Ethics, AI, and Human-centered Computing with Dr. Casey Fiesler

hi I'm Scott Hanselman this is another hi I'm Scott Hanselman this is another episode of Hansel minutes today I'm episode of Hansel minutes today I'm episode of Hansel minutes today I'm chatting with Dr Casey fer she's an chatting with Dr Casey fer she's an chatting with Dr Casey fer she's an associate professor at the department of associate professor at the department of associate professor at the department of information science and computer science information science and computer science information science and computer science at the University of Colorado Boulder at the University of Colorado Boulder at the University of Colorado Boulder she's a wonderful tech Communicator that she's a wonderful tech Communicator that she's a wonderful tech Communicator that I met on Tik Tok at Professor Casey how I met on Tik Tok at Professor Casey how I met on Tik Tok at Professor Casey how are you you I'm great thank you for are you you I'm great thank you for are you you I'm great thank you for having me thanks for hanging out so I've having me thanks for hanging out so I've having me thanks for hanging out so I've got this Theory I want to run by you as got this Theory I want to run by you as got this Theory I want to run by you as the framing for our conversation that Ai the framing for our conversation that Ai the framing for our conversation that Ai and chat Bots as a proxy are a new kind and chat Bots as a proxy are a new kind and chat Bots as a proxy are a new kind of user interface that we have not yet of user interface that we have not yet of user interface that we have not yet figured out and the way the reason that figured out and the way the reason that figured out and the way the reason that it feels weird is that we're kind of it feels weird is that we're kind of it feels weird is that we're kind of anthropomorphizing these text boxes anthropomorphizing these text boxes anthropomorphizing these text boxes thinking that we're talking to like a thinking that we're talking to like a thinking that we're talking to like a friendly agent and it's just not a friendly agent and it's just not a friendly agent and it's just not a person but don't realize that so this person but don't realize that so this person but don't realize that so this user interface remains rough that is an user interface remains rough that is an user interface remains rough that is an interesting Theory I think I could buy interesting Theory I think I could buy interesting Theory I think I could buy that I mean I think one of the biggest that I mean I think one of the biggest that I mean I think one of the biggest problems right now with the way that problems right now with the way that problems right now with the way that people are interacting with chat Bots people are interacting with chat Bots people are interacting with chat Bots and the way that they're being used sort and the way that they're being used sort and the way that they're being used sort of broadly is that people don't of broadly is that people don't of broadly is that people don't understand what they are and even understand what they are and even understand what they are and even recently in some of my social media recently in some of my social media recently in some of my social media comments people have said oh well it's comments people have said oh well it's comments people have said oh well it's just a search engine or it's just an just a search engine or it's just an just a search engine or it's just an encycl cedia and I'm like no it is it is encycl cedia and I'm like no it is it is encycl cedia and I'm like no it is it is not either of those things that is not either of those things that is not either of those things that is really valid I one of my favorite demos really valid I one of my favorite demos really valid I one of my favorite demos and the one that I do on almost all of and the one that I do on almost all of and the one that I do on almost all of my AI 101 talks is I do text completion my AI 101 talks is I do text completion my AI 101 talks is I do text completion and I say it's a beautiful day let's go and I say it's a beautiful day let's go and I say it's a beautiful day let's go to the and I ask the audience what's the

to the and I ask the audience what's the to the and I ask the audience what's the right answer and there isn't a right right answer and there isn't a right right answer and there isn't a right answer so then if I treat the chatbot as answer so then if I treat the chatbot as answer so then if I treat the chatbot as an information retrieval database and it an information retrieval database and it an information retrieval database and it returns Beach or park or Zoo people are returns Beach or park or Zoo people are returns Beach or park or Zoo people are go oh well the answer the right answer go oh well the answer the right answer go oh well the answer the right answer is Beach but that's just like playing is Beach but that's just like playing is Beach but that's just like playing Family Feud you know and just CU Steve Family Feud you know and just CU Steve Family Feud you know and just CU Steve Harvey says show me beach doesn't mean Harvey says show me beach doesn't mean Harvey says show me beach doesn't mean it was correct I think Family Feud is it was correct I think Family Feud is it was correct I think Family Feud is actually a really great uh comparison to actually a really great uh comparison to actually a really great uh comparison to use there one way that I've been use there one way that I've been use there one way that I've been describing it is being a sort of bundle describing it is being a sort of bundle describing it is being a sort of bundle of of of tropes one of the reasons that I think tropes one of the reasons that I think tropes one of the reasons that I think about it that way is I I got in a little about it that way is I I got in a little about it that way is I I got in a little bit of an argument once with someone in bit of an argument once with someone in bit of an argument once with someone in my social media comments somewhere about my social media comments somewhere about my social media comments somewhere about whether chat gbt could write good whether chat gbt could write good whether chat gbt could write good fanfiction cuz I was saying it writes fanfiction cuz I was saying it writes fanfiction cuz I was saying it writes terrible fanfiction because it's all terrible fanfiction because it's all terrible fanfiction because it's all very like exactly what you would expect very like exactly what you would expect very like exactly what you would expect like it's cool that you can say write me like it's cool that you can say write me like it's cool that you can say write me a piece of Star Trek fanfiction where a piece of Star Trek fanfiction where a piece of Star Trek fanfiction where Kirk is Dr House and Spock is Sherlock Kirk is Dr House and Spock is Sherlock Kirk is Dr House and Spock is Sherlock Holmes and and it can do that and it's Holmes and and it can do that and it's Holmes and and it can do that and it's very cool that it can do that but what very cool that it can do that but what very cool that it can do that but what you get back is exactly what you like you get back is exactly what you like you get back is exactly what you like the most predictable thing because the most predictable thing because the most predictable thing because because that's how it works and so I because that's how it works and so I because that's how it works and so I think that once you get past that think that once you get past that think that once you get past that novelty of like oh my gosh because it's novelty of like oh my gosh because it's novelty of like oh my gosh because it's giving you exactly what you want to hear giving you exactly what you want to hear giving you exactly what you want to hear which is cool that it can do that but which is cool that it can do that but which is cool that it can do that but not a particularly interesting not a particularly interesting not a particularly interesting fanfiction it's just a bundle of tropes

