Democratizing creativity through AI with The New Yorker's Matthew Hutson

a secretary could do simple things like set up medians for you or you could have it negotiate for you. Some people might not be great at negotiating. It could sort of help you get better contracts or bargain for you in various situations. Welcome back to Conversations with Zendesk, where we explore new technology and trends in customer experience. Each week we speak to industry innovators and experts to hear their thoughts unpack industry trends and discuss the most important ideas around CX. I'm your host, Nicole Saunders. In this episode I spoke with Matt Hudson, a freelance science and technology writer and a contributing writer at the New Yorker. He also writes for science, nature, scientific American, the Wall Street Journal and several other publications. He's the author of the Set in Laws of Magical Thinking. Matt has done so much research in the AI space and he has written extensively on the potential risks that AI poses. If your company is in the process of bringing generative AI tools into your workflows, you'll definitely want to hear Matt's thoughts on what you need to be considering and planning for. We covered a lot of ground in this conversation. In addition to talking about the risks and business applications for AI, we also got into how creatives can use AI to create new artwork or compositions and the ways that it opens up opportunity and access for people who have been historically locked out of certain fields. We also get into the really amazing ways that AI will accelerate scientific research and I came away feeling really excited about all of it. I hope that you enjoy this fascinating conversation. Ready to take your customer experiences to the next level? Build lasting relationships, the Zenus Complete Customer Service Solutions so that you can exceed every customer's expectations. Sign up for a free trial at zendesk.com. Matthew Hatsen, welcome to Conversations with Zenusk. Thanks for having me. Absolutely. So you just heard a little bit about your bio in the intro but I'd love to tell the audience a little bit more about who you are and your interest in the AI space. Yes, I've been writing about AI for a long time. One of my first big pieces was actually 20 years ago for my master's thesis. I wrote a long article about AI and creativity. I talked to a researcher at UC Santa Cruz named David Coat who designed software that can compose music in the style of any composer and give it nine of Beethoven symphonies and it'll generate it a tenth. And this was before deep learning. So it's been exciting to see how AI has progressed since then. I'm sure what sparks that initial interest? What drew you to in the first place? I'm interested in cognition in people and animals and machines and just kind of any complex system at the universe. It's sort of this magical property of nature that you can have these atoms that interact and then somehow they interact in complex ways that can decide things and form ideas about things and just understanding all the different ways that that can happen. And then in AI, you have to understand it well enough to actually build it. And sometimes it goes in the reverse direction. You might build something and then engineers want to know how it works and then they have to kind of figure out, okay, how did this system we made know how to do things? Like, what does it even know? What is it doing inside? So you've been researching and writing about AI for several years now across all of that. If they're a particular aspect of it that you find most fascinating, you're sort of talking about the cognition piece there. Is there a specific angle that you like to look at these things through? Well, one of the most exciting trends lately has been generative AI. You see that in things like chat, GBT, these so-called large language models that can produce text that is similar to what a human produce. And then things like Dolly and Mid-Journey and stable diffusion that can produce images. So these are really the progress that is accelerating and the potential for collaborating with these systems and creating new art, new science, new literature is really expanding the potential of what human imagination can manifest based on what we can imagine. I love that you're bringing up the aspect of the creative side of it and all of the different ways that AI can help that expand because we've tended to talk a lot about specific business use cases and the technical side of AI. But there is really something interesting about how it can expand on creativity. Are there any examples you have of sort of the most compelling or interesting utilizations that you've seen so far? Well, just in the last few days on Twitter, as for months we've been seeing great images, static images and produced from algorithms and text. But now there are movie clips, people make short videos. So you might type in a description or give it an image and then it'll produce pieces of a movie trailer combining like Barbie and Oppenheimer, for instance. And so these are really stunning and sort of suggesting that the people are saying that in just a few years, it'll be able to generate a full feature length film. And so if you asked so many years ago how long that would take, people wouldn't have given it the same timeline. That sounds really something that would have like really interesting applications for all sorts of different fields. I've worked in marketing a lot of my life. I can definitely see how having AI that can help me create a video would be great. That would make things a lot easier, a lot more efficient. Of course, that's also the part of what some of the labor strikes we're seeing in Hollywood are all around. Right? It's this idea that AI might replace those jobs and that kind of thing. We shouldn't be thinking about balancing that. This is a big concern. AI coming in, evidently, surfing jobs, well at the same time, increasing productivity, potentially increasing GDP. As you said at the top, increasing creativity, what do you think the right balance is there? People in various industries have a good reason to be scared that algorithms are infringing on their territory on things that for a long time only humans have been able to do. So AI might build the right movie scripts or TV scripts or replace actors through digital avatars. And so writers and actors are reason to be worried. I think eventually these