The AI industry is at a major crossroads

Support for this show comes from Adobe. The all-new Adobe Acrobat Studio is reimagining how we use and interact with PDFs and the powerful impact they can have for your business and personal projects. And with their AI-powered PDF spaces, you can collect files, synthesize information, and even chat with an AI assistant for fast insights. It's time to do more with PDFs than you ever thought possible. And you can do that with Acrobat.

Learn more at adobe.com slash do that with Acrobat. That's adobe.com slash do that with Acrobat. Support for Decoder comes from Anthropic, the team behind Claude. When you're analyzing a complex tech strategy or trying to understand the forces behind major industry decisions, Claude can help. Claude is AI that goes beyond easy answers to help you explore the nuanced questions that drive real understanding.

Whether you're dissecting platform strategies or exploring regulatory implications, Cloud works through complexity alongside you. Try Cloud for free at cloud.ai slash decoder. Avoiding your unfinished home projects because you're not sure where to start? Thumbtack knows homes, so you don't have to. Don't know the difference between matte paint finish and satin? Or what that clunking sound from your dryer is?

With Thumbtack, you don't have to be a home pro. You just have to hire one. You can hire top-rated pros, see price estimates, and read reviews all on the app. Download today. Hey there, and welcome to Decoder. I'm Hayden Field, Senior AI Reporter at The Verge and your Thursday episode guest host. I'm subbing in for Neelai while he's still out on parental leave, and I'm excited to keep diving into the good, the bad, and the questionable in the AI industry.

It's been a very big news week in AI, and a lot of it had to do with OpenAI. The company hosted its annual Dev Day in San Francisco on Monday, and I'm still here in person covering all the news. They announced a bunch of ChatGPT product features and new agent tools, and executives also laid out a pretty bold vision for the future of AI. At the same time, the new Sora iOS app has shoved AI-generated video into the mainstream.

creating all sorts of unintended consequences, and even surprising OpenAI CEO Sam Altman, who's become the face of sore memes across the internet. And earlier this week, the New York Times published a great story about how AI-powered job screening has become so prevalent that applicants are starting to sneak in hidden messages to chatbots inside their resumes, effectively trying to prompt inject the automated job screening process for a better chance at an interview. I brought in Kanjun Kyu.

CEO of AI startup MBU and a close watcher of the industry to help me break all this down. Ken June has been both a tech founder and investor, and her perspective on AI and the broader tech industry in general is a very unique one. So I wanted to chat with her about this week's biggest AI stories to break down what's really happening, why it's happening, and the societal implications of it all. Okay, MBU CEO Kan Jun-kyu on the good, the bad, and the questionable in the AI industry this week.

Here we go. Ken June Q, CEO of MBU, welcome to Decoder. Thanks, Hayden. I'm glad to be here.

So let's jump right into our first story of the week. Yesterday, OpenAI at its annual Dev Day event announced ChatGPT apps making... apps available within ChatGPT. Developers can build them. We saw it launch with a bunch of them. So Booking.com, Canva, Coursera, Expedia, Figma, Spotify, Zillow. And soon they're also going to have DoorDash open to... uber target in there so right now that obviously means you can as a user of chat gbt ask for one of these apps to do something for you within the

ChatGPT interface. I want to know if you saw that announcement and how big of a deal that was to you. Did you think that was the biggest announcement of the day or, you know, just a small incremental step forward? I think that this is all to be expected. You know, this is kind of where the entire Asian ecosystem has been going toward. And the way I think about it is it's kind of like the iOS of AI.

where you have this single interface that OpenAI is trying to build that is the way that you as a user get into all of these different apps through this single interface, ChatGPT. And I think that's a big deal. And I think that there are going to be really major implications on the power dynamics with AI going forward from that kind of approach. How so? Yeah, I think right now we're kind of at a crossroads on whether AI becomes another walled garden platform.

situation, just like the internet has been the last 10 years, or whether we end up having AI that actually democratizes and decentralizes power in the way that the original internet did or the personal computer did. Right now, we're in this kind of platform network effect ecosystem right now where you have these AI model providers. and they are providing something like this that lets you integrate with a lot of other apps through partnerships.

