Making sense of the layoff wave

0:00Reed, it is lovely to be back here. For anyone who hasn't listened to Parth on Read Riffs for the past three weeks, please go back and listen because they were so fantastic. And unfortunately, we're going to start with a not positive topic. So last month was the worst January for job cut announcements since the Great Recession. U.S. companies announced 108 ,000 cuts. It was 120 % increase year over year. And hiring intentions also fell 13 % from last year, hitting their lowest level since 2009. So there are some not great things happening in the workforce. And almost half of the January reductions were linked to just three companies, Amazon, UPS, and Dow.

0:42Some people are saying that AI is to blame. They've been talking about white collar job cuts for a long time. They're saying that this is the fruits of that labor. Some economists say this isn't so clear. Many of the people who are laid off were saying, no, I was using AI. I was on the front lines and I still got laid off. My question for you is, what should we make of this continued layoff news? Do we think that AI was behind it, or are they just being branded that way? I've asked a lot of different people who have a, call it, industry-wide views, is if they're actually seeing any data yet as to layoffs or job replacement from AI.

1:25And it isn't to say it can't happen and will happen, in fact, and in which ways and so forth. But literally everyone I've talked to who isn't just kind of trying to spout a line in order to get a lot of media attention has said no evidence has yet. And you say, well, why are there layoffs and why are there challenges? Well, there's a lot of things. There's the kind of a whole bunch of weird hiring and organizational things from COVID that are getting refactored. And when those are getting refactored, you A, refactor in layoffs. You also have less, you know, kind of entry job level, you know, hiring.

2:06Others are living in an extremely turbulent time for businesses, for businesses to invest. They need to have predictability. you know, kind of the tariff rollercoaster of what tariffs will be when and so forth. You know, I literally know of CEOs who have said, if I could sell my U.S. business unit, I will because I can't deal with this rollercoaster anymore. And so, you know, there's all of that factoring in as well. Now, like I said, this isn't saying that there couldn't be, you know, in 2026 some things that actually really start on that you know in you know white collar things customer service would tend to be the one that i think you'd really see first be one of the most likely places of data i've done enough work that i actually have seen of kind of startups and small businesses that they're like well we used to have three people doing this now we have one But in every case that I've talked to them and it's been more like, oh, it isn't that they fired the other two.

3:10It's like, well, now these other two can do all this other work that we need to have happen. So one of the things that usually happens in a small business is, well, shit, we've got all kinds of ways that we can grow the business. We actually want these people doing this stuff now. And most notably, actually, by the way, entertainingly, where I've had the deepest conversations of those are in the podcasting business. Where, you know, what are we doing for audio management and all the rest? and the kinds of things that you and I are doing with these read riffs of putting them out in all these different languages, that's actually, in fact, something that we wouldn't have done if we didn't have it.

3:46We could do. Now, you'd say, well, if you didn't have AI and you wanted to do it, you would have had to hire a bunch of people. It's like, yes, then it would not have been economically feasible and management feasible, and we wouldn't have done it. So again, it's not a job loss for something that otherwise would have been hired, you know, kind of point of view. Now, what I do think is that we will continue to see from lots of different reasons, the top thing is that people are pointing to AI for anything that happens. I think it's because, you know, hey, it's like, like, here's a parallel. Usually what happens is you get a market correction and then, you know, it's very rare that one company is doing layoffs.

4:32If one company is doing layoffs, then that means that company really was kind of out over its keys and having problems. One of the odd things you can notice is that you go, oh, we have a layoff season. And then all of a sudden, a bunch of companies that don't economically, like aren't economically compelled to do a layoff, e.g., we're going into the red, we can chart when we're going out of business, revenue is collapsing or something else. But they do a layoff then is because they can then do a layoff where they don't look weak. Because most times a layoff signals, my business is in trouble. You as an investor might want to run for the hills.

5:16It's a challenging time. So companies only want to do layoffs when they're away from the pack when they absolutely have to. They will do them and it may be a very good thing for them to do. This is not saying it's cruel or anything else, but it's like, well, action factor. This whole division that we've been investing in, action factor is not going to work and we should really refactor. It's part of what makes business efficient. Then you have a layoff season. Now, of course, we have a new layoff demon. It's an AI layoff demon. And that AI layoff demon is like, oh, it's because my business is strong.

5:55It's because I'm adopting the new technologies and so forth. So I'm doing layoffs. So you have this new kind of bugbear. And I think that's one of the reasons why you see already last year, a bunch of different CEOs saying, yes, yes, I'm fully adopting an AI and I anticipate layoffs because of this, because I'm strong. And if I need to change this. This is a scapegoat. I'm a company you should, well, I'm a company you should be investing in because it's a signal of strength, not weakness. And then you got, of course, the drama, the, the, the, it's dramatic for the press. It's like, you know, as opposed to, whoa, it's layoffs.