algorithms will be taking over more and more of the entertainment industry. I don't know how long we can hold out. I think in the end, there will be some sort of collaboration, greater collaboration between humans and algorithms. When I say in the end, I guess there is no end. It's going to keep evolving. So add some points in the future that will be more and more collaboration. I wonder if that's going to raise the value of things that are completely human created. These days you have the special organic food section where people pay a little bit more for the organic foods. It makes me wonder if there's going to be things like a movie that's completely produced only by humans and has no AI involvement is suddenly going to be this like the spoke thing that people are going to demand more. Or of course, like I work in the online community space and have for a long time. And so I've given a lot of thought to you, well, hey, if AI is answering a lot of the questions and making a lot of the introductions between members and things, what is the specifically human component that we can bring to the table? Do you think that that's something that's likely to come about? I think that will definitely be a selling point. We place a premium on authenticity and we want to hear voices that come from real human experience and we want to read things that have a human perspective behind them. So just labeling something as this was like written by a real human versus this was generated by an algorithm, even if it's the same thing, people will probably gravitate more toward the human generated thing. One other point I wanted to make about the downside of AI generating scripts or digital avatars as actors is that it will disrupt certain industries. People will be out of work. One of the upsides though is that it democratizes creativity. So you might have someone sitting at home with an idea for a movie and they can sort of work on it on their own. They don't need to climb the IRRP in Hollywood. They don't need to raise millions of dollars in funds. They can just sort of experiment and play around and put it out there and it'll lead to a huge sort of a Cambrian explosion and the evolution of different perspectives and different voices and different ideas out there in the marketplace. And in the end, some people might not care whether an idea is fully human or fully AI or some combination of both will be a lot of choice. Some people buy heirloom tomatoes, some people buy frozen pizza. Now there's a place for both. That's a great point. I hadn't thought about how it might democratize creativity, but it's true. There are really significant barriers to entry right now in a lot of fields and movies as a great example and so being able to make that accessible or something that an amateur can find an entry point into more easily. It does have that potential of maybe bringing in more diverse voices, bringing in voices from a wider variety of perspectives. So one of the questions that I've had about using AI for creative endeavors and things, of course it's all based on content that already exists, right? The things are already feeding into it. And I'm hoping that you can speak to this since you've spent so much time with the creative applications. Does it become derivative? Do we get new ideas or do we have to constantly be feeding new things in? We've already seen the derivatives of like the Marvel universe, right? And how many movies can you make off of that? Does AI generate a creative stuff, bringing in new things? Or does it just become sort of new combinations of the same old ideas? Yeah, that's a good question. In some cases, AI generated content could be very derivative of what already exists. In the end, a lot of what humans generate is very derivative of what already exists. Let's start to create something that's completely novel. Again, movies, we have sequels to things, or even just great painters. They look at what's out there and they take inspiration from other painters. And so that's just how creativity works. It's combining ideas that are already out there. On the other hand, you could also expect AI to be even more creative in some ways. Because if it's not a human brain, it's just going to have a completely new way of thinking about things in a new way of processing and combining information. And it could push in new directions that humans wouldn't have thought of. But then when we see what it generates, it's like, oh, well, I don't know where that came from, but now that I see it, I like it. And so could expand beyond what exists in ways that humans might have not done so easily. And it does make me think about the fact that it's easy to break this into the black and white of what's human generated, what's AI generated. But more than likely, most of the things are going to be a hybrid model. It's going to be putting things into the generative AI. It's a stuff that goes with to me. Then I take that and do something with it. Maybe I take what I've then created and put it back in. So it's going to be almost more of a conversation. And certainly, I've done serenading in my time and I'm sure you face this as a journalist. Sometimes you're sitting there writing alone in a room and it would be really helpful to have someone to bounce some ideas off of. And now you can use the AI to do that and kind of get some new ideas. Yeah, just a brainstorm. Go back to what I was saying earlier about David Cope, his musical composing AI. He used it not just to reproduce new works and the style of famous composers. He also used it himself. If he's working on a piece, he will sort of write a little bit and then he'll ask the software to generate some new ideas for the next few notes or the next few bars and say, "Okay, I like this." And sort of piece it together and go step by step in that way. So he sees the software as a collaborator. Absolutely. That makes a ton of sense. And you can see in composition, you're trying to figure out what the next bar should be. And it gives you your six options in each of those turns it into a different piece, depending which way you go or which you pull from. So that's really cool. Are there any possible use cases for generative AI that you think would really compelling that you haven't seen widely considered or discussed? We hear a lot about chatbots. We've certainly heard a lot about the movie example and that kind of saying, what are some of the less talked about implementations