fanfiction it's just a bundle of tropes fanfiction it's just a bundle of tropes well this is great I love extended well this is great I love extended well this is great I love extended analogies that hop around so remember analogies that hop around so remember analogies that hop around so remember when we were kids and we would play Mad when we were kids and we would play Mad when we were kids and we would play Mad Libs and it would be like okay give me a Libs and it would be like okay give me a Libs and it would be like okay give me a color give me a famous person give me a color give me a famous person give me a color give me a famous person give me a piece of clothing and then you're like 8 piece of clothing and then you're like 8 piece of clothing and then you're like 8 years old and you go and tell a story years old and you go and tell a story years old and you go and tell a story yeah this person went to the mall and yeah this person went to the mall and yeah this person went to the mall and bought a purple dress and saw you know bought a purple dress and saw you know bought a purple dress and saw you know Jimmy Stewart at the mall like Ah that's Jimmy Stewart at the mall like Ah that's Jimmy Stewart at the mall like Ah that's hilarious because it's all it's so it's hilarious because it's all it's so it's hilarious because it's all it's so it's such an anachronism such random things such an anachronism such random things such an anachronism such random things all put together and you're eight and all put together and you're eight and all put together and you're eight and you think it's hilarious that like you think it's hilarious that like you think it's hilarious that like there's a purple dog or whatever that's there's a purple dog or whatever that's there's a purple dog or whatever that's just the same thing as temperature on a just the same thing as temperature on a just the same thing as temperature on a very large language model where you go very large language model where you go very large language model where you go hey that's not creative enough let's hey that's not creative enough let's hey that's not creative enough let's turn the temperature up and get a little turn the temperature up and get a little turn the temperature up and get a little more randomization and maybe Kirk and more randomization and maybe Kirk and more randomization and maybe Kirk and Spock and Sherlock Holmes are all on the Spock and Sherlock Holmes are all on the Spock and Sherlock Holmes are all on the moon now but it's just like Mad Libs moon now but it's just like Mad Libs moon now but it's just like Mad Libs they're just ridiculous random things so they're just ridiculous random things so they're just ridiculous random things so I feel like we're ascribing creativity I feel like we're ascribing creativity I feel like we're ascribing creativity to the temperature slider bar when it's to the temperature slider bar when it's to the temperature slider bar when it's in reality not creativity it's simply in reality not creativity it's simply in reality not creativity it's simply Randomness yeah gosh I I have never Randomness yeah gosh I I have never Randomness yeah gosh I I have never thought about what I'm about to tell you thought about what I'm about to tell you thought about what I'm about to tell you in the context of large language models in the context of large language models in the context of large language models yet but when I was a first year PhD yet but when I was a first year PhD yet but when I was a first year PhD student I actually spent a year studying student I actually spent a year studying student I actually spent a year studying improv improv improv comedy little known fact about me it was comedy little known fact about me it was comedy little known fact about me it was because there was some funding and they because there was some funding and they because there was some funding and they were like that law school thing is weird were like that law school thing is weird were like that law school thing is weird but you used to be a cognitive but you used to be a cognitive but you used to be a cognitive psychologist why don't you come work on psychologist why don't you come work on psychologist why don't you come work on this project where we studied real this project where we studied real this project where we studied real professional improv troops in order to

professional improv troops in order to professional improv troops in order to figure out how they improvise to build figure out how they improvise to build figure out how they improvise to build AI you know I coded hours and hours of AI you know I coded hours and hours of AI you know I coded hours and hours of videos of them saying what they were videos of them saying what they were videos of them saying what they were thinking when they were doing something thinking when they were doing something thinking when they were doing something in an improv game and we ended up in an improv game and we ended up in an improv game and we ended up building these little like micro agents building these little like micro agents building these little like micro agents that did improvising and one of the that did improvising and one of the that did improvising and one of the things that was in the code was things that was in the code was things that was in the code was basically the temperature slider doing basically the temperature slider doing basically the temperature slider doing something that was slightly unexpected something that was slightly unexpected something that was slightly unexpected like if they were going to be a like if they were going to be a like if they were going to be a character you would take like The character you would take like The character you would take like The Stereotype of that character but then Stereotype of that character but then Stereotype of that character but then take one element on it and change that take one element on it and change that take one element on it and change that temperature slider and that was what temperature slider and that was what temperature slider and that was what made it interesting I never thought made it interesting I never thought made it interesting I never thought about that before but this was you know about that before but this was you know about that before but this was you know 2009 mhm and it's how we were thinking 2009 mhm and it's how we were thinking 2009 mhm and it's how we were thinking about that at the time but I think it's about that at the time but I think it's about that at the time but I think it's the same kind of thing now people say the same kind of thing now people say the same kind of thing now people say like oh that wasn't interesting enough like oh that wasn't interesting enough like oh that wasn't interesting enough do something that's a little wackier and do something that's a little wackier and do something that's a little wackier and suddenly you're like finishing each suddenly you're like finishing each suddenly you're like finishing each other's sandwiches instead of finishing other's sandwiches instead of finishing other's sandwiches instead of finishing each other's sentences which is another each other's sentences which is another each other's sentences which is another example that I use in my in my as well example that I use in my in my as well example that I use in my in my as well it's a great example because because it's a great example because because it's a great example because because there's no right answer to it's a there's no right answer to it's a there's no right answer to it's a beautiful day let's go to the because it beautiful day let's go to the because it beautiful day let's go to the because it doesn't have enough context and funny I doesn't have enough context and funny I doesn't have enough context and funny I did Comedy Sports and standup for a did Comedy Sports and standup for a did Comedy Sports and standup for a while and I'm a big yes and leader at my while and I'm a big yes and leader at my while and I'm a big yes and leader at my company I don't like stop energy and I'm company I don't like stop energy and I'm company I don't like stop energy and I'm also a big fan of a comedian named Mike also a big fan of a comedian named Mike also a big fan of a comedian named Mike Berbiglia who does a podcast dedicated Berbiglia who does a podcast dedicated Berbiglia who does a podcast dedicated to how comedy is developed it's called to how comedy is developed it's called to how comedy is developed it's called working it out and in that he talks working it out and in that he talks working it out and in that he talks about the very very subtle line you have about the very very subtle line you have about the very very subtle line you have to walk between absurdist and true like

to walk between absurdist and true like to walk between absurdist and true like comedy and like that temperature scale comedy and like that temperature scale comedy and like that temperature scale has got to be perfect otherwise you're has got to be perfect otherwise you're has got to be perfect otherwise you're was just weird like that was was just weird like that was was just weird like that was uncomfortable and not uncomfortable and not uncomfortable and not hilarious yeah totally trying to figure hilarious yeah totally trying to figure hilarious yeah totally trying to figure this stuff out though is you and I this stuff out though is you and I this stuff out though is you and I anthropomorphizing apply ascribing human anthropomorphizing apply ascribing human anthropomorphizing apply ascribing human behavior to a thing that isn't human but behavior to a thing that isn't human but behavior to a thing that isn't human but is still derived on a corpus of human is still derived on a corpus of human is still derived on a corpus of human stuff so it's like don't think it's a stuff so it's like don't think it's a stuff so it's like don't think it's a person don't think it's a person but person don't think it's a person but person don't think it's a person but it's doing everything it can to make me it's doing everything it can to make me it's doing everything it can to make me feel like it's a person which brings me feel like it's a person which brings me feel like it's a person which brings me back to my original question is that the back to my original question is that the back to my original question is that the wrong user interface does it have to wrong user interface does it have to wrong user interface does it have to talk to me like talk to me like talk to me like is a computer from Star Trek I mean I is a computer from Star Trek I mean I is a computer from Star Trek I mean I think it probably depends on the context think it probably depends on the context think it probably depends on the context right for how important it is that you right for how important it is that you right for how important it is that you constantly remember that this is not a constantly remember that this is not a constantly remember that this is not a person I mean I I feel like it's person I mean I I feel like it's person I mean I I feel like it's probably good for people to constantly probably good for people to constantly probably good for people to constantly remember that this is not a person in remember that this is not a person in remember that this is not a person in most contexts but I suppose one could most contexts but I suppose one could most contexts but I suppose one could make arguments about customer service make arguments about customer service make arguments about customer service and this kind of thing and then the and this kind of thing and then the and this kind of thing and then the question becomes is it okay if you know question becomes is it okay if you know question becomes is it okay if you know like as long as you're reminded that like as long as you're reminded that like as long as you're reminded that this is a chat box and the problem is this is a chat box and the problem is this is a chat box and the problem is with customer service chatbots people with customer service chatbots people with customer service chatbots people aren't being told and that actually aren't being told and that actually aren't being told and that actually really makes a lot of people upset really makes a lot of people upset really makes a lot of people upset people like to know when they're people like to know when they're people like to know when they're speaking to a human versus not I think speaking to a human versus not I think speaking to a human versus not I think and I think that's going to be a and I think that's going to be a and I think that's going to be a continuous problem and so I think that continuous problem and so I think that continuous problem and so I think that at least you should always know when at least you should always know when at least you should always know when you're talking to a chatbot versus not you're talking to a chatbot versus not you're talking to a chatbot versus not but if it's an issue of just like what