And what that means is that just like with iOS, Apple, they're kind of trying to lock you into their platform ecosystem. And what we've seen, there's a really good talk from Cory Doctorow on what he calls in-shittification, which is... platforms, they'll start out doing great things, you know, being really useful to your users like Facebook. And then over time, what they try to do is they lock in their users and then we end up being at their mercy and we can't exit. And so we're really...

in a place right now where the question is like, do we control AI or does it control us? Do we own it or do we just rent it from these centralized platforms? And I think there's a risk today of people losing the ability to shape the digital systems that control our lives. And you can see it a little bit in...

how we relate to our digital devices right now. Like my phone, I feel really aversive to my phone. You know, I need to detox from my phone or my TV or my computer. And that's a really weird relationship to an object in my life. Like if I had to detox from my sofa, then I would... get a different sofa. Right? But we don't think about that. So why? Why do we have to detox from our devices? It's because our device is full of...

things that are built right by other people and have incentives that are not necessarily aligned with ours. And as AI gets more powerful, those incentives, those will become more powerful as well. And so we want to be in a different world. And so that's kind of the risk I see. I think it's great, like really cool, very empowering. And also like we're kind of at a turning point in time right now.

That makes sense. I like what you said about the crossroads about whether it becomes a walled garden. And right now, I wanted to also ask down the line, at the moment for these apps, OpenAI is controlling distribution. Only large businesses are really being allowed to enter this space right now. But down the line, the current vibe is that anyone, any developer will be able to build an app and maybe it...

is integrated within ChatGPT. So what happens when this gets opened up like that? Do you think controls and safeguards from OpenAI will be enough? You got at this a little bit a minute ago, but yeah, I'd love to hear your thoughts.

Yeah, I think that part is going to be quite similar to what we saw in the last 10 years with the mobile ecosystem, with iOS and Android, lots of people building quote-unquote apps and getting distribution through the platform. And that's not a problem. I think there will be plenty of safeguards. AI has great incentive to try to make a really healthy ecosystem and good apps that don't cause people problems.

However, it doesn't solve for the control problem of do we actually have control over our digital environments?

Right. Let's talk a little bit about MPU's approach to AI agents compared to what we saw yesterday. So we already touched on this a little bit, but you guys are all about the decentralized approach. How does that play into everything we're seeing elsewhere? And talk a little bit about how you guys are hoping to do things differently.

I think it's a really hard problem. And the moment in time we're at is kind of where we are in our digital environments today is what I said about our phones or our devices, where we're really renting. these apps we don't own them somebody else is making them and we're using them and that means that you know there are incentives where people who make them want to make money off of them and that's generally okay

The way in which we really get into trouble is these devices, digital systems, they have very fine-grained controls that capture our attention. They notify us. embedded in our workflows. And so because other people are controlling these apps, it's also easy for them to start controlling us. I think actually right now, today, we have a window of opportunity to really change this dynamic of control.

In the past, the reason why we have this dynamic with software creation is because it's been really hard to make software. So you need to hire developers who are really expensive, build software, pay for the development, and then sell it for a profit.

But we're starting to get into a world where, as OpenAI Dev Day demonstrated, you can generate apps, generate things using natural language, and these language models are starting to get really good at writing code. And so in theory, we could go into a world where... a lot of people can write code. And a lot of people could not just write code, but take a piece of software and change it to make it fit for themselves.

An example of that is like a doctor who is dealing with an electronic medical record system where they have to put in all these fields. It's not suited for them. And a lot of doctors encounter a lot of burnout from the situation. In theory, they could, in this future world, just have a system that helps them change the software at the point of use. We call this modifying at the point of use. And so I as the user know what's best for me. I know what I need.

And I can change my digital environment right in the moment so that it suits me better instead of suiting whoever designed the software. However, that requires inventing a few different pieces, just like how...

You know, the moment we're in, it reminds me of the 1960s. In the 60s, people were really excited about supercomputers. They thought, everyone's going to timeshare on supercomputers through these terminals. Like, we're all going to be in terminals, timesharing, and using supercomputers. It's going to be super cool. And then in the 70s, late 60s, a small group of people at Xerox PARC invented the mouse, the GUI, files, folders, windows, everything that allowed us to...

personal computers, and that led to personal computing today. And what that did is it made the technology capabilities more accessible so that people could use computing for themselves and modify their computer for themselves. And I think we have an opportunity kind of to make the same inventions for the AI ecosystem. Got it. That makes sense. And back to yesterday at Dev Day, they made a lot of agentic AI announcements. Obviously, this is kind of just playing into...