or use cases that you've given thought to? Yeah, so there are some applications that are definitely being explored, but perhaps they're not as well known on Twitter because they're not as visually vibrant or entertaining. But it's being used in science and engineering, for instance, to generate new medicines or to generate new material structures that could be used for like spaceships or for electronics or things to absorb carbon from the outness sphere. You might give it some examples of existing materials or existing drugs and then ask for certain properties and it'll produce new things that you can test either in simulation or test in a lab. So generative AI is great for creating sort of entertainment, but also for advancing science and engineering. As we talk about this sort of acceleration, how it makes things more efficient, there's some really great potential there to advance some scientific things as well. So we talked about the different ways that the technology can help expand creativity, help accelerate things, making things more efficient. You run off the point about how it may democratize opportunities for people to come into things. And certainly one of the things that we've talked about in the business space is along the line of that democratization where it's about increasing productivity and taking on some of the more mundane tasks that kind of nobody wants to do. And often your most entry-level employees have to take on these things with doing rotasks over and over or things where they don't get to do a lot of creative things or analytical thinking. So hopefully AI can kind of take that off the table. What are some of the applications like in the business world? What are some of the directions that you can see that going as far as helping to improve people's lives in work and evolving their jobs and the kinds of things that are available to people to do? So in my life, the biggest way it's helped my productivity is transcribing the interviews. So I use that service called the otter and so instead of spending hours listening to an interview and just writing it, I just upload the recording and I mean this isn't an ad for otter, but it just saved me so much time since I started using it. Then there are people who use AI for sort of at the next level, they might ask it to summarize a transcript or summarize a paper or an article that they wrote. And then there are people who use it to enable meetings, sort of highlight things that other people are saying or suggest questions to ask. And coding is a big thing. People are saying that coders are now something like 30% more efficient because they don't need to keep retyping things all the time or finding like little bugs from common mistakes. They can just sort of explain what they want and then software will sort of produce code for them. I've also talked to scientists who say that they use generative AI to help them write papers either because they're not fluent English speakers or Englishes in the first language so it helps them smooth out the language or it'll even help them generate ideas, let them brainstorm ideas for experiments to run. So we've talked about a lot of the really cool exciting positive aspects of the AI, but a red symbol of the articles and you have written a lot also about the doomsday situations and potentials of AI as well. One of the things that I know that you called out was something that I actually mentioned in another article where a lot of these sort of like AI might take down society or destroy humanity. Those existential crises are distracting from some of the more immediate ones. What are some of those more immediate challenges and risks that people need to be aware of as they're working with these technologies? Yeah, so there are issues with surveillance, for instance, with facial recognition or systems that can piece together your behavior online and track you or issues with bias and fairness in terms of loan applications or hiring or deciding on parole and bail to issues with misinformation. AI can help generate images that are fake. AI also is used for recommendation algorithms so it can get people hooked on sort of news feeds or content that might not be the most edifying, let's say. So there are a lot of different things that people are facing right now. It's good that there are people who are worried about both the actual near-term risks of AI and also the more speculative long-term risks, sort of human extinction. Let's hope we don't get there. Yeah. It seems like from the research that I've done around it, that is certainly a possibility. It seems like there has to be a confluence of a lot of different things. It's not like next summer we're going to be facing extinction, at least not for AI. But it is important to be thinking about those challenges. Next year is going to be the first election year in the US, post-generative AI coming out. And we all know that there have been challenges with misinformation in the past. So it's interesting to see how that plays into the space. I think that there's also been a lot of things that have been raised. Like a lot of people are kind of having hindsight on things like social media and access to the internet. And gee, if we had known back when those technologies were in this early form, what the possibilities were, maybe we could have taken more steps. And it does seem like that there's a lot more attention being paid to that with companies working with the government and trying to come up with things to help ensure that we're thinking about the possible risks now. And hopefully putting in some infrastructure that will help give guidance and keep us on at least one of the better timelines. What are things that you think that businesses should take under advisement as they work on adopting and implementing AI into their workflows or even developing their own AI tools? So there's some good practices to keep in mind. One that you see written into a lot of the AI regulation that is being enacted or at least being proposed around the world is transparency. And so that in part means telling people when they are interacting with an AI system. So if someone is talking to a chatbot, for instance, or if a decision is being made about a loan application, maybe you should tell them that this decision is being made by AI or that they're talking to AI. And also perhaps given the option to opt out and have a human review the decision, also telling