If you or your doctor is not using AI models as a second opinion for anything medical, it's really kind of malpractice. There are still these major diseases that are killers. AI is basically almost like a superpowered search engine. And part of what you're searching for is what's the right molecular structures that can allow us to do cures for these things that are fatal.
What AI tools do you use daily that are actually increasing Reid Hoffman's agency? ChatGPT, Copilot, Gemini, and Claude. AI now can not only do the translation but it can also do the language generation, put the language in your voice. Like I've listened to myself speaking Hindi, literally don't know a single word in Hindi.
What is your most contrarian view on the future of AI?
This is going to be very surprising, but essentially even without a lot of intervention, AI will actually help humanity. It is not an existential risk. It is a humanity elevator.
I'm Sammy Cohen and this is Social Currency, a podcast that unpacks stories that are shaping business, culture, and the intersection of the two. Before we dive in, I want to ask for your help. The show is growing very quickly, and my next milestone is subscribers. So, if you're not subscribed wherever you're listening, please take a second and do that now because I don't want you to miss any conversations like the one that you are about to hear. Today, I'm talking to someone who has quite literally shaped how the modern professional world works and someone that I've followed closely for years. Reid Hoffman is the co-founder of many influential companies including LinkedIn and the drug discovery company called Manas AI. Reid is also a legendary investor, podcaster, and author. His most recent book is called Super Agency, where he argues that AI isn't just about productivity, it's about expanding human agency. And as an aside, Reid sent me a copy of this book with a personalized book jacket. So, if you're watching this on YouTube or Spotify, I'll show you the visual because it is easily the coolest book branding that I have ever seen. And it's also a meta look at the way that Reid uses AI to expand his own agency. So, in our conversation, we get tactical about how to navigate the risks and opportunities around AI, like what to do if you're worried AI will replace your job, and the three AI skills that everyone should be learning right now to stay ahead. We also get into the AI risks that exist in today's ever-changing tech landscape. His take on Claude and whether or not digital twins will replace human creators someday. Without further ado, this is Reid Hoffman, Social Currency. Reid Hoffman, welcome to Social Currency.
Reid, my whole brand is built around telling the hottest business and tech lore, and I'm so excited to sit down with you because honestly, your career is filled with the top tier level tech and business lore. So, this is a dream conversation for me and we're going to talk a lot about AI, where AI is going today. But first, everyone talks about how AI is making us more productive. But in your book, Super Agency, you frame it as AI expanding human agency. So what's the difference?
Part of the reason I wrote Super Agency is because when you look at a lot of the worries, the fears, the concerns, the rejection of AI, it actually comes down to concerns about human agency. It's like loss of a job or transformation of a job, loss of agency. Tech companies doing something with my data that's outside of my control, loss of agency. Questions around like the tech companies changing industries that I'm in, loss of agency. You even have what we Silicon Valley people and other people call existential risk, you know, like loss of agency. And yet what I think the real truth of the matter is with AI, inclusive of when you look at this dialogue which happens every time we have major new technology, it's actually going to result in major increase in agency. Now it's a transformation and people tend to say no, no, I'm happy where I am. And it's like, well, yeah, no, the people who were working in the horse industry before the car were happy where they were but we're changing to a car industry, right? Like here we are in LA, there are no horses on the streets, there are no stables, and that's because that was the transport mechanism back then for cities and local. So that's changing and so you go, but I don't want it to change, that's a loss of agency. Yes, that's a transformation but if you then say, okay, I'm leaping from the horse to the car and I can take agency. Agency is in part a mindset. It's the okay, I'm not going to go boohoo, it's no longer the age of horses. It's I'm going to go, I get it. It's the age of cars. What should I learn? How do I do that? How do I participate? How do I benefit from it? How do I benefit my family, my community, etc.? What do I do? And that's when you can begin to take agency. And part of the super agency is when the real superpowers unlock is when lots of people all get the agency superpowers. So, for example, without cars, I can drive places. I have freedom of mobility. But it's not just that. Like say for example your parent or your child is ill, can't move from your house, well now the doctor can drive to you, right? That's also an increase in agency. So it's not just what you can do with AI but for example today if you or your doctor is not using AI models as a second opinion for anything medical, it's really kind of malpractice. You really should be doing it today, both of you. But that's not just you, oh I can check AI and I can say, well, is the doctor giving me the right advice? But so can the doctor.