OpenAI and AI companies in general's broader agent goals. But I would love for you to chat a little bit about how their agent updates. differ from the decentralized approach you guys are taking. So one corporate example they used was Albertson's grocery stores using OpenAI's agent builder to ask about why ice cream sales were down and create a plan for getting them back up again using their own data.

and their own traditional marketing. So what did you make of those announcements and how does that kind of play into what you were just getting at in terms of changing things up at the point of use for your own devices? There's a difference between empowerment and control. And so the agent kit is super empowering. I can now do things I couldn't do before. And that's super cool. Like that's with all technology is empowering, you know, and like.

That means I can build apps, I can answer questions I couldn't answer before. All of that is really exciting. The difference is, with OpenAI's system, you can build apps, but only within their platform. Your agent runs on their servers. It follows their rules. It pays their API fees. You get distribution to their billion users, but you risk losing control over.

what you're trying to do. One kind of silly example is Imbue. Recently, we last Tuesday actually launched Sculptor, which is a tool for building software using coding agents. And it's a Mac app. And when we were trying to register with the Apple App Store, Apple banned us, even though we are a legitimate company. And it took us several days and pulling a lot of internal strings at Apple to figure out how to unban ourselves.

And that's an example. It's a very trivial example, not huge downside effects, but it's a trivial example of a way in which we have lost power relative to a centralized platform. And so... The question is like, what would it look like if you controlled AI instead of platforms controlling AI and you using it through their platforms? We need to take a quick break. We'll be right back.

Support for this show comes from LinkedIn. When you're a small business owner, your business is on your mind 24-7. So when you're hiring, you need a partner that works just as hard as you do. That hiring partner is LinkedIn Jobs. When you clock out, LinkedIn clocks in. LinkedIn makes it easy to post your job for free, share it with your network, and get qualified candidates that you can manage all in one place.

LinkedIn's new features can help you write job descriptions and then quickly get your job in front of the right people with deep candidate insights. Either post your job for free or pay to promote. Promoted jobs get three times more qualified applicants. At the end of the day, the most important thing to your small business is the quality of candidates. And with LinkedIn, you can feel confident that you're getting the best. Based on LinkedIn data, 72% of SMBs using LinkedIn

Say that it helps them find high quality candidates. Find out why more than 2.5 million small businesses use LinkedIn for hiring today. Find your next great hire on LinkedIn. Post your job for free at linkedin.com slash partner. That's linkedin.com slash partner to post your job for free. Terms and conditions apply.

Here's something to think about. Why do we assume simple and affordable has to mean basic or just for beginners? Professionals want to save time and money too, and they deserve tools that actually make life easier. That's the idea behind Spaceship, a platform that simplifies buying, connecting, and managing domain names, with over 4 million domains registered already. That's alongside offering web hosting, business email, and more advanced business products like Virtual Machines and Hyperlift.

which lets you create apps, APIs, websites, and more without touching infrastructure. You're ready to launch new ventures in seconds. Plus, Spaceship comes with extra highlights, including Unbox, making it super easy to connect your products to your domain, or ALF, your AI buddy, that can handle domain transfers, DNS updates, and all the tedious tasks, and roadmap.

a community-driven way to suggest and vote on features so the platform evolves around what professionals actually need. So for exclusive deals, visit spaceship.com slash decoder and use the promo code decoder. That's spaceship.com slash decoder. Promo code decoder. Support for this show comes from Adobe, who are introducing the all-new Adobe Acrobat Studio now with AI-powered PDF spaces. Look, I'm sure when I say PDF, you have a very specific thing in mind.

And I'm guessing it's an email attachment. Certainly not a dynamic asset that can help elevate your business. But Adobe Acrobat is changing that. It's time to do more with PDFs than you ever thought possible. Need AI to turn 100 pages of market research into five insights with a click? Do that with Acrobat. Need templates for a sales proposal that'll close that deal? Do that with Acrobat.

Need an AI specialist to tailor the tone of your market report to sound real smart in real time? Do that with the all-new Adobe Acrobat Studio. It's time to reimagine and rethink what a PDF can actually do. Learn more at adobe.com slash do that with Acrobat. That's adobe.com slash do that with Acrobat.

We're back with MBU CEO Kanjun Q, talking about the biggest AI stories this week. Before the break, Kanjun and I were discussing OpenAI's Big Dev Day announcements. And Kanjun's perspective was that we're at a crossroads in the AI industry between an open, their controlled vision for AI and the more walled garden, platform-centric approach that dominated during Web 2.0. But now I want to ask Ken June about the other major open AI story going on right now. That is, of course, the Sora app.

