This transcript is generated with the help of AI and is lightly edited for clarity.
//
ARIA:
America is helping to build this technology, and yet Americans are among the most worried about it.
REID:
So I actually think AI inclusion in the hiring process will broadly be much better.
REID:
You want to return to manufacturing strength in the US? AI and robotics are your only hope.
ARIA:
What do you think changes when the advertising platform is also this assistant that you’re supposed to trust and use to make really important decisions in your life?
//
ARIA:
Reid, so thrilled to be chatting with you today, as always.
REID:
And likewise.
ARIA:
So let’s get right into it: AI optimism. Do people like it? Do they not? We saw that last month, Gallup and Microsoft released findings from a survey of adults in 37 countries. And this was about their awareness, use, and views of AI. And one of their major findings was, among people aware of AI, positive feelings outweighed negative ones in 34 of those countries. That was a big surprise to me. But what was not a surprise was that the US is an exception.
74% of AI-aware Americans feel worried, and only 36% expect AI to mostly help their country. And that’s compared to 93% of people in China who think AI is mostly going to help them. And across countries, the median share who had never used AI is also 57%. So Reid, America is helping to build this technology, and yet Americans are among the most worried about it. What do you think explains this gap? We’re the ones who build it; we have high adoption, and yet we are not excited about it?
REID:
I think there’s three things. One, I think there’s a general—call it concern, fear, worry—that the future is going to be less good than the now, and that AI is just an amplification of that. Like, I think even before AI, there was that worry, partially because I think that, you know, we’ve done a not-great job as a country of bringing, call it, the majority of the country along on our economic prosperity, and how to make sure that—you know, maybe the simplest way of putting it is, make sure we get the middle class a raise in the work and jobs and opportunities they’re doing. I think the second thing is that, generally speaking, when you have technological waves, the people who already, in their own minds, feel like challengers versus incumbents are benefited from that.
So it’s kind of not surprising when you go to China, which is, you know, growing its economics and feels that it’s, you know, kind of coming on strong to create, you know, a whole number of industries, some of which already very successfully. But even, for example, when you get to the Global South, the primary thing that they request around AI is, make sure we’re included. We don’t want to have this be economic prosperity for other people and not for us, so make sure we’re included. But, you know, since the US has been the richest country in the world, it tends to have more of an incumbent perspective.
Even though historically, you know, we psychologically identify more with the challenger in terms of, you know, growing our industries, growing a new country, et cetera, et cetera. But I think people are like, well, we don’t want to lose what we have. So I think that’s the second part of the American psychology. And then the third is that we have—and this is kind of classic when you get to the kind of shadow ages of empires, whether it’s Rome or England or now for us Americans—there’s a lot more internal dispute. It’s like, well, if there’s going to be an AI, you know, boon, who is it going to benefit? Is it going to benefit me or benefit you?
And you’re not focused on the US as a, you know, kind of defining country within the world. It’s more like, you know, my political tribe versus your political tribe, my region versus your region, my, you know, racial identity versus your racial identity. And so I think there’s a lot of internal grain. And so I think all of that sets the stage for most people being inherently somewhat skeptical. So, for example, a lot of people are, you know, kind of anti-big tech companies and big tech, you know, CEOs and creators anyway, and they’re like, this is just another reason to be anti. And even when you say, well, wait a minute, this could create a medical assistant that could, you know, essentially create high-quality health care across, you know, the entire scope of the world—
But also, you know, 340 million Americans, of which only something substantially less than 100 million have really good, high-quality health care. Yeah, right. And even if you could do that, they say, well, no, but that doesn’t matter, because it’ll be benefiting, you know, the tech companies and the owners of the tech companies more than it is us. And you’re like, well, that’s the wrong place to be. You should be thinking about what are the things that kind of raise the capabilities for everyone. And that’s, I think, the reason why—since the other countries don’t have these internal divisions, they’re clearheaded about the fact that this is going to create the next generation of globally relevant prosperity that they want to be bought into.