And that adds to the super agency part of it.
And so much of what you're saying, the fact that people can use this agency to enhance their lives in so many ways. Some of the terms that you use in your book, you say that you are a bloomer rather than a doomer. And you've also seen the risks of AI firsthand, though. Can you walk me through a specific moment where you had to reconcile your optimism with AI with maybe a genuine AI risk that concerned you?
Well, I have a number of AI risks that concern me. That doesn't necessarily mean that I think it's realistic or good to particularly slow down. Some of it is in say for example bioterrorism, especially in too many open source models that can be run in a bunch of different places could enable someone, like for example we're in LA. Maybe people have seen the movie 12 Monkeys, you know, enable some kind of very bad kind of bioterrorist kind of activity. There's cybersecurity, like we already have an issue where people are more vulnerable to kind of cyber hacking and economic and other kinds of issues. So that gets amplified. Those are kind of real things. So there are real risks here where the transformation brings new risks. Now I also think that AI can help protect against those too, right? So, like for example, you get a phone call. It sounds like it's your sister, your brother asking for money and you got to wire it to me right now. I'm in trouble. And then the AI agent comes onto your phone and says, you know, this could be an AI phishing attack. Why don't you ask something that really only you, your brother, or your sister actually really know? Like some memory from childhood or something that isn't really recorded anywhere or something else and just check.
Right, you know, so it could be the defense too.
So there's a whole stack of things that I think are real risks and by the way I think the economic transformation that we're going to get from AI is going to be magnificent in the medium long term, but it's going to be very transition difficult for many people in the short term. And that's another thing that I worry about.
Yeah. And actually speaking on that, there's been countless conversation around the displacement of jobs with AI. And obviously that's one of the main things that people bring up in perhaps more of the fear-mongering camp. One of the items that you've cited and the projections are nearly 170 million new jobs will be created versus 92 million that are displaced. So what would be your advice for someone who finds themselves in that 92 million camp and they want to embrace AI to stay in the job market?
One, sympathy. It's not easy, right? Two, in terms of advice, it's start using AI, right? Too often people go, oh, I don't know. I'm not that good with technology or like I've used an iPhone or I've used a smartphone or I've used a laptop, but I'm not particularly like other people are much much better. I don't know anything about coding or anything. And you're like, no, no, no, the benefit of AI is AI can translate your intentions. Like what you say, speaking is actually much better than typing with AI because it actually gives you a much richer thing. We call this voice pilling as ways of operating and just start using it, right? And whenever you're encountering a problem, say, oh shoot, my, for example one of the industries that I think will probably first hit would be customer service. So you say you're in the customer service industry and shoot, my customer service center jobs are going away. The one that I was working at, what should I do? Oh, there isn't more customer service around here. What should I do? Well, one of the benefits of AI is you can go to AI and say, hey, my customer service job is going away. You can say because of AI. AI doesn't have feelings. That's fine. Help me. What can I do? Right? And an AI will go, well, let's talk about some of your skills. Can I see your LinkedIn profile? Like this kind of thing. And started saying, well, maybe you could be good at sales or maybe you could be good like working for you because formerly was industrial policies like well customer service people were good on the phone so what other kind of phone jobs can there be, right? And it's actually in fact, well, yeah, sure, that's some experience with that and some experience doing that but it's like, oh, actually in fact you might be pretty good at the IT help desk, right? You just need to learn this, this, and this here. Help me. Let me, because for example you can go to an AI and say like, well, do I going to sign up for a class, have to commit for three years? Like, no, no. Ask the AI to give you three to seven YouTube videos that you can look through and then talk about with the AI and then say, oh, actually, I like this or I actually have an affinity for this. Hey, let's go down this path. Or, nah, this didn't really work for me. Let's talk about other things. So, AI can be part of the solution for you. It can be part of the what jobs might there be out there or be growing that I'd be good for? Which things could I possibly learn even as you get into the job? How do I do the job? How do I do it better? Etc. And so, adopting it to help you with your navigation.