Let's get into our second story of the week, which is Sora 2 and AI slop in general. Obviously, over the last week, you, I, and everyone else we know probably has seen AI-generated video blowing up. It's because of Sora 2, OpenAI's new...

iOS app that, though it's invite-only, has started flooding social media feeds with videos. of Sam Altman, Nintendo characters, Nickelodeon characters, and even some unsettling political commentary. So I wanted to see if you have access to Sora. Are you using it? What have been your thoughts so far on the first week of this strange? new platform.

I think Sora shows what happens when AI capabilities meet platform incentives. It's really interesting. Platform incentives we talked about with social media, they optimize for engagement and attention capture. And that's what allows for making money, profit, ads, et cetera. Generally, this is not a huge problem until you take it too far. It's actually a perfect case study when we talk about the shape of tech power.

My friends at OpenAI talk about how it's actually cool. They're trying to create a video generation app that creates joy, gives people moments of entertainment and joy. And I think from my perspective, that's a great thing. And... Anytime you're trying to create joy and also maximize engagement to maximize profit, it is not necessarily, I think it's like very difficult to trust.

And so I think what I would expect to happen is that this platform engages a lot of people, gets a lot of users using it and creating lots of videos, and there will be... people showcasing videos and how cool what they made was. And that's not such a bad thing. However, over time, a lot of people will feel similarly about this as they feel about TikTok. Hey, three hours went past and that's not what I intended. What happened?

And it goes back to kind of this idea of control. We're in a world where our digital technology controls us and not the other way around. That makes sense. I was going to ask you if you think OpenAI is trying to cash in on the entertainment industry side of generative AI with Sora. Is it just a money-making effort at the expense of our attention spans? And yeah, you just gave a pretty good answer to that.

They had a Sora Cinema pop up where you could watch some of the videos that people have been making. And I and a few other reporters did feel a little depressed watching some of them. It was an interesting case study, like you said. I would also love to ask you.

I've seen OpenAI moving pretty quickly here to try to make a good faith effort to rights holders, or at least they say they are, and prove that they're not in the business of rampant copyright infringement, which is something that was super controversial the first week of this app. life cycle. But the bigger question here beyond the licensing and rights issue seems to be more about whether this is actually creating any type of creative value.

or in fact, just contributing to an already toxic internet. So where do you land on AI generated video? I know you already mentioned a little bit about how it may end up just being like TikTok and just a thing that... take some of your time away unwittingly and no worse than other social media apps. But what about the misinformation side of it and the fact that this system specifically is a lot more realistic than most others? You're not seeing any six-fingered hands anywhere.

I think it's a really complicated situation because on the one hand, it is true that AI-generated video lets a lot more people express themselves. Who couldn't before? They didn't have the tools. Or the tools were super hard to use. And I think that's really cool. That's really powerful. And on the other hand, like, I always say people are not bad. Incentives are bad.

Like on the other hand, we're in this incentive ecosystem of the internet right now where there's kind of like rampant attention capture as the primary way of making money off of things. And what I would expect is some people will use it to really deeply express themselves, and it's going to be really cool. And other people will use it in order to capture attention and optimize for that. And OpenAI, as the platform, also has an incentive toward

capturing attention and increasing engagement so that they have more users. And so I think that the... The technology itself, Sora, generating video is like really, really cool. And yeah, it has misinformation problems. We have to figure out how to do like watermarking or something. to verify the reality of something. I think that has to be solved a different way. But the problem with it is that it's being launched into an environment that has rampantly inhumane, I think, incentives.

where as humans we are losing control over our attention. And that really sucks. I don't think that's a future we want to live in. It's also interesting because... The watermark aspect of it all, there have already been a ton of tools proliferating that I've seen, whether it's tutorials for removing the watermark, people coding their own ways to remove the watermark.

or using like magic eraser type tools to remove it. I've seen a bunch of these types of videos on other social media apps with no reference to it being created with Sora. And in some ways, OpenAI isn't responsible for what people do once. The video leaves the platform, but in another way, they did make the technology and we're seeing the president of the United States.

share a lot of AI generated videos. So I'm wondering, when we get into another election cycle, and it's hard to tell what's real and what's not, how this is going to affect everything, just as the tech becomes more and more advanced, I used to pride myself on being able to always tell. when something was an AI-generated video. I mean, we all could. Now it's a little bit more thorny. Yeah, it's really hard to tell these days. I think to your point, the current technology ecosystem...

we relinquish responsibility over the technologies that we built. And I think that's actually not a very healthy moral philosophy of technology builders. As technologists, I believe that we should be like... ethically, morally responsible for the way that what we build impacts society. And you said, in a way, it's not OpenAI's responsibility, but I think in a way it is. And we are responsible for what we create and what we allow people to do.