And so, for example, the most surprising part of these findings is, you know, only 36% expect AI to mostly help their country. It’s like, well, that’s an obvious thing in terms of economic productivity. Like, to say, hey, I worry about the reshuffle of power economics within the country, is, I think, something to say, how do we navigate that? But the creation of prosperity for the countries that adopt is actually, in fact, really straightforward. And that’s the reason why you see the 93% in China. And I think, you know, there’s a number of other factors that come into it. I think the Chinese government, you know, kind of understands—just as they’ve been trying to raise their pharma industry and the car industry and a bunch of other things—they understand this is an economic multiplier.
It’s part of the reason why, a number of years ago, they said they want to be the leader globally in AI by 2030. That’s just around the corner in terms of what they’re trying to target. And so I think, you know, for us, that’s part of the reason why, you know, a couple weeks ago, I came out kind of arguing for a dual mandate, which is we both need to win at AI, because that will give us as a society the economic prosperity to do everything from trying to fund the fact that we are massively in debt—we’re like a society that’s maxed out all our credit cards, and a huge percentage of our taxes are going to paying off the debt.
And this gives us the growth to navigate that and create economics. And then the second is to use that—you know, to earn the money by being, you know, first in AI—to then benefit the bulk of the country. To say, how is it that, in building AI, it isn’t only going to be a beneficiary to one industry or to a small number of companies, but to many of the industries, to many of the workers?
And some of that you can do through creating assistants—medical assistant, legal assistant, educational assistant, government services—and some of it you do by supporting the industries. Like, you know, to close off this opening monologue—sorry for the length—is to say, look, you want a return to manufacturing as a strength in the US? AI and robotics are your only hope. And the question would be, how do we deploy our strengths in AI? We’re actually behind the Chinese on robotics. So, increased robotics to return us to a manufacturing powerhouse, or at least in contention with China, Japan, Germany, et cetera. And I think that would be obviously very good for the middle class if we succeed in doing so.
ARIA:
So it’s very clear that there is a lot of work to be done, both by the companies themselves, but also by the American people, on using AI to help make their own lives better, but also understanding how it can sort of keep the US as an economic powerhouse as we head into the future. I’d say one of the places where people are worried is jobs. And specifically, I see headlines like, AI is writing resumes and cover letters, you know, thousands a day, and AI agents are rejecting thousands of people a day. It adds to both the nervousness, but also the dehumanization that people are feeling. Like, I’m putting myself out there, I’m trying to get this job, but, you know, I just have a robot on the other side, or I have a machine telling me no.
And so HackerRank is a technical hiring platform, and it’s used by thousands of companies, and it’s expanding its AI interviewer, Chakra. And so this is actually an AI that is doing the interviewing. I think the interesting thing here, though, is what it tests. Candidates can work through practical problems with AI, and then they face questions about their choices, their reasoning, whether they check the AI’s work. This company had been known for straight coding assessments, but this signals a broader change in how developers prove their ability. Employers increasingly need to assess how someone directs and evaluates AI. Like we’ve said so many times, in the future—or maybe the present—everyone is going to be a manager, because you’re going to be managing AI.
And employers need to know how their potential employees are going to do that. And so you and I both have told young people, including my children, that being AI-native is going to be part of their opportunity. How do you think hiring should change so that employers can actually recognize that talent, but at the same time, we need to make sure we’re not sort of dehumanizing the entire experience?
REID:
Well, I mean, the hiring experience is already somewhat less than ideal, so it’s not, you know, this kind of perfect experience. And actually, there’s various ways in which I hope that the transformation of the hiring experience through AI will actually, in fact, elevate the various elements of humanity. Now, one of them is a project you and I work on together, along with Byron Auguste and all the people at Opportunity@Work, which is this notion of Tear the Paper Ceiling, and the fact that just having a piece of sheepskin is not actually, in fact, the right credential. And so it creates a differentiation, and also this dependency on going massively into debt to go to some colleges in order to do it.