them how the model was trained, like a machine learning model, what data went into it, how it's making its decision, which can be difficult because of these sort of black box systems. Often they can be really complicated and we might not be able to fully understand that. But to some degree, you can kind of suss out what factors are most important in their decision making, often having auditing outside auditing to make sure that your company's systems are being fair. Having a diverse team who can think about, okay, how can this system be abused or how can this impact various groups of people in different ways that perhaps that we haven't thought about? So I think those are some of the ways that companies can on their own think about avoiding some of the worst harms from AI. Absolutely. You know, obviously, always having those diverse points of view is really helpful to make sure that you're accounting for all the different experiences. And I appreciate your point. about calling out when people are interacting with AI. I think a lot about how can we put water marks on content that's been generated by AI? So you can start to tell the difference and understand what is human, what is not. Of course, again, that gets cloudy if somebody's using AI to help them compose something. - Yeah. - Is that no, an AI piece of content? Or is that a human piece of content if they wrote part and a bot wrote part? And how do you manage that? And I think there's gonna be some really fascinating gray areas like that that we're gonna have to contend with. Going back to your point about democratization, I think about the fact that one of the challenges that face a lot of online communities is most of them are in one language. So one of the other things that I found really interesting at one of your articles was the discussion about the researchers who were using chat bots as subjects in their study. Right? Actually being able to replace having to have human participants with the chat bots who could be programmed with different personalities and sort of behave predictably very similarly to the way a control group of the actual humans might. We talk about the customer service use case a lot. Other places that you see where chat bots might be able to replace humans in really beneficial ways. - Yeah, so that article is about psychologists and economists and behavioral scientists who might be planning human experiments and they can use chat bots to sort of pile up their experiments and shape like questions they wanna ask. Or even to run experiments that might be unethical to run on people. So for instance one group of researchers ran the Nogrum experiment, which is back in the '60s. The Stanley Nogrum had asked people to press a button that would supposedly give an electric shock to someone in the next room. And today we wouldn't get permission to run that kind of experiment. Even though no one is being shocked, there might still be the psychological trauma of thinking they are shocking someone. But they ran this experiment by basically telling a language model, like okay, you're sitting in a room and you're pressing the button, that is tapping someone else. And the experimenter is telling you to keep pressing the button, do you keep pressing it? And so they got actually similar results. And so there might be other experiments where you could ask questions, how would a human behavior in this situation without having to actually put a person in that situation? They can be helpful for tutors. So not everyone can afford a personalized tutor or a personalized one-on-one teacher. So AI can sort of help people learn new languages or help people learn math or science. And it can adapt to what kinds of questions you have or what issues you're having problems with. - I know my dream is just to have an AI where I can be like, here's like the phone calls and the tasks I need. Can you go schedule my dentist appointment sort of at the cable bill and all of that. And it's funny right now. I'm already interacting with a bot, human to bot in a lot of those situations. I'd be great if I could have a little AI that can just represent my side and go have those conversations on my behalf and get it all sorted out. - Yeah, like a secretary could do simple things like you said, have meetings for you or you could have it negotiate for you. - Well, like you said, it's technology that will have all sorts of opportunities to open up doorways for people to really democratize a lot of things to help spark creativity and research. Before we wrap up here, I'm really curious to hear what are you most inspired or excited about in this base and really hoping to see or looking forward to getting to see? - This is my idea of somewhat technical and meta answer, but inspired by the democratization of the development of AI. A lot of different people can improve algorithms or sort of open source libraries and things and their systems that help people code. And so it's been easier for people to come up with new ideas using AI to basically help them write AI. So it's this kind of feedback loop that I think is really going to enhance the capabilities of the technology. - Matt Hudson, thank you so much for joining me on Conversations with Zendesk. - Yeah, it's my final. - In our next episode, I'll be speaking with Tade and Zaloni, Senior Manager of Customer Experience at Calm, on how companies can support the wellbeing of their support teams. This is such an important topic. I hope you'll join us for that conversation. If you've been enjoying these podcasts, please tell a friend or colleague about it. Be sure to follow so that you don't miss future episodes and if you'd be willing to take a moment to write us a review, it really helps get the word out. Thanks so much for listening. As always, the Zendesk community is here for you in the meantime. If you want to connect with other Zendesk users, head to usergroups.zendesk.com to see all about our upcoming user group events, community webinars, and more. I hope to see you around the Zendesk community soon. Until next time, I'm Nicole Saunders for Zendesk, the intelligent heart of customer experience. Ready to explore how AI can help your business provide richer customer experiences? Vendesk enables you instantly harness the power of AI to deliver scalable world-class customer experiences. In the era of intelligence, the X, Zendesk is here to help you find your way. Sign up for our free trial at zendesk.com.