To your point about watermark removal, I actually think longer term we need a different mechanism for trust and verification on the internet of information and of data, and we just haven't figured that out yet. Yeah, based on that, do you think likeness laws, copyright? Are we going to need to update our entire conception of intellectual property to figure out how this tech fits in? I do think we need to update our conception of intellectual property.

interesting to think about who's responsible for what here and yeah we're going to be dealing with a lot of these questions over the next few weeks months years we need to take another quick break we'll be right back Support for Decoder comes from Anthropic, the team behind Claude. The best tech journalism takes a seemingly straightforward announcement and reveals all the strategic layers underneath. That's exactly the kind of thinking Claude excels at.

Cloud is an AI thinking partner that goes beyond easy answers to help you work through the nuanced questions that drive real understanding. Whether you're trying to understand why a platform made a particular policy decision or what a merger really means for the industry, Cloud can help you explore connections you might not have considered. From analyzing corporate strategy to exploring regulatory implications,

to just trying to understand how these systems actually work, Claude thinks through complexity alongside you. It's not about getting quick takes. It's about developing the deeper insights that matter. Try Claude for free. Support for this show comes from Agency. We're talking about multiple AI agents working at once without human intervention. Agency. That's A-G-N-T-C-Y.

is working to ensure AI agents aren't siloed off from one another and instead are able to securely discover, connect, and work across any framework. With Agency, your organization gains open, standardized tools and seamless integration. That includes robust identity management to help you identify, authenticate, and interact across any platform. Agency is leading the way in established trusted identity and access management and empowers you to employ multi-agent systems.

It's already working with industry leaders like Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and more to set the standard for secure, scalable AI infrastructure. Now, Agency is an open-source Linux foundation project. If your enterprise is ready for the future of Agentic AI, visit agency.org to explore use cases now. That's A-G-N-T-C-Y. Support for this show comes from Airbus. It took 100 years for electric vehicles to catch on. Modern solar panels, half a century.

Lithium batteries? Decades to go from their debut to daily use. It's a pattern. Energy tech breaks through, stalls, and then something tips the scales. But for every success, there are even more almosts. So what if there was a breakthrough sitting at this crossroads right now? The potential to cut carbon emissions with Sustainable Aviation Fuel, or SAF, is real. SAF works just like Jet A1.

which is a standard jet fuel used worldwide. But the difference is its contribution to aviation emissions reduction. it could change the future of flight. And that's why several major U.S. carriers have already increased their commitments to fuel with SAF. But it's not just up to the airlines. What's needed is an industry-wide push to increase the supply to meet growing demand. You can learn more about how Airbus is contributing to accelerating this journey at fly.airbus.com.

We're back with MBU CEO Kan Jun-kyu. Before the break, we were discussing Sora. the unintended consequences of unleashing ultra-realistic AI-generated video across modern social media, and the implications of technology like this despite guardrails.

Now I want to turn to the last big story of the week that we'll discuss. How AI is transforming the relationship between companies and job seekers, many of whom are now facing an increasingly uphill battle, trying to break through automated systems to land an interview.

Let's move into our third story of the week, which is recruiters using AI to scan resumes and applicants trying to trick the system. The New York Times. posted a story about this. There's a lot of concern about AI screening and how it might be factoring into everything from housing to hiring, of course. And this story today...

was about how AI, now a pillar of corporate job screening, has put all kinds of pressure on job seekers to try to game the system, making sure their resume is seen and not just going into a black hole so they can land an interview. And the opening detail of the story is pretty remarkable. It's about...

recruiter in the UK who found that one applicant was hiding instructions for ChatGPT on his resume in a different font color, instructing it to select him as an exceptionally well-qualified candidate. Do you think these concerns about AI screening are justified?

And what do you make of all this? So this is actually a perfect example of when AI mediates human relationships. So we're going into a world where algorithms, AI, I use them interchangeably, and algorithms are making more and more decisions. about our lives, whether we get a job or whether we get our mortgage or whether we get approved for something or prisoner recidivism. There's all of these different things where the algorithm is making the decision.