but if it's an issue of just like what but if it's an issue of just like what is the user interface should it feel is the user interface should it feel is the user interface should it feel more natural when you're talking to it I more natural when you're talking to it I more natural when you're talking to it I think it totally depends on the context think it totally depends on the context think it totally depends on the context yeah one other thing that I think is yeah one other thing that I think is yeah one other thing that I think is interesting for me to think about and interesting for me to think about and interesting for me to think about and I've been trying to ascribe that I think I've been trying to ascribe that I think I've been trying to ascribe that I think that something quietly happened between that something quietly happened between that something quietly happened between Alexa Siri and Cortana and the modern Alexa Siri and Cortana and the modern Alexa Siri and Cortana and the modern chatbots is that they appear to have chatbots is that they appear to have chatbots is that they appear to have removed gender we spent a lot of time removed gender we spent a lot of time removed gender we spent a lot of time ordering disembodied women's names ordering disembodied women's names ordering disembodied women's names around you know saying Alexa turn off around you know saying Alexa turn off around you know saying Alexa turn off the lights and stuff like that and uh the lights and stuff like that and uh the lights and stuff like that and uh you know at least you know female coded you know at least you know female coded you know at least you know female coded names and I would always encourage my my names and I would always encourage my my names and I would always encourage my my family and my kids to like say please family and my kids to like say please family and my kids to like say please and thank C but even now I can only and thank C but even now I can only and thank C but even now I can only imagine what all the actual women named imagine what all the actual women named imagine what all the actual women named Alexa they must be just really hating Alexa they must be just really hating Alexa they must be just really hating life right now oh there's some kind of life right now oh there's some kind of life right now oh there's some kind of movement because someone a like movement because someone a like movement because someone a like organization of alexas kept tweeting at organization of alexas kept tweeting at organization of alexas kept tweeting at me like asking me to talk about the like me like asking me to talk about the like me like asking me to talk about the like there's unionized all the there's yeah there's unionized all the there's yeah there's unionized all the there's yeah there's there's some kind of organized there's there's some kind of organized there's there's some kind of organized movement around people I mean one one of movement around people I mean one one of movement around people I mean one one of my best friends is named Alexa and um my best friends is named Alexa and um my best friends is named Alexa and um I'm actually really surprised mine did I'm actually really surprised mine did I'm actually really surprised mine did not go off which just I actually muted not go off which just I actually muted not go off which just I actually muted mine before I said mine before I said mine before I said it but yeah I mean and that I actually it but yeah I mean and that I actually it but yeah I mean and that I actually think that is a very interesting think that is a very interesting think that is a very interesting unintended consequence for you know all unintended consequence for you know all unintended consequence for you know all of these poor people with with that name of these poor people with with that name of these poor people with with that name it is interesting that the gender thing it is interesting that the gender thing it is interesting that the gender thing sort of went but I think that's because sort of went but I think that's because sort of went but I think that's because Chachi BT wasn't like voice first and my

Chachi BT wasn't like voice first and my Chachi BT wasn't like voice first and my understanding is when you set up voice understanding is when you set up voice understanding is when you set up voice on because you can you know talk to chat on because you can you know talk to chat on because you can you know talk to chat gbt I think the first thing it does is gbt I think the first thing it does is gbt I think the first thing it does is ask you to pick the voice so there isn't ask you to pick the voice so there isn't ask you to pick the voice so there isn't a default and I bet you that the voice a default and I bet you that the voice a default and I bet you that the voice this is me guessing but I think it's a this is me guessing but I think it's a this is me guessing but I think it's a reasonable guess that it's not using reasonable guess that it's not using reasonable guess that it's not using that voice selection on the output as an that voice selection on the output as an that voice selection on the output as an input into how it behaves so there's input into how it behaves so there's input into how it behaves so there's it's not ascribing gender to it it's own it's not ascribing gender to it it's own it's not ascribing gender to it it's own context it's own prologue have you seen context it's own prologue have you seen context it's own prologue have you seen the person on Tik Tok who is making the person on Tik Tok who is making the person on Tik Tok who is making videos of her talking to Dan chat gbt videos of her talking to Dan chat gbt videos of her talking to Dan chat gbt have you seen this I no I haven't what have you seen this I no I haven't what have you seen this I no I haven't what are they what are they trying to do are they what are they trying to do are they what are they trying to do she's just having this chat with you she's just having this chat with you she's just having this chat with you know a version of of chat GPT that's know a version of of chat GPT that's know a version of of chat GPT that's like pretty snarky and inappropriate and like pretty snarky and inappropriate and like pretty snarky and inappropriate and I mean they have they have millions of I mean they have they have millions of I mean they have they have millions of of views and I guess people are trying of views and I guess people are trying of views and I guess people are trying to make versions of CHA gbt as like a to make versions of CHA gbt as like a to make versions of CHA gbt as like a have a discussion with a romantic have a discussion with a romantic have a discussion with a romantic partner kind of thing but one of the partner kind of thing but one of the partner kind of thing but one of the things I thought was interesting about things I thought was interesting about things I thought was interesting about it is that the innerface for talking to it is that the innerface for talking to it is that the innerface for talking to chat TBT I keep meaning to go back and chat TBT I keep meaning to go back and chat TBT I keep meaning to go back and look but I think it actually looks look but I think it actually looks look but I think it actually looks exactly like in that episode of Black exactly like in that episode of Black exactly like in that episode of Black Mirror where the woman brings back her Mirror where the woman brings back her Mirror where the woman brings back her dead spouse as a robot eventually but dead spouse as a robot eventually but dead spouse as a robot eventually but there's the there's the interim phase there's the there's the interim phase there's the there's the interim phase where she's talking to it it and the where she's talking to it it and the where she's talking to it it and the interface to me looks exactly the same interface to me looks exactly the same interface to me looks exactly the same and I thought that was funny because um and I thought that was funny because um and I thought that was funny because um I made a Tik Tok recently about voice I made a Tik Tok recently about voice I made a Tik Tok recently about voice cloning and people have been fighting in cloning and people have been fighting in cloning and people have been fighting in the comments about like the idea of the comments about like the idea of the comments about like the idea of bringing someone back to the dead as a bringing someone back to the dead as a bringing someone back to the dead as a chatbot is okay right yeah see that's