Yeah. And I love this idea of the thought partner in really all aspects of your life because I think when LLMs really became big now a few years ago, I think that they were used in perhaps more specific use cases, but really being able to think about creative problem solving using AI is really incredible. And I think one thing that I'd love to hear from you because you speak so much about AI and I would imagine that your workflows with AI must be through the roof. So what AI tools do you use daily that are actually increasing Reid Hoffman's agency?
All of the ones I use increase my agency. I think I most often use some combination of ChatGPT, Copilot, Gemini, and Claude. I do use some other ones as well. For example, when I have either something that's unusual like a creative prompt where I'm looking to kind of establish some kind of creativity, even though you say well on creativity Claude usually a little bit better.
That's what I was going to ask. Do you have your specific usually?
But like for example like recently one of the things that I was doing for a project I was working on was said look it'd be really fun to create some stuff around social media for a number of the strange holidays, right? Like Speak Like a Pirate Day or International Towel Day off Douglas Adams. And so it would be fun to kind of do that and I was like okay I want an index of these. I thought Claude would give me the best index. Gemini did, who knew? Right? And now by the way doing it in all of them, all of them had at least one that the others didn't get that was good. Right? Now, some of them were like, yeah, yeah, fine. You know, like all the standard ones, but it was like the Oh, that's a funny one. Right. And so, like, for example, sneak a zucchini onto your neighbor's porch day. Is that actually a real one? Yes. Right. And you're like, that could be funny. Right. So it's like, you know, so that kind of thing. So basically, anytime that I'm doing something of any seriousness whatsoever, I ask myself the question, should I engage with one or more AI models in order to do it? Right? And I recommend this to everybody. It's like, oh, you're going to have a difficult conversation with your parent. Oh, I'm going to Rome, and what would be fun for me to do in Rome? You necessarily want to do the psychic energy of figuring out what the prompting is or what the agent setup is or anything else for every single thing because sometimes it's like, eh, don't really need to do that. But that's one basis. Then another basis is let's take for example one of the ideas that I've had over a couple years but intensely this last fall was part of how AI can be humanizing. So I do, as you know, a podcast called Possible and one of the things we have is Arya, my co-host, and I do this Reid Riffs thing and I was like well this would be really easy to put in a lot of languages and a lot of languages would be humanizing. And AI now can not only do the translation but it can also do the language generation, put the language in your voice. And so it's like I've listened to myself speaking Hindi, it's very strange. I literally don't know a single word in Hindi. Obviously the media thing of a broader reach but also more humanizing. People really prefer to be spoken to in the languages that they're native in, right? It's the rich fabric that we go into. It's like, oh, we could do this. Now, you think, oh, I just prompt the AI and it doesn't. And you're like, it doesn't actually work. Maybe someday it'll work that way. But to figure it out, we ultimately had like six different agents, each taking a different role, each kind of doing a different part of it, some of them cross-checking some of the work and so forth. And so as of today, we can go from the two of us are talking because we have our voice from ElevenLabs. We have a video thing. We can generate that the next day in 14 other languages, right? But part of that is again the setup of the agents and the agents working as a team. So it isn't just you talking to ChatGPT, you talking to Copilot, you talking to Gemini, it's oh no, they're a set of essentially Claude bots that are taking a role in a production process in order to generate it through. And by the way, to get it right, it still took a bunch of different human interventions. When I first had this study, I was like, well, I'm going over France. Let's do it in French, right? They're leaders in AI and in the EU, so let's do that. So we generated and I called a French friend of mine and said look could you listen to this because I don't speak French so could you listen to this and he called me back and he said well I got good news and bad news and I'm like okay and he's like well the good news is the French was coherent, all the points that you're making I recognize Reid points. The bad news is you sounded like a Canadian which to us French sounds kind of hickish. You need to change that. Very helpful to have the human feedback.