This is like a perfect example of what you'll see in that situation when there's no recourse for the human. So when there's no recourse for the person who's being affected by the algorithmic decision. we will see these arms race effects. And so, yeah, I think this is basically, let's see, I actually built an ML recruiting startup in my last company and saw that when AI systems are black boxes, people will try to game them.

And ultimately, I think the real solution is better incentive alignment between the creators of the algorithm and also the users, quote unquote, and also laws. better laws that protect people it's about power and in a way about control like when you have a human making these decisions you can generally appeal like hey

I think you misunderstood or something like that. But right now, our legal infrastructure doesn't cover algorithms very well. Like, we don't have very many laws that talk about how should we govern the way that algorithmic decision-making affects people. When we have people... making decisions that affect people. We do have laws that govern, like discrimination and things like that. And some of those laws also cover algorithmic decisions. But it's not complete by any means.

And internally at Imbue, we have this notion we call lawless spaces, where a space is lawless if there are not many laws, like the Wild West back in the day, where you could like... go steal money from a bank and not get punished. Or you could do whatever you wanted. And that's great for innovation and trying things and freedom. But when it's lawless, then like...

You go into this lawless space, you get bonked on the head, there's no recourse. And so the digital world right now and the internet is kind of a lawless space. Our laws haven't really caught up to the pace of technological improvement. So this is a good example of that. I'm really glad you brought that up because for as long as AI has been used for any number of things like mortgage algorithms, government agencies, facial recognition, pretty much anything, it's...

disproportionately affected vulnerable communities, minorities. It's faced a lot of scrutiny. due to harmful ripple effects. Police use of AI has led to a ton of wrongful arrests. In California, voters rejected a plan. I remember to replace the state's bail system with an algorithm because of concerns it would increase bias, which...

It definitely would have. So it is interesting to think about lawless spaces and how regulation and rules just haven't quite caught up with this technology yet. And I think a lot of us are paying the price.

That's exactly right. This may seem obvious now, but why do you think we're seeing AI injected into areas like hiring and recruiting in the first place? Is this just one of those areas where people want to solve a problem that doesn't really exist and maybe AI would be better used? not in this area at all? Or do you think there's some value to it? I think AI is so fundamental, a technology, it is intelligence, that it will be in every space. It will shape everything, how we do everything.

in the digital world and maybe even sometimes in the physical world with robotics. Fundamentally, AI is trying to replicate human intelligence. as close as we can. And right now, as a society, we use humans to do lots of things, right? Like humans are useful means of production for society economically.

And so because of that, AI will also become, will be one of the major means of production in the future, AI and software. I use them interchangeably because AI just is software. And so what that means is that when we think about... the future that we're building, AI is going to reshape business decision-making, human decision-making. It's going to be involved in all of this stuff. And because of that, we actually need to rethink a lot of things.

When we just built software, software like automates some processes and it lets us communicate with people and like sends information across the world and things like that. But AI is not just. sending information or automating like really hard-coded things. It's like it is intelligence. It's replicating intelligence. And so when it comes to

not just our laws, but our infrastructure for how software works, how the internet works, like our expectations of how all this stuff works. I actually think there needs to be a lot of things that change in order for us to get the world that we want and not be victims of this. technology. And so like one example of a type of thing earlier we talked about, I talked a little bit about, oh, software should be modifiable at the point of use.

What does that mean? It means like AI and software needs to be both explainable and controllable by the people it affects. And to this point about like recruiting decision making, like I need to be able to. like know how these decisions are being made and I need to be able to change how these decisions are being made. A simpler example that we all experience day to day is notifications or our social media feeds.

We neither know how those decisions are made about what notifications we get, nor do we have any power to change the feeds except by upvoting and downvoting things and tossing stuff into a black box. And so as a result, we don't end up getting the decisions that we wanted from these algorithms. I always say, like, social media feeds are the first AI agent. And it is a runaway AI agent in a way.

changing our lives and making decisions on our behalf in a way that isn't necessarily what we want. So I actually think going into the future, it's important for most software to be open source. for each individual person to be able to take the source code and to use an agent to modify it.

using natural language in a way that feels more intuitive to us. And at Imbue, it's something that we're working on. We call it common source. Right now, we have open source where, like, people, volunteers make stuff for free. Other side, we have closed source where enterprises use all of this free software that has been built by volunteers and make money off of it, but don't pay the volunteers.