chatbot is okay right yeah see that's chatbot is okay right yeah see that's really interesting I've got 900 shows really interesting I've got 900 shows really interesting I've got 900 shows I've got over 500 hours of content the I've got over 500 hours of content the I've got over 500 hours of content the Corpus is there I don't know if my sons Corpus is there I don't know if my sons Corpus is there I don't know if my sons would want to have a Hanselman chatbot would want to have a Hanselman chatbot would want to have a Hanselman chatbot they just like let some things die one they just like let some things die one they just like let some things die one of my colleagues Jed breaker this is one of my colleagues Jed breaker this is one of my colleagues Jed breaker this is one of his major areas of research he does of his major areas of research he does of his major areas of research he does work on postmortem data and this kind of work on postmortem data and this kind of work on postmortem data and this kind of thing simple versions of this has been thing simple versions of this has been thing simple versions of this has been have been around for a long time there have been around for a long time there have been around for a long time there actually was a startup for a while where actually was a startup for a while where actually was a startup for a while where the tagline was when your heart stops the tagline was when your heart stops the tagline was when your heart stops beating you keep tweeting oh no see beating you keep tweeting oh no see beating you keep tweeting oh no see that's the other thing I've got I've got that's the other thing I've got I've got that's the other thing I've got I've got like what do I have now I've been on like what do I have now I've been on like what do I have now I've been on Twitter for like 17 years like there's Twitter for like 17 years like there's Twitter for like 17 years like there's more than enough for it to like you know more than enough for it to like you know more than enough for it to like you know be crap posting for the rest of my life be crap posting for the rest of my life be crap posting for the rest of my life about Beyonce and Neil Patrick Harris about Beyonce and Neil Patrick Harris about Beyonce and Neil Patrick Harris yeah no simple and I mean and I when yeah no simple and I mean and I when yeah no simple and I mean and I when that episode of black Mir came out and I that episode of black Mir came out and I that episode of black Mir came out and I was talking to my students about it I was talking to my students about it I was talking to my students about it I was like we're here like that middle was like we're here like that middle was like we're here like that middle technology like we're basically there technology like we're basically there technology like we're basically there and we're definitely there now like with and we're definitely there now like with and we're definitely there now like with voice cloning and what you just voice cloning and what you just voice cloning and what you just described like the fact that I haven't described like the fact that I haven't described like the fact that I haven't already heard about a startup that is already heard about a startup that is already heard about a startup that is doing exactly that right now surprises doing exactly that right now surprises doing exactly that right now surprises me a little bit because I'm sure it's in me a little bit because I'm sure it's in me a little bit because I'm sure it's in the works okay so here's an Ethics the works okay so here's an Ethics the works okay so here's an Ethics question you probably remember many question you probably remember many question you probably remember many years ago I think my guess is you're years ago I think my guess is you're years ago I think my guess is you're maybe a little younger but when we maybe a little younger but when we maybe a little younger but when we started cloning when they cloned Dolly started cloning when they cloned Dolly started cloning when they cloned Dolly the sheep it was the biggest news on on the sheep it was the biggest news on on the sheep it was the biggest news on on the world the whole you know was like the world the whole you know was like the world the whole you know was like number one news story for weeks but I number one news story for weeks but I number one news story for weeks but I don't know how this organized but the don't know how this organized but the don't know how this organized but the world self-organized and said we're not world self-organized and said we're not world self-organized and said we're not going to clone people so a technology

going to clone people so a technology going to clone people so a technology appears a thing happens and then the appears a thing happens and then the appears a thing happens and then the world decides to do it or not do it but world decides to do it or not do it but world decides to do it or not do it but then Dolly the sheep being cloned is then Dolly the sheep being cloned is then Dolly the sheep being cloned is pre- Big Tech like Tech was medium Tech pre- Big Tech like Tech was medium Tech pre- Big Tech like Tech was medium Tech at that point so now we have big Tech at that point so now we have big Tech at that point so now we have big Tech big capitalism I feel like if they clone big capitalism I feel like if they clone big capitalism I feel like if they clone Dolly you know a couple weeks later we'd Dolly you know a couple weeks later we'd Dolly you know a couple weeks later we'd have Jurassic Park like you know what I have Jurassic Park like you know what I have Jurassic Park like you know what I mean like things are too big now do you mean like things are too big now do you mean like things are too big now do you think the same thing is happening in think the same thing is happening in think the same thing is happening in large language models that they're going large language models that they're going large language models that they're going to make AGI no matter what ethics be to make AGI no matter what ethics be to make AGI no matter what ethics be damned that is a really good question I damned that is a really good question I damned that is a really good question I mean I have to say I have been mean I have to say I have been mean I have to say I have been disheartened by the speed at which disheartened by the speed at which disheartened by the speed at which things have been happening and I think things have been happening and I think things have been happening and I think that speed is because of Market that speed is because of Market that speed is because of Market competition issues so for example a competition issues so for example a competition issues so for example a concept that I've been thinking about concept that I've been thinking about concept that I've been thinking about for a while that kind of came out of U for a while that kind of came out of U for a while that kind of came out of U my lab myself and a former graduate my lab myself and a former graduate my lab myself and a former graduate student wrote an opet about this like student wrote an opet about this like student wrote an opet about this like four years ago uh ethical debt as an four years ago uh ethical debt as an four years ago uh ethical debt as an analog to technical debt and this comes analog to technical debt and this comes analog to technical debt and this comes from moving too fast like not thinking from moving too fast like not thinking from moving too fast like not thinking about things ahead of time and and like about things ahead of time and and like about things ahead of time and and like technical debt is I mean technical debt technical debt is I mean technical debt technical debt is I mean technical debt wasn't even meant to be a negative thing wasn't even meant to be a negative thing wasn't even meant to be a negative thing like taking on debt can be good as long like taking on debt can be good as long like taking on debt can be good as long as you know that you're taking it on and as you know that you're taking it on and as you know that you're taking it on and so when it's technical debt like you so when it's technical debt like you so when it's technical debt like you sort of know but the problem with sort of know but the problem with sort of know but the problem with ethical debt is like you might have ethical debt is like you might have ethical debt is like you might have absolutely no idea what these absolutely no idea what these absolutely no idea what these consequences could be if you don't stop consequences could be if you don't stop consequences could be if you don't stop to think about it at all and I feel like to think about it at all and I feel like to think about it at all and I feel like especially in the early days of chat gbt especially in the early days of chat gbt especially in the early days of chat gbt there was just this sense of like from there was just this sense of like from there was just this sense of like from you know other big tech companies like