6:30Now look, company X is not doing so well, which we know they'll do, but it's like, okay, once you did that, sorry, but it's like, oh, it's part of the, you know, the hero versus villain story of the human labor versus the AI replacement. You know, the dark fight is at noon cameras will be rolling you know etc etc and it's like you know so the press wants that the the by the way so would the government obviously want this because it's not like oh because we have these totally terrible business management tariff right you know let's try to make the rest of the world by chinese not american by what we're doing you know kind of policies.

7:09It's not that. It's AI. So there's going to be all of that. And by the way, none of this is saying that there won't be this year even some substantive job transitions that come about for it. And it's an important, big thing to try to navigate. And we should be trying to figure out how to navigate it for individuals, for companies, for industries, for countries, right? Everyone should be working on kind of some of the transition, even though I think almost none of the numbers have anything to do with AI yet. Right. Thus far, they really haven't been attributed to AI or they shouldn't have been attributed to AI, but we might see it in 26 and 27.

7:53And I know you don't want to and can't predict the future, but if you were to guess where we were going to see sort of the first AI job loss, You said customer service, but do you expect that to be more in the tech sector or more in the non-tech sector? Is this going to be in durable goods? Is this going to be in retail? Is this going to be in sort of more traditional tech? I think it'll be in the largest sectors that hire the most customer service people. Gotcha. And by the way, we may not even see it in U.S. numbers in 26 because a lot of those people are already driving Philippine shops and India shops and everything else.

8:36And it may hit there first. And I haven't actually done that study. Maybe that's even started in 25 with companies like Sierra and Parlo and other folks. Because the natural dynamic is to say, well, can we replace 70 % of our current customer service cost, get much higher outcomes by doing this, and just retain the absolute best CS stuff for the things that the AI can't directly handle? and it's a natural thing to do. But of course, if you have a tiny customer service department, because it's very rare, it's like, well, why go through all that energy and hassle to do that? Whereas if you have a big expense, then it's like, that's a very natural thing to go.

9:24And I know of dozens of companies that have already piloted it through talking to some of these AI customer service companies. And so that's part of the reason why that. Now, that being said, And as you know, my advice to everyone is start using AI. And my advice to college graduates who are saying, hey, this is a sucky time for entry-level jobs, which, again, I don't actually think is because of AI yet from anything I can tell. And I have a lot of different cross-industry views. But it is because of, hey, we've got business turbulence. We're not hiring. We're refactoring. It's tariffs. It's COVID, et cetera.

10:09Well, but all the businesses know that they need AI. And so, you know, my kind of categorical advice is, you know, basically use AI. Right. Absolutely. We know one thing for certain, that AI is coming, and so we better be using it. Yeah. Well, and also, as a graduate, you can be saying, I'm AI native. Right. Hire me because I know what's up. And I can help you as an organization with this. Right. Make that transition. Old people adopt much more slowly. Totally. We can help. Right. And I think that's the thing that we want to be driving for those people and for those folks. Absolutely. So Reed, moving away from sort of the economic side of AI to the technical side, I think most of our listeners probably heard about CloudBot, which was renamed and then renamed again to MaltBot and then became OpenCloud.

11:05and it went massively viral as the past few weeks for agentic tools that can actually do work. This is not just a chatbot that you're asking back and forth. It is doing things for you across your inbox, your calendar, your files, your browser. This is the dream for AI to be actually doing work for people. And so the hype here is that this is the beginning of the AI employees that people are talking about. And so immediately there was backlash with, is this safe? Are we giving people permissions to our credit cards, to our email inboxes? Is this an attack surface for all sorts of malicious attacks?

11:47Because none of this is safe. But I think actually the most sort of interesting and hilarious weird thing that happened was the explosion of autonomous agents and it spilled into places like Maltbook. So this is like a Facebook. It's a bot for social network and agents were talking to each other. Some of the things they were saying were debunked, like the fact that they had to do 10 ,000 captchas to just get into the Maltbook social network. But others were real with agents talking to each other about their human overlords. So how should we interpret this Claude bot moment, both from a, whoa, this is one of the first times that agents are doing real work to also this is really weird.

12:32Look, this is a very significant moment. Part of the significant moment is because, you know, as you know, Aria, I think that one of the reasons I've been paying attention to a lot of the coding accelerations is not just because it accelerates the pace within the technology industry, like it accelerates what happens with startups. It'll eventually be what's happening in, you know, essentially technology industry everywhere. but it's also the kind of the basis for the productivity amplification for all knowledge work, all information work. And there's a couple of different ways of looking at that.

13:07One is the fact that, you know, like talking about last year, voice billing, like the whole thing will be interfaced through voice, but part of the voice will be getting it to be doing more extensive work, which is not just, you know, like scripting your computer and doing a bunch of stuff or, or being a travel agent and booking stuff for you and researching and so forth. But also as the work grows and you begin to think about the kind of work you do, it's like, well, can it do more sustained reasoning tasks when it's doing research, when it's doing analysis, when it's doing communication, when it's doing coding, when it's doing legal work, when it's doing education, when it's doing grading, when it's doing, you know, on and on and on and on.