And there has to be something in the middle where you can ship your software to other people, let it be open source so that the users, your users can change it, but you still get to make money off of it. But that doesn't exist today. And so this is an example of like...

needing to reshape the technical infrastructure, the payments infrastructure, the economic infrastructure and policy infrastructure that that like operates our world today to make it suitable for a world in which AI exists so that we can live the lives we want. That's really interesting about common source. So we'll have to chat about that later. Back to the resume thing. It's obviously pretty clear how this can go wrong. AI can come with...

and does come with actually always, all sorts of implicit bias baked in when screening job candidates and considering labor and discrimination laws. I even remember years ago, Amazon scrapped one of its tools that... was trained to vet applicants by observing patterns and resumes submitted over a 10-year period, and eventually it taught itself that male candidates were preferable. It penalized resumes that had the word women's in it, like I think women's colleges.

or certain clubs. They eventually, of course, scrapped the system, but who knows how many ripple effects it had before then. And so I think some companies use these types of tools because it can... help with bias sometimes. A lot of times it'll anonymize names or other aspects of a resume to ideally put people on an equal playing field. But in other ways, it's perpetuating a ton of the biases we see every day. So what do you make of the...

potential benefits and the potential pitfalls of this technology being used in hiring. Like you mentioned, it's kind of a... inevitable, I guess. It's in every sector. But yeah, it just seems like there's a lot of potential obstacles that we've been seeing for like 10 years. I think this is part of why it's important for these systems to be more explainable and controllable.

where ideally you could like see what's happening in the decisions and like be able to change it. Today, it's kind of like a black box with some reinforcement loops. Like you're like, yes, no, yes, no. And then you don't know what the black box is learning and what it's starting to index on.

I think more general models might actually solve some of this problem. Like the more general the model is, the more it understands about the world, the more it understands like, hey, we shouldn't be optimizing for this like. We shouldn't be optimizing against minorities. Or, oh, maybe we could flag if we are ending up optimizing against minorities. Or maybe we could analyze the data at the end and take a look and see what's happening. And so...

I actually think that we can use these models to make much better. The optimistic take is that we can use these models because they give us a lot more intelligence capability to analyze the effects of what's going on in, say, hiring.

and what's going on with these algorithms and to change them to our liking. And this is kind of a world, you know, I imagine where we are controlling the AI and we are directing it to create the kind of outcomes that we want to see that's more in line with our intentions. I don't think anyone...

intending to bias against women at Amazon, but they couldn't see it very well. And so that ended up happening. But we could use AI to have better monitoring, to become wiser, really. And that's really the opportunity here. That makes a lot of sense, actually. Sometimes companies will...

I've heard automatically downvote or disqualify someone if they do find secret instructions for a chatbot hidden in someone's resume or a white font of skills listed like at the bottom, only meant to be seen by AI.

recruiting system. Do you think it's fair for... companies or recruiters to look down on people that use those types of tactics when at the same time they're using similar tactics to screen applicants? I think we're in a game where no one's winning right now. It's like this war of attrition.

more information into the model or does the model like catch that job seekers fake information faster ultimately i think we need where the value is is to get out of this war of attrition and it's hard to know how to do that you know part of what's happening is you know there aren't that many like it's hard to find jobs right now

Maybe people aren't upskilled in the right ways. And so maybe we could direct more of our energy and attention toward upskilling people and trying to teach them how to have the skills that are needed. And with AI systems. There are a lot of creative ways to use AI systems to learn really fast, learn new things really fast. This also goes to how we think about AI, where instead of letting AI automate us out of a job, somebody else can automate us out of a job.

a job today, there is an alternate world where we can use AI to automate ourselves out of our jobs. And that actually feels a lot better and is awesome. I make little scripts and agents to automate myself out of some of the processes that I have to do every day. What that means is a world in which we get all of these tiny AI systems that are embedded throughout our lives that we're creating and we're controlling those workflows. And so then we are the ones who get to...

decide like, oh, do I want to automate this? Do I want to change this in this way? And in that way, like AI can become more an expression of who we are and what we're trying to do, as opposed to like a centralized thing that's like someone else is using to automate us. So I think I'm of the philosophy that this war, this job seeker-recruiter fight, or we're going to see these fights in all sorts of different places. These fights are, they're just going to happen.