you know other big tech companies like you know other big tech companies like we have to pick up the pace we have to pick up the pace we have to pick up the pace um and I saw you know quotes from people um and I saw you know quotes from people um and I saw you know quotes from people in the news like internal documents that in the news like internal documents that in the news like internal documents that were like we need to pick up the pace we were like we need to pick up the pace we were like we need to pick up the pace we can think about the consequences later can think about the consequences later can think about the consequences later and you know Market forces capitalism and you know Market forces capitalism and you know Market forces capitalism not not a fan of Market forces I saw a not not a fan of Market forces I saw a not not a fan of Market forces I saw a really great tweet from uh Emily Freeman really great tweet from uh Emily Freeman really great tweet from uh Emily Freeman who is a best-selling devops author and who is a best-selling devops author and who is a best-selling devops author and a community leader and she said stop a community leader and she said stop a community leader and she said stop shipping gen chatbots were being flooded shipping gen chatbots were being flooded shipping gen chatbots were being flooded with identical tools get creative so with identical tools get creative so with identical tools get creative so many other potential uses you know it's many other potential uses you know it's many other potential uses you know it's funny I would think about it in the funny I would think about it in the funny I would think about it in the context of like early web design we context of like early web design we context of like early web design we started making text boxes and I remember started making text boxes and I remember started making text boxes and I remember when text boxes was just like text box when text boxes was just like text box when text boxes was just like text box and a button and then we'd have these and a button and then we'd have these and a button and then we'd have these validators remember where it's like oh validators remember where it's like oh validators remember where it's like oh you have to validate that this is really you have to validate that this is really you have to validate that this is really a social security number you have to a social security number you have to a social security number you have to validate that this is really a phone validate that this is really a phone validate that this is really a phone number we we did all this work because number we we did all this work because number we we did all this work because we were told in the early days of the we were told in the early days of the we were told in the early days of the web don't trust user web don't trust user web don't trust user input and here we are again with a giant input and here we are again with a giant input and here we are again with a giant text box and a button and we're not text box and a button and we're not text box and a button and we're not supposed to trust user input and the supposed to trust user input and the supposed to trust user input and the user input could make my my coffee user input could make my my coffee user input could make my my coffee ordering chatbot like write a python ordering chatbot like write a python ordering chatbot like write a python script to take over the world and that's script to take over the world and that's script to take over the world and that's that's clearly ethical debt that no one that's clearly ethical debt that no one that's clearly ethical debt that no one is thinking about yeah things keep is thinking about yeah things keep is thinking about yeah things keep happening that like the amount of how happening that like the amount of how happening that like the amount of how did you not know this could happen just did you not know this could happen just did you not know this could happen just like I mean even the recent one with like I mean even the recent one with like I mean even the recent one with like the the New York City chatbot where like the the New York City chatbot where like the the New York City chatbot where they have a chatbot like answering they have a chatbot like answering they have a chatbot like answering questions for small businesses and some questions for small businesses and some questions for small businesses and some journalists from the markup tested it journalists from the markup tested it journalists from the markup tested it and it turns out it's telling people to

and it turns out it's telling people to and it turns out it's telling people to do Elite things M which of course it do Elite things M which of course it do Elite things M which of course it is or like you know there was the eating is or like you know there was the eating is or like you know there was the eating disorder help line their workers tried disorder help line their workers tried disorder help line their workers tried to unionize and so they had chatbot to unionize and so they had chatbot to unionize and so they had chatbot answering people instead and within like answering people instead and within like answering people instead and within like five days it was gone yeah and then was five days it was gone yeah and then was five days it was gone yeah and then was that Air Canada chat Bots can be that Air Canada chat Bots can be that Air Canada chat Bots can be convinced to give people upgrades yep convinced to give people upgrades yep convinced to give people upgrades yep and the car dealership like selling and the car dealership like selling and the car dealership like selling someone a car for a dollar because they someone a car for a dollar because they someone a car for a dollar because they just like integrated chat gbt into their just like integrated chat gbt into their just like integrated chat gbt into their site without site without site without yeah I mean some of these things happen yeah I mean some of these things happen yeah I mean some of these things happen because whatever it person for the car because whatever it person for the car because whatever it person for the car dealership decided to do that like you dealership decided to do that like you dealership decided to do that like you know may not have known a lot about chat know may not have known a lot about chat know may not have known a lot about chat Bots um but then there's other examples Bots um but then there's other examples Bots um but then there's other examples where you should have known that this where you should have known that this where you should have known that this kind of thing could happen and it's hard kind of thing could happen and it's hard kind of thing could happen and it's hard to tell whether it really is not to tell whether it really is not to tell whether it really is not thinking about it at all or how thinking about it at all or how thinking about it at all or how difficult it is for people to imagine difficult it is for people to imagine difficult it is for people to imagine these things which sometimes surprises these things which sometimes surprises these things which sometimes surprises me a little bit I mean I guess I guess me a little bit I mean I guess I guess me a little bit I mean I guess I guess this is my I mean I guess I do this all this is my I mean I guess I do this all this is my I mean I guess I do this all day right like there's there's this day right like there's there's this day right like there's there's this really great tweet that I'm reminded of really great tweet that I'm reminded of really great tweet that I'm reminded of it was a political tweet from it was a political tweet from it was a political tweet from 2015 but I think it has AI implications 2015 but I think it has AI implications 2015 but I think it has AI implications now and the Tweet was I never thought now and the Tweet was I never thought now and the Tweet was I never thought the leopards would eat my face sobs the the leopards would eat my face sobs the the leopards would eat my face sobs the woman who voted for the leopards eating woman who voted for the leopards eating woman who voted for the leopards eating people's faces party and now like people's faces party and now like people's faces party and now like leopards ate my face is like become like leopards ate my face is like become like leopards ate my face is like become like the Clarion call of like I didn't think the Clarion call of like I didn't think the Clarion call of like I didn't think it would happen to me and now here we it would happen to me and now here we it would happen to me and now here we are you know maybe getting ethical debt

are you know maybe getting ethical debt are you know maybe getting ethical debt around AI I highly recommend the around AI I highly recommend the around AI I highly recommend the leopards ate my face leopards ate my face leopards ate my face subreddit good for many examples I mean subreddit good for many examples I mean subreddit good for many examples I mean you know if you want to like weep you know if you want to like weep you know if you want to like weep uh trying to think about these things uh trying to think about these things uh trying to think about these things though when you are doing this kind of though when you are doing this kind of though when you are doing this kind of public scholarship like we're here public scholarship like we're here public scholarship like we're here complaining but is there a solution is complaining but is there a solution is complaining but is there a solution is there an answer and is the answer just there an answer and is the answer just there an answer and is the answer just slow down because I'm doing talks I'm slow down because I'm doing talks I'm slow down because I'm doing talks I'm doing public education on AI you're doing public education on AI you're doing public education on AI you're doing a public education on AI but doing a public education on AI but doing a public education on AI but simply standing in front of like the simply standing in front of like the simply standing in front of like the Avalanche you know the classic Babylon 5 Avalanche you know the classic Babylon 5 Avalanche you know the classic Babylon 5 quote the Avalanche has begun it's too quote the Avalanche has begun it's too quote the Avalanche has begun it's too late for the Pebbles to vote uh you and late for the Pebbles to vote uh you and late for the Pebbles to vote uh you and I are the pebbles in this context and I are the pebbles in this context and I are the pebbles in this context and I'm out there selling AI because I'm I'm out there selling AI because I'm I'm out there selling AI because I'm excited about aspects of it but I'm not excited about aspects of it but I'm not excited about aspects of it but I'm not excited about other aspects of it so I'm excited about other aspects of it so I'm excited about other aspects of it so I'm also complicit I mean I guess it depends also complicit I mean I guess it depends also complicit I mean I guess it depends on who the audience is right so for me on who the audience is right so for me on who the audience is right so for me I'm speaking to the general public who I'm speaking to the general public who I'm speaking to the general public who may not understand enough about AI to be may not understand enough about AI to be may not understand enough about AI to be critical of it or to know how it's critical of it or to know how it's critical of it or to know how it's appropriate to use it and how it's not appropriate to use it and how it's not appropriate to use it and how it's not that's what I'm trying to do I don't that's what I'm trying to do I don't that's what I'm trying to do I don't have any delusions that my Tik toks are have any delusions that my Tik toks are have any delusions that my Tik toks are changing any you know anyone at open changing any you know anyone at open changing any you know anyone at open ai's mind about ai's mind about ai's mind about anything you don't think you've saved anything you don't think you've saved anything you don't think you've saved anyone with your Tik Tok I don't think anyone with your Tik Tok I don't think anyone with your Tik Tok I don't think I've saved anyone with my T I mean I'm I've saved anyone with my T I mean I'm I've saved anyone with my T I mean I'm in an interesting position because I am in an interesting position because I am in an interesting position because I am a researcher I have written many papers a researcher I have written many papers a researcher I have written many papers about technology ethics and a a little about technology ethics and a a little about technology ethics and a a little bit about Ai and ethics so that's not bit about Ai and ethics so that's not bit about Ai and ethics so that's not really like my main thing more I'm more really like my main thing more I'm more really like my main thing more I'm more a communicator around that and I'm