Whereas actually, in fact, you could just, you know, use the skills and capabilities. And I think one of the things that you get in an AI interviewer is—currently, the use of a human interviewer is very expensive, off the things that they’re doing. And so they tend to go, well, let’s just narrow the channel as much as possible. So, do you have a bachelor’s? Like, we’re interviewing you to become a janitor—do you have a bachelor’s degree? Right, you know, kind of, you know, in the absurdity of it. But even, for example, you don’t need a bachelor’s degree to necessarily be a software developer. You don’t need a bachelor’s degree to necessarily, you know, be any number of other kind of white-collar jobs.
Sure, you should be certified if you’re an accountant, because you have to understand that kind of thing. But, you know, part of, you know, kind of an AI interviewer is you go, well, so-and-so looks interesting. It’s not that expensive to run a two-hour AI interview process with them, and you could get a really interesting assessment. And what’s more, you’ll get a better assessment, necessarily, than simply a, oh, look, you know, does the chemistry, you know, feel right in a, you know, hour interview where you’re kind of going through stuff, because you can also do four hours with AI, and you could have it do it in chunks and follow up and all the rest.
So I think there’s a lot of things you could do there that enable a much broader range of talent that you can potentially find. You know, kind of the diamonds that are out there in the field, not just in the rough. I think there’s also, you know, kind of questions around AI where, you know, like, I think that you will still have human interviewing, but I think that, you know, the thing you’ll do is say, okay, the AI will do a bunch of stuff on competence and capabilities and learning curves and other things. And then the human will assess, you know, kind of cultural fit, team fit, you know, other kinds of things, and then cross-check some of the stuff. And I think that will make a good experience.
And then, of course, also, you know, one of the things that an AI can also then spend time doing is, like, helping the candidate know whether or not this particular job or company culture or path is right for them, because it could also spend time answering and selling as a way of doing it, because it’s a better—it’s a batch. So I actually think AI inclusion in the hiring process will broadly be much better. Now, I think people will kind of go, whoa, I’m worried that—like, before, I was completely like, could I get an interview or an informational interview, and could I use, you know, a really clever cover letter or a college opportunity or a network connection to make those happen? And I worry about getting further away.
I think those connections will still exist. So I think those are there. But I think that the question around—as opposed to, I just can’t get anyone to respond to me, you can now get some response and possibly some feedback. So I’m overall very bullish, and not a surprise to you or the people who’ve read Superagency. But I think when this becomes more common, a little bit like when cars became more common, I think it’ll elevate the capabilities all around the ecosystem, both on the job seeker and on the hiring side. And it’ll make it more transparent, engaged, closer to the truth, talent-wise. So overall, very positive.
Hey, y’all. Masters of Scale Summit returns October 20 to 22 in San Francisco. If you’ve joined me at Summit before, you know how different it feels from other business or tech conferences. Van Jones calls it “the highest concentration of heart with smart you’re ever going to find.” Every element—the lineup of speakers, the curated audience of founders—is designed to spark ideas that transform your business. You’ll see people you know from Possible—Reed Hastings, Amjad Masad, Kara Swisher—take the Summit stage live, alongside leaders like Nobel laureate Maria Ressa, Walmart’s John Furner, and Cloudflare’s Matthew Prince. Be in the room this year and walk away feeling it’s possible to scale your next great idea. Join us. Visit MastersofScale.com/Summit to learn more.
ARIA:
I mean, we’ve talked about this before. There’s already so much bias, discrimination, whatever, built into the hiring process. To your point, do you have, you know, chemistry with the person you’re interviewing? There’s certain jobs that are much better suited for in-person interviews, because those are the skills you need for the job, whereas for others, those aren’t the skills that you need for the job. So how could this be the job? I mean, it was funny—last week, I was talking to someone who’s sort of in the job search process, and they had been on 20, you know, first-round interviews, and they were kicking themselves, because we were like, oh, wait, obviously you recorded them and asked AI for feedback after each interview. And she was like, ah, no.