And the way to get around them is to try to solve the underlying problems more and to kind of empower each individual to have more opportunity and to be able to express themselves and to have more economic opportunity in their own way.

by empowering them with AI. Yeah, I think we are going to see this fight happen a lot. And one last thing I wanted to ask you about this situation is another kind of... area we're seeing this tension happen, which is AI interviews. I've had multiple friends talk to me about their experiences with AI interviews, and the New York Times story included that too. It seems to be an increasing trend to screen people. using ai basically it's like a recording

I used to be an actress, and it's kind of like recording a self-tape in a way when you're sending in a video of yourself for an audition. But instead of, you know, you just talking to yourself, you're talking to an AI. system that is asking you questions and then weirdly sometimes affirming you afterward. So yeah, what do you make of that trend? It's pretty crazy. Yeah, I think this is a trend in the...

in the category of AI exacerbating existing power dynamics. Right. Where I, as the applicant, have a lot less power than the recruiter or the company that is potentially going to hire me. And so it does feel, you know, when we talk about fairness, like what is it that feels unfair? What feels unfair is people who are already disempowered being more disempowered. And this kind of automated video screening, like.

That takes less time on the company side, but it takes just as much time on the applicant side to go through every AI video company screening. And so in this way, there's this power differential where... As a person who's not, able to use AI in my application. If I need to record myself as a human and I can't use AI, then I can't scale myself in the way that the company was able to scale itself. And that's why I keep going back to like decentralizing, enabling people to create their own.

AI agents, enabling people to create software, common source. All of these things are about how do we give people the ability to scale themselves and use AI systems to give them more power. There are already enough of these power dynamic differentials in the world. And we're already in an increasingly lopsided power differential curve.

We need efforts that push against that and really try to flatten the power dynamics because the fundamental thing that AI does and will do is it will give you power. It's like a way, it's a source of power. It scales a bunch of things. It automates a bunch of things. It understands a bunch of things. It can process tons of information and can take actions. Those are literal sources of power. And because of that.

Like, what we need is a way to give more of that power to people who have less power today. to kind of equalize the power dynamic. And there's an opportunity. I think we have this like window of opportunity to do that today before we end up with like lots of lock-in again, just like as happened the last 10 years.

That's fundamentally, I think, the thing that needs to happen in order for us to see, in order for us to stop feeling so weird about what's happening. I think the weirdness that you're feeling. is real. It's pointing at a real moral dilemma that feels really strange. This doesn't feel fair. There's something ethically weird going on. And I think those are real signals.

that we're getting. And those signals point at this power differential problem. Right. Yeah, I really, I'm so glad you put it that way about the power dynamics, because, you know, that's, when people ask me what I focus on as an AI reporter. You know, six years ago when I started on the beat, I could just say I was an AI reporter and that was niche enough. But now everyone's an AI reporter. It's kind of the same as saying you're a business reporter.

Everything is business. Everything is AI. Now I say that I focus a lot on the shifting power dynamics within both the industry and how these systems and the people that lead them relate to society and the people that... systems are affecting. So yeah, it's definitely going to be a more and more important and more talked about topic as the months and years.

go on. But yeah, thank you so much for your time and for coming on. I'm really glad we got to talk about all this. Yeah, me too. I'm glad we got to talk about the power dynamics. I guess one more thing on that, on the power dynamics, I'm really glad that that's what you're thinking about.

And I think that often people think about the power dynamics in terms of, oh, what laws can we enact?

But I actually think that we can build the technology in ways that are fundamentally more empowering and fundamentally less empowering, just like the supercomputer analogy I gave before. And so when we were talking about OpenAI Dev Day and these platforms and Sora and what's happening... I think that there are things that we can change about how we build these technologies and the environment that they're released into.

that would actually distribute power and change the power dynamics. And so that's, I think, really the opportunity there, but we have to get creative. Thanks for coming on. I'm really glad we got to have this talk. Yeah, same. I'd like to thank Ken Joon for taking time to speak with me and thank you for tuning in. I hope you enjoyed this episode.

If you'd like to let us know what you thought about this show or what else you'd like us to cover, drop us a line. You can email us at decoder at the verge. We really do read every email or hit me up directly on X blue sky or threads. I'm at Hayden field on all platforms.

Decoder also has a TikTok and Instagram and now also a YouTube channel. Check those out at DecoderPod. They're a blast. If you like Decoder, please share it with your friends and subscribe wherever you get your podcasts. And if you haven't already, please consider subscribing to The Verge.

Decoder is a production of The Verge and is part of the Vox Media Podcast Network. Our producers are Kate Cox and Nick Statt. This episode was edited by Xander Adams. The Decoder music is by Breakmaster Cylinder. See you next time.