a communicator around that and I'm a communicator around that and I'm trying to help people understand these trying to help people understand these trying to help people understand these things so they can be more critical at things so they can be more critical at things so they can be more critical at the very beginning of Chad gbt when I the very beginning of Chad gbt when I the very beginning of Chad gbt when I was first thinking about it I went back was first thinking about it I went back was first thinking about it I went back and reread the paper that Joseph weisen and reread the paper that Joseph weisen and reread the paper that Joseph weisen bom published about Eliza from like bom published about Eliza from like bom published about Eliza from like 1966 or whatever and a really 1966 or whatever and a really 1966 or whatever and a really interesting thing from that paper is I interesting thing from that paper is I interesting thing from that paper is I mean the whole thing like Eliza the mean the whole thing like Eliza the mean the whole thing like Eliza the chatbot you know for folks who might not chatbot you know for folks who might not chatbot you know for folks who might not be as familiar with that the be as familiar with that the be as familiar with that the psychoanalyst turns out it's very easily psychoanalyst turns out it's very easily psychoanalyst turns out it's very easily to create a procedural like mad Li style to create a procedural like mad Li style to create a procedural like mad Li style psychoanalyst you know uh I'm sad about psychoanalyst you know uh I'm sad about psychoanalyst you know uh I'm sad about my mother today oh tell me about your my mother today oh tell me about your my mother today oh tell me about your mother but he seemed sort of surprised mother but he seemed sort of surprised mother but he seemed sort of surprised and Disturbed when his secretary like and Disturbed when his secretary like and Disturbed when his secretary like even knowing that it wasn't real was even knowing that it wasn't real was even knowing that it wasn't real was very like sort of taken in by it and he very like sort of taken in by it and he very like sort of taken in by it and he says in this paper technology can seem says in this paper technology can seem says in this paper technology can seem like magic it can fool even the most like magic it can fool even the most like magic it can fool even the most experienced Observer but as as soon as experienced Observer but as as soon as experienced Observer but as as soon as you can explain it in ways that someone you can explain it in ways that someone you can explain it in ways that someone can properly comprehend it then that can properly comprehend it then that can properly comprehend it then that magic crumbles away um and I think magic crumbles away um and I think magic crumbles away um and I think that's totally true for a lot of kinds that's totally true for a lot of kinds that's totally true for a lot of kinds of technology and the problem is it is of technology and the problem is it is of technology and the problem is it is very hard to make the magic of very hard to make the magic of very hard to make the magic of generative AI crumble away because even generative AI crumble away because even generative AI crumble away because even when we tell people like oh it's when we tell people like oh it's when we tell people like oh it's basically autocomplete that's not really basically autocomplete that's not really basically autocomplete that's not really sufficient understanding for what it is sufficient understanding for what it is sufficient understanding for what it is and it's very very hard to describe and it's very very hard to describe and it's very very hard to describe especially in a short amount of time

especially in a short amount of time especially in a short amount of time what is actually happening here when it what is actually happening here when it what is actually happening here when it seems so magical yeah the the elevator seems so magical yeah the the elevator seems so magical yeah the the elevator speech to explain it to non-technical speech to explain it to non-technical speech to explain it to non-technical parent is the elevator ride is too short parent is the elevator ride is too short parent is the elevator ride is too short to explain it so when non-technical to explain it so when non-technical to explain it so when non-technical parent calls and is like do I want this parent calls and is like do I want this parent calls and is like do I want this is the AI it's like well it kind of is the AI it's like well it kind of is the AI it's like well it kind of depends watch this 20 minute Tech talk depends watch this 20 minute Tech talk depends watch this 20 minute Tech talk or this 90-minute leure and then come or this 90-minute leure and then come or this 90-minute leure and then come back with questions and that's the wide back with questions and that's the wide back with questions and that's the wide reaching ramif ation of all of this reaching ramif ation of all of this reaching ramif ation of all of this moving so fast and then applied to all moving so fast and then applied to all moving so fast and then applied to all of the other pressures postco and of the other pressures postco and of the other pressures postco and politics is adding up to a lot of stress politics is adding up to a lot of stress politics is adding up to a lot of stress and some people think that this is the and some people think that this is the and some people think that this is the way out of our problems and also very way out of our problems and also very way out of our problems and also very large language models being conflated as large language models being conflated as large language models being conflated as artificial general intelligence is artificial general intelligence is artificial general intelligence is another huge problem yeah totally that another huge problem yeah totally that another huge problem yeah totally that and like there is a a particular genre and like there is a a particular genre and like there is a a particular genre of pedantic uh social media comment that of pedantic uh social media comment that of pedantic uh social media comment that I've been getting lately which is um I've been getting lately which is um I've been getting lately which is um this is not AI stop using the word AI do this is not AI stop using the word AI do this is not AI stop using the word AI do you get that or is it just me I have not you get that or is it just me I have not you get that or is it just me I have not gotten that although I have said that I gotten that although I have said that I gotten that although I have said that I feel like AI has an interesting branding feel like AI has an interesting branding feel like AI has an interesting branding situation where it's fun to say Ai and situation where it's fun to say Ai and situation where it's fun to say Ai and chat GPT has good mouth feel but ML and chat GPT has good mouth feel but ML and chat GPT has good mouth feel but ML and data science doesn't well that is very data science doesn't well that is very data science doesn't well that is very true there's an interesting Spectrum true there's an interesting Spectrum true there's an interesting Spectrum here there's the people who think that here there's the people who think that here there's the people who think that we are like 5 days away from G or sen we are like 5 days away from G or sen we are like 5 days away from G or sen five minutes away five minutes away from five minutes away five minutes away from five minutes away five minutes away from yeah there are tons of people who are yeah there are tons of people who are yeah there are tons of people who are like we're practically there right so like we're practically there right so like we're practically there right so there's like those folks and then there's like those folks and then there's like those folks and then there's the folks who are like this is there's the folks who are like this is there's the folks who are like this is all literally just linear regression I