And so, to your point, there are definitely ways that AI can help you do this. But on the other hand, I was talking to someone who said, you know, forget AI, I would never hire someone that I hadn’t seen face-to-face. Like, we do, you know, Zoom interviews or whatever for the first rounds, but before hiring them, I have to see them in person, I have to have that. Is there any job today, Reid, that you would say the AI can do it all? Like, we don’t need a human in the loop—these are the jobs that we can just hire with AI, and we should be good to go. Maybe a really fast-scaling startup? Is there a world where that could happen now? Or what would we need for it to be true in the future?
REID:
Well, I don’t think—I mean, obviously, there’s some already incredibly amazing capabilities with the AIs, which are very helpful and therefore an unabridged positive. I think there’s a lot of context awareness, and there’s still a little bit of a, you know, kind of almost like—you know, one of my favorite expressions is, in theory, there is no difference between theory and practice. It’s almost like an evocation of the theory. Like, it’s kind of like, what would I expect the hiring practices to be? You know, the kind of thing that a mid-level MBA course would have as kind of a checklist, which is not bad, but by no means what you really want to be doing.
And you’re not just hiring for an individual and their capability, most often, but for a kind of part of a team. And so, you know, part of interviewing is not just that cultural fit, but also it’s kind of that set of agreements of, oh, yeah, we as a team think that this person is good, and we as a team want to see them be successful, and we as a team are buying into co-owning this process. And so it isn’t just like—you know, when you have a company, as, you know—DoSomething and all the other things you’ve done—when you have a company, you want the hiring process to be a team process.
ARIA:
Totally.
REID:
You don’t just bring in Sarah or Bob and say, hey, everybody, meet your new colleague. You know, it’s like, even if you say, well, we really think Sarah or Bob will be really good, and, you know, we’re cross-checking with you—you don’t need to do a full interview. Sometimes you do that, but you still do that. And so I don’t think there’s any place yet where you would hand it completely to the AI, except maybe—and this is a very limited case—say you’re doing something kind of the equivalent of project work, and you’re hiring essentially a contractor for a particular task, and literally all you want is the task output back. You might have AI hiring, deploying, integrating only with that.
But as an ongoing team relationship, I don’t think—you know, it may be never. Not just a question of AI capability, just on these kinds of principles of forming the cultures of teams.
ARIA:
I love it. Human in the loop, perhaps forever. All right, so we already touched on jobs, which is making people fearful. Now we’re going to move to advertising and trust, which I think is another place where people have a lot of differing opinions. So some people say that advertising is the greatest—it creates the greatest business models the world has ever seen. It allows many consumers to not pay. Other people say, you know, this is why social media became so predatory, because it wanted eyeballs, and we can no longer trust it. And so OpenAI announced it’ll be testing visual ads during ChatGPT image generation later this month, and they’re going to be starting in the US with users on its Free and Go plans.
And these ads, they say, are going to be clearly labeled. It’s going to be separate from the image being created. The advertising, they say, won’t influence ChatGPT’s answers, and they also argue that this will make it more accessible. Obviously, a lot of people are worried that if you have this trusted AI assistant that you’re asking important, you know, intimate questions, that all of a sudden it has sort of a vested interest to give you different answers based on advertising interests. What do you think changes when the advertising platform is also this assistant that you’re supposed to trust and use to make really important decisions in your life?
REID:
So, a couple things. One is, I do think that there will be some paid, non-advertising things, and if you have a particular worry about that, you’ll be able to do that, whether it’s from OpenAI or from others. And so I think there’s a range of choice that gets there. The second thing is, you know, I think one of the reasons why, you know, Google so far has created the best, you know, call it business model yet invented in human history—or at least the best media and so forth—is part of what they did with AdWords was it said, this is the advertising thing, this is the non-advertising thing. Right? And then made the advertising thing quality enough that people would also treat it most often as an alternative class of search result. Like, I actually know people who, on certain kinds of searches, specifically do the search and then look at AdWords, thinking, I’m looking for a commercial answer to this particular query, and the intersection of the dollars and the click-through actually makes that a good search result.
