03]: That's right.
• 47:33
  [SPEAKER_01]: Yeah, it's a go.
• 47:34
  [SPEAKER_01]: But anyways, the point I was going to make was that, yeah, so we all have to think about whether in our particular fields, what are the good ways of using AI, what are the parts of our disciplines of practices where we actually don't need AI and where actually we should wait with using AI until we have learned the skill and then we can use it in a kind of productive way.
• 47:54
  [SPEAKER_01]: And so, again, we're trying to find that a high-speed sport, that's a elusive goal, maybe, but still, we do have to question this assumption that for any problem the solution is called AI, so to speak.
• 48:05
  [SPEAKER_01]: On the other hand, we shouldn't be to total AI skeptics and say that AI is not going to help us with anything, so let's just go back to hunter-gatherers, a site or something like that.
• 48:13
  [SPEAKER_01]: That's, that's installed to be the wrong person, so it's just, that's think more about
• 48:19
  [SPEAKER_01]: Where can we use AI in a fruitful and useful and good way and where is it actually not useful?
• 48:25
  [SPEAKER_01]: I've mentioned a lot of tech companies and let me bring up one more Google just put out a new music making generative AI tool a couple of days ago, that's recording this.
• 48:34
  [SPEAKER_01]: And on the website they say that this new version of Gemini understands musicality.
• 48:40
  [SPEAKER_01]: because it's been developed with the help of musicians that helped develop the tool.
• 48:44
  [SPEAKER_01]: And as I read that, I just thought, okay, good.
• 48:47
  [SPEAKER_01]: If it's true that this new AI tool is conscious and understands the conscious experience of music, which I think is probably the most important part of music.
• 48:56
  [SPEAKER_01]: What it's feels like, how you experience it.
• 49:00
  [SPEAKER_01]: If that's all true, then yes, you might have an AI tool that understands musicality and music.
• 49:05
  [SPEAKER_01]: But if you have what is more likely to be the case,
• 49:09
  [SPEAKER_01]: For sure, without consciousness, that cannot experience or feel music, then the aspect of musicality that it can understand to a quote, I'm going to be the less, the not important aspect, the most important part of music is what it sounds like, what it feels like in our experience.
• 49:26
  [SPEAKER_01]: And that, until you have a conscious form of AI, you're not going to have this statement that our new AI tool understands musicality,
• 49:33
  [SPEAKER_01]: That's just shows that maybe there's a bit of a wrong idea of what is important in the field of music, for example, is the patterns in data sets, maybe on some level, that is part of the, what the music is, but on the other hand, it's now with that development.
• 49:47
  [SPEAKER_03]: Reducing or emotional issues, other people to something like neurological explanations.
• 49:51
  [SPEAKER_01]: Sure.
• 49:51
  [SPEAKER_01]: Sure.
• 49:51
  [SPEAKER_03]: That's that some level of explanation, it's true, it's just not helpful, or the full picture, or the most important part of the picture.
• 49:57
  [SPEAKER_01]: That's right.
• 49:57
  [SPEAKER_01]: So it behooves us to reflect on what do we value in different parts of life, what is music, whether it's research, whether it's whatever it might be medical, yep.
• 50:05
  [SPEAKER_03]: All right.
• 50:07
  [SPEAKER_03]: Sven, this has been great.
• 50:08
  [SPEAKER_03]: We'll have to reconnect it in a little while and see that.
• 50:10
  [SPEAKER_03]: Human flourishing is on the decline.
• 50:12
  [SPEAKER_03]: All right.
• 50:12
  [SPEAKER_03]: Thanks a lot.
• 50:13
  [SPEAKER_03]: All right.
• 50:14
  [SPEAKER_03]: Thanks, Sven.
• 50:17
  [SPEAKER_03]: Thanks everyone for listening.
• 50:18
  [SPEAKER_03]: I hope you enjoyed the episode.
• 50:19
  [SPEAKER_03]: And I want to thank a few people who make this all possible.
• 50:22
  [SPEAKER_03]: Delano Alexander, he's my producer.
• 50:24
  [SPEAKER_03]: He does such an amazing job on everything that he touches.
• 50:27
  [SPEAKER_03]: Garrett Lane is the sound engineer.
• 50:28
  [SPEAKER_03]: That's honestly, he's just, he's too good for me.
• 50:31
  [SPEAKER_03]: Authamia Jairam has joined us recently.
• 50:33
  [SPEAKER_03]: He's helping us find the best guests to bring on to this podcast.
• 50:36
  [SPEAKER_03]: And Ashley Wasselenko helps us organize all this.
• 50:39
  [SPEAKER_03]: Get people on the schedule.
• 50:40
  [SPEAKER_03]: Get people's bios, headshots, and really keeps the train on the track.
• 50:43
  [SPEAKER_03]: So thanks guys, really appreciate all the hard work you do.
• 50:56
  [UNKNOWN]: Thank you.