all literally just linear regression I all literally just linear regression I don't see why anyone's getting so worked don't see why anyone's getting so worked don't see why anyone's getting so worked out about it well then you have to go out about it well then you have to go out about it well then you have to go and explain the regression to random and explain the regression to random and explain the regression to random people like it just it gets complicated people like it just it gets complicated people like it just it gets complicated fast and it's hard for me to want to I fast and it's hard for me to want to I fast and it's hard for me to want to I want to explain this stuff to folks in a want to explain this stuff to folks in a want to explain this stuff to folks in a way that is inclusive like as a as an way that is inclusive like as a as an way that is inclusive like as a as an educator I think we all want people to educator I think we all want people to educator I think we all want people to like feel powerful and not afraid and like feel powerful and not afraid and like feel powerful and not afraid and that's a reasonable thing to want that's a reasonable thing to want that's a reasonable thing to want totally um I mean and one of the totally um I mean and one of the totally um I mean and one of the problems with the amount of hype around problems with the amount of hype around problems with the amount of hype around gen AI is like the hype is either oh my gen AI is like the hype is either oh my gen AI is like the hype is either oh my God this is the most amazing thing ever God this is the most amazing thing ever God this is the most amazing thing ever this is going to change everything this is going to change everything this is going to change everything tomorrow nothing's going to be the same tomorrow nothing's going to be the same tomorrow nothing's going to be the same like this is this is the most amazing like this is this is the most amazing like this is this is the most amazing Tech and then there's the the world's Tech and then there's the the world's Tech and then there's the the world's going to be over next year it's going to going to be over next year it's going to going to be over next year it's going to destroy Humanity like right away and destroy Humanity like right away and destroy Humanity like right away and it's sometimes it's the same people so it's sometimes it's the same people so it's sometimes it's the same people so either way the hype is like making the either way the hype is like making the either way the hype is like making the technology seem much more powerful than technology seem much more powerful than technology seem much more powerful than I think it actually is especially right I think it actually is especially right I think it actually is especially right now well yeah plus the Mechanical Turk now well yeah plus the Mechanical Turk now well yeah plus the Mechanical Turk problem I was recently giving a talk to problem I was recently giving a talk to problem I was recently giving a talk to some people in turkey and use the term some people in turkey and use the term some people in turkey and use the term Mechanical Turk and then explained it to Mechanical Turk and then explained it to Mechanical Turk and then explained it to them that the idea that a lot of the AI them that the idea that a lot of the AI them that the idea that a lot of the AI that you're seeing is not necessarily AI that you're seeing is not necessarily AI that you're seeing is not necessarily AI it may be ghost workers right it's funny it may be ghost workers right it's funny it may be ghost workers right it's funny too because that is also an issue of of too because that is also an issue of of too because that is also an issue of of not understanding how these systems work not understanding how these systems work not understanding how these systems work okay so even for a chatbot like I I you okay so even for a chatbot like I I you okay so even for a chatbot like I I you know I explained something about ghost know I explained something about ghost know I explained something about ghost work recently and someone said are you work recently and someone said are you work recently and someone said are you saying that like there's some dude saying that like there's some dude saying that like there's some dude that's like very quickly typing the

that's like very quickly typing the that's like very quickly typing the answer when I asked Chad gbt something I answer when I asked Chad gbt something I answer when I asked Chad gbt something I was like no no that's not what I'm was like no no that's not what I'm was like no no that's not what I'm saying however it is still very common saying however it is still very common saying however it is still very common for people to forget the huge amounts of for people to forget the huge amounts of for people to forget the huge amounts of human input that went into creating this human input that went into creating this human input that went into creating this system in terms of data labeling and system in terms of data labeling and system in terms of data labeling and supervised learning and this kind of supervised learning and this kind of supervised learning and this kind of thing and when we forget the humans that thing and when we forget the humans that thing and when we forget the humans that are part of the system it a seems more are part of the system it a seems more are part of the system it a seems more magic um and B it can obscure sources of magic um and B it can obscure sources of magic um and B it can obscure sources of bias and that sort of thing I mean and bias and that sort of thing I mean and bias and that sort of thing I mean and this is a this is a huge issue like all this is a this is a huge issue like all this is a this is a huge issue like all of this only works because of data of this only works because of data of this only works because of data labeling and whether that labeled data labeling and whether that labeled data labeling and whether that labeled data is coming from captioned images online is coming from captioned images online is coming from captioned images online or whether it's humans that are looking or whether it's humans that are looking or whether it's humans that are looking at things and labeling them or whether at things and labeling them or whether at things and labeling them or whether it's you know Amazon Mechanical trikers it's you know Amazon Mechanical trikers it's you know Amazon Mechanical trikers who also are are humans sorry that was a who also are are humans sorry that was a who also are are humans sorry that was a weird way to put that or unpaid weird way to put that or unpaid weird way to put that or unpaid YouTubers as we just learned who had YouTubers as we just learned who had YouTubers as we just learned who had their transcripts including my own and their transcripts including my own and their transcripts including my own and my blog and you know actually do you my blog and you know actually do you my blog and you know actually do you know that someone actually accused this know that someone actually accused this know that someone actually accused this podcast of being AI generated because podcast of being AI generated because podcast of being AI generated because they couldn't conceive of 900 episodes they couldn't conceive of 900 episodes they couldn't conceive of 900 episodes over 18 years being actual oh that's over 18 years being actual oh that's over 18 years being actual oh that's kind of funny this is not a this is a kind of funny this is not a this is a kind of funny this is not a this is a real thing that that's funny they tried real thing that that's funny they tried real thing that that's funny they tried to get it delisted interestingly you to get it delisted interestingly you to get it delisted interestingly you know a a lot of my research um over the know a a lot of my research um over the know a a lot of my research um over the past eight years has been around past eight years has been around past eight years has been around research ethics for public data which research ethics for public data which research ethics for public data which you know until recently was largely you know until recently was largely you know until recently was largely things like um machine learning things like um machine learning things like um machine learning researchers you know sucking up researchers you know sucking up researchers you know sucking up everything from a subreddit to build a everything from a subreddit to build a everything from a subreddit to build a classifier about whether someone's classifier about whether someone's classifier about whether someone's depressed I you know like that sort of