So I think that the key principle here will be to keep that—like, this is advertising, this is not advertising—and that the non-advertising isn’t being influenced the way that you see in other things. Like, for example, you know, when you’re searching for stuff on Amazon and you see it’s sponsored, then surprisingly, that sponsored is also very high in the search results, and so you wonder what the blend is. You have to keep them, you know, quite distinct in terms of how this is playing. And then the third, I think, is not just the kind of transparency, which is the second point, but also kind of making a contractually binding commitment that you could be held accountable to.
Like, you know, for example, the search quality team at Google is completely distinct and says, we don’t have anything like that ad space and everything else—we don’t look at that at all in how we’re making our search results. And I think you need to have some kind of commitment like that within it, because when you hold those things—exactly what OpenAI—and it’s part of the reason why I think people strongly prefer to have a free search with Google and Bing and then navigate the ads, because they’d like to. The free opens it up to lots more people, by the universality. It also creates a—like, it’s just kind of like, you know, these kinds of technology projects are best when at massive volume. So as opposed to all the nuances—
And this is, you know, like—in a separate context, they made this argument of, you know, the CEO of Apple and the Uber driver have the same iPhone, broadly. Right? Because it’s made for a broad mass-market audience, and sure, maybe the Uber driver’s got the $800 version and the CEO has the $1,600 version, but they’re largely the same thing. And so I think that’s one of the things about having a very universal product, and it’s one of the reasons why having a free AI product with advertising as the economic model creates that universality. That’s very important. But of course, you need to keep to these principles of integrity, which is the reason why I started there.
ARIA:
And how do you think—like, how do you get consumers to trust that, or who should be the governing body? I mean, I want to know, if I’m like, hey, you know, everything you know about my son, what is, you know, the best college for him? And then, you know, I find out that Middlebury is, you know, paying ChatGPT however much a month for advertising. How do we get the consumer trust? Is it just reps over time? Just, the more I use ChatGPT or Claude or whomever, I’m more excited about it? Is it that, again, the government has weighed in? How do we create this ecosystem?
Because I think that, again, the dream is—I mean, I will say, I saw this fantastic tweet yesterday that said that so far—this is a little uncouth—agents are so far being people’s husbands. They do discrete tasks that are fine, but we really need someone to be a wife. We need an agent who thinks of the whole picture, who knows everything about me, and who can plan my whole calendar. And so in order for these agents to really be people’s wives—which is what I’m looking for—they need to know so much information. Like, how do we bridge that trust gap? Because as we talked about earlier, the trust isn’t there right now in the US.
REID:
By the way, I don’t think it’s uncouth. It’s saying that wives have a higher trust dependency than husbands. And I think that’s a—yeah, they just do more.
ARIA:
They’re better. I’m just saying wives are better. But yes, continue.
REID:
So, and I think that—look, the way I would start is, I’d say companies should make assertions that they are held accountable to about how they operate. You know, everything from press to, if necessary, government inquiry can see if they are actually, in fact, holding themselves accountable to those assertions. If the assertions are too weak, then you put pressure on them, first by customers selecting other things, by press, by, you know, critique, possibly by government inquiry. And if that doesn’t work, then you start stepping up regulation to say, you have to do the following things.
And as you know, if I was stepping up regulation, the way I would step it up—and it’s something I’ve been very positive on for AI for a while—is, how do you have essentially auditing of it by third-party entities, where that auditing is then available to government scrutiny, is then available to challenge and to, you know, verification? That’s working, because when you have a third party—the same reason we have financial auditors—when you have a third party who doesn’t have an incentive to lie here, because, you know, this is their business, then you have a much higher level of reliability and truth.
ARIA:
Reid, that makes so much sense. Thanks so much for being here today. Appreciate it.
REID:
Always a pleasure.
Possible is produced by Palette Media. It’s hosted by Aria Finger and me, Reid Hoffman. Our showrunner is Shaun Young. Possible is produced by Thanasi Dilos, Katie Sanders, Spencer Strasmore, Yimu Xiu, Aman Suri, Danny Garrison, Trent Barboza, and Tafadzwa Nemarundwe.
ARIA:
Special thanks to Surya Yalamanchili, Saida Sapieva, Ian Alas, Greg Beato, Parth Patil, and Ben Relles.