depressed I you know like that sort of depressed I you know like that sort of thing or social scientist studying thing or social scientist studying thing or social scientist studying people's tweets and a lot of that work people's tweets and a lot of that work people's tweets and a lot of that work has been around how people react to this has been around how people react to this has been around how people react to this and you know a lot of times the reaction and you know a lot of times the reaction and you know a lot of times the reaction is why would scientists be reading my is why would scientists be reading my is why would scientists be reading my tweets that's weird um but I feel like tweets that's weird um but I feel like tweets that's weird um but I feel like I've seen a lot of the same kinds of I've seen a lot of the same kinds of I've seen a lot of the same kinds of patterns in people's reactions to patterns in people's reactions to patterns in people's reactions to thinking about their inclusion in thinking about their inclusion in thinking about their inclusion in training data and that it feels training data and that it feels training data and that it feels distractive it does I've stopped distractive it does I've stopped distractive it does I've stopped blogging because I'm tired but also blogging because I'm tired but also blogging because I'm tired but also because I'm not like I don't know if I because I'm not like I don't know if I because I'm not like I don't know if I want to feed the giant MW that is AI but want to feed the giant MW that is AI but want to feed the giant MW that is AI but that means then how is it ever going to that means then how is it ever going to that means then how is it ever going to get smarter if the Corpus doesn't get get smarter if the Corpus doesn't get get smarter if the Corpus doesn't get smarter and we've already seen in the smarter and we've already seen in the smarter and we've already seen in the last year and a half Google's search last year and a half Google's search last year and a half Google's search results have turned into garbage because results have turned into garbage because results have turned into garbage because everything is AI generated everything is AI generated everything is AI generated spogs yep so I think there's this like spogs yep so I think there's this like spogs yep so I think there's this like sort of fundamental question here which sort of fundamental question here which sort of fundamental question here which is you know a lot of people will tell is you know a lot of people will tell is you know a lot of people will tell you that the only solution is for this you that the only solution is for this you that the only solution is for this technology to not exist like that the technology to not exist like that the technology to not exist like that the technology itself is fundamentally technology itself is fundamentally technology itself is fundamentally unethical because the only way that it unethical because the only way that it unethical because the only way that it exists is it was created you know on the exists is it was created you know on the exists is it was created you know on the backs of people without their consent I backs of people without their consent I backs of people without their consent I me you hear this a lot from artists and me you hear this a lot from artists and me you hear this a lot from artists and I you know I think I and I think that is I you know I think I and I think that is I you know I think I and I think that is a very valid argument now do I think a very valid argument now do I think a very valid argument now do I think there is any there is any there is any possibility is out of the bottle like possibility is out of the bottle like possibility is out of the bottle like it's hard to imagine this technology not it's hard to imagine this technology not it's hard to imagine this technology not being used at all for this reason you being used at all for this reason you being used at all for this reason you know except for the extent that know except for the extent that know except for the extent that copyright things start to happen I copyright things start to happen I copyright things start to happen I personally think this is unlikely but if

personally think this is unlikely but if personally think this is unlikely but if copyright cases all went like wildly not copyright cases all went like wildly not copyright cases all went like wildly not on the side of the tech companies and on the side of the tech companies and on the side of the tech companies and all training data is copyright all training data is copyright all training data is copyright infringement and shut it all down what's infringement and shut it all down what's infringement and shut it all down what's probably going to happen is that we will probably going to happen is that we will probably going to happen is that we will get a bunch of AI models that are not as get a bunch of AI models that are not as get a bunch of AI models that are not as good yeah or a bunch of Open Source ones good yeah or a bunch of Open Source ones good yeah or a bunch of Open Source ones that are put on Torrance that are that are put on Torrance that are that are put on Torrance that are trained on regular data anyway and can't trained on regular data anyway and can't trained on regular data anyway and can't be stuff and I think people will be stuff and I think people will be stuff and I think people will reasonably disagree about whether that reasonably disagree about whether that reasonably disagree about whether that is a better solution or not right like is a better solution or not right like is a better solution or not right like even folks in the like AI ethics even folks in the like AI ethics even folks in the like AI ethics Community disagree vastly about many of Community disagree vastly about many of Community disagree vastly about many of these things I don't think that's these things I don't think that's these things I don't think that's surprising and I think there are a lot surprising and I think there are a lot surprising and I think there are a lot of reasonable arguments on different of reasonable arguments on different of reasonable arguments on different sides of all of this I mostly just want sides of all of this I mostly just want sides of all of this I mostly just want us to get to a place where if this us to get to a place where if this us to get to a place where if this technology is going to exist which it technology is going to exist which it technology is going to exist which it almost certainly is at this point what almost certainly is at this point what almost certainly is at this point what are the ways that it can be developed are the ways that it can be developed are the ways that it can be developed and deployed as responsibly as possible and deployed as responsibly as possible and deployed as responsibly as possible and one of the problems is that you and one of the problems is that you and one of the problems is that you can't get to that place with the speed can't get to that place with the speed can't get to that place with the speed at which things are moving and without at which things are moving and without at which things are moving and without having some kind of agreement on having some kind of agreement on having some kind of agreement on priorities and that's why people are are priorities and that's why people are are priorities and that's why people are are like clamoring for regulation is because like clamoring for regulation is because like clamoring for regulation is because there is an assumption tech companies there is an assumption tech companies there is an assumption tech companies are not going to do this on their own are not going to do this on their own are not going to do this on their own some kind of a gdpr for AI yeah and you some kind of a gdpr for AI yeah and you some kind of a gdpr for AI yeah and you know we have the AI act in the EU now know we have the AI act in the EU now know we have the AI act in the EU now and I do think that could have some and I do think that could have some and I do think that could have some impacts I think it's probably likely to impacts I think it's probably likely to impacts I think it's probably likely to not have the kinds of widespread impact

not have the kinds of widespread impact not have the kinds of widespread impact that people think that it will in part that people think that it will in part that people think that it will in part because of some of the difficulty of of because of some of the difficulty of of because of some of the difficulty of of implementation of some pieces of it but implementation of some pieces of it but implementation of some pieces of it but there are some things that I think the there are some things that I think the there are some things that I think the things that are most gdpr like I things that are most gdpr like I things that are most gdpr like I actually think are the things that are actually think are the things that are actually think are the things that are more likely to happen like requiring more likely to happen like requiring more likely to happen like requiring information about training data and that information about training data and that information about training data and that sort of thing like things that are very sort of thing like things that are very sort of thing like things that are very sort of rot yeah yeah ex exactly and I sort of rot yeah yeah ex exactly and I sort of rot yeah yeah ex exactly and I do think those things could be helpful do think those things could be helpful do think those things could be helpful like you know traceability of data like you know traceability of data like you know traceability of data behind things is a hard problem right behind things is a hard problem right behind things is a hard problem right now we've reached the end of our podcast now we've reached the end of our podcast now we've reached the end of our podcast and we've asked a lot of good questions and we've asked a lot of good questions and we've asked a lot of good questions answered a few I want to point folks to answered a few I want to point folks to answered a few I want to point folks to a recent research article that you were a recent research article that you were a recent research article that you were a part of called teaching ethics in a part of called teaching ethics in a part of called teaching ethics in Computing where you went through 40 Computing where you went through 40 Computing where you went through 40 Years of published research on teaching Years of published research on teaching Years of published research on teaching Ethics in higher education and you Ethics in higher education and you Ethics in higher education and you publish that as part of the ACM I can publish that as part of the ACM I can publish that as part of the ACM I can see it up there in their recent January see it up there in their recent January see it up there in their recent January publication along along with some of publication along along with some of publication along along with some of your colleagues and that just give you a your colleagues and that just give you a your colleagues and that just give you a taste of your the work that you've done taste of your the work that you've done taste of your the work that you've done uh and of course folks can check you out uh and of course folks can check you out uh and of course folks can check you out at your website Casey fester.com and at your website Casey fester.com and at your website Casey fester.com and I'll link to you in the uh in the show I'll link to you in the uh in the show I'll link to you in the uh in the show notes and you've got links to your Tik notes and you've got links to your Tik notes and you've got links to your Tik Tok and your socials there as well yes Tok and your socials there as well yes Tok and your socials there as well yes thank you so much thanks so much for thank you so much thanks so much for thank you so much thanks so much for spending time with me today this has spending time with me today this has spending time with me today this has been another episode of Hansel minutes been another episode of Hansel minutes been another episode of Hansel minutes and we'll see you again next week and we'll see you again next week and we'll see you again next week [Music] [Music] [Music] a