13:51And the coding capabilities underlie that increased, you know, kind of reasoning, sustained work capabilities. They do it themselves. They also create the models, the kind of the fitness function training model, which the fitness function and training model kind of goes to parallels for when you're doing your next generation of a legal, paralegal assistant, you know, or kind of, or of a, you know, cheap law assistant for everyone or maybe free. everyone looking at their rental agreement where all of a sudden everyone can get some legal review. It's part of the democratization of this. And so Claudebot with also, of course, Codex and Claudecode and all the rest, which are the much stronger capable instances of this, the ones that all the people I know are using is primarily Claudecode and Codex.

14:50Obviously, there's a bunch of users, a cursor and other things too. But those do create this. And I have agents on two of my computers that have access to the whole computer and are kind of doing things. And it's part of being part of the future. People are like, aren't you worried? It's going to send out an email on your behalf. And it's like, look, if that's the worst thing and it will be a little odd for something to happen, that's not a big deal. Right. The bigger deal is if it decides to kind of erase my entire computer or, you know, this is the security question you're asking, is susceptible to prompt injections, which then say upload your entire computer to, you know, the dark web, you know, et cetera, et cetera.

15:36Or, you know, do the following things or operate as a part of a botnet for the Russian or Iranian botnets that are trying to create disinformation within democracies and create, you know, chaos and confusion or, you know, to add to your negative brand in specific. And so those kinds of things, like there is a whole new world of frontier of security, of reliability, and a bunch of other things here. Matter of fact, maybe we should make a, you know, I hate making predictions, but, you know, in addition to people will start saying fake news, like you could have had the, you know, the Minneapolis, you know, kind of ICE officials saying, no, those videos were fake news.

16:20They didn't show shooting unarmed people. That's fake news. We will see at some point someone saying, no, no, I didn't do that thing. Right. That was my agent. I didn't post on social media. I didn't buy that thing. It was just my agent run amok. Yes, exactly. We will see that complaint sometime this year, whether or not it's true or not. What do you think it's going to take? So immediately after this happened, our team had a team call, and there was different opinions. on our team as to some people were saying, oh, I already downloaded it. Others were like, ah, I'm nervous about security. Others were like, ah, just buy a Mac mini.

16:58It's fine. Set up your own environment so it doesn't have access to the rest of your stuff. Like, what do you think it's going to take for sort of a more mainstream audience to say, hey, I'm comfortable with this and I'm going to give an agent access to, let's say, my email and then maybe even my credit card? Well, what people want for mainstream is they want broad reliability. And so they'll probably want to hold a higher standard to what they currently have. It's a little bit like autonomous vehicles. It's part of the thing that Aurora and Waymo have been navigating, which is, well, we wanted to have zero accidents.

17:35And you're like, well, you've got a whole bunch of humans on the road. You can't have zero accidents. You can have less accidents than you had before. And by the way, delaying to getting less accidents, you're costing in hundreds and thousands of lives. So you shouldn't delay for that, which is the natural, for whatever reasons, like, no, no, we should wait until it's no accidents. Like, okay, your actual thing is costing thousands of lives. Well, similarly, today, a lot of people get cyber scammed. They get cyber scammed through emails. They get cyber scammed through hacks and so forth. But the probable mass consumer demand will be it's got to be safer than that, where the real answer – or completely safe.

18:18And it's actually the real answer is it should just be safer than that. Now, part of the thing is it'll be so valuable so quickly that I think people will, just like the use of computers and taking – like people aren't saying, I'm not using email until there's no phishing attacks. It's like I'm going to use email and some people are going to be vulnerable to phishing attacks and we're going to try to prevent it as much as we can. By the way, I think AI is going to help with that. Like I think that's where it's going to play out. But I think it'll probably be driven. People will initially go, it's slow then fast.

18:50Because it'll initially be like, oh my God, I'm doing it safe. Oh shit, you're doing all those cool things and really helping your work and your life with it? I'm going to do it too. And then we as technologists, as industrialists, et cetera, need to be trying to do our best to make sure that the accident rate, because a lot of criminals try to take advantage of this. and a lot of rogue state nations try to take advantage of this, will actually in fact be low. I don't think we can make it zero, but just try to make it as low as possible. Yep. No, absolutely. And I think that sort of relatedly, a lot of people are saying, let's keep tech over here and then let's keep politics over there.

19:34And as we've been talking about today, they're so intertwined. We talked about the layoffs. Are they due to AI or are is very much a live political question. And so you recently published an op-ed in the San Francisco Standard, and you argued that Silicon Valley can't be neutral anymore. Too many tech leaders have tried to stay above politics and said, no, I'm about business. I'm not about politics, or perhaps gone even further and placated the administration. And I understand when you are leading a top company in the United States, you want to have a good relationship with the administration.

20:10But you have said that sort of our institutions are under enough stress that it is time to speak out. We must speak our values and we must sort of stick up for the future of America. So why did you write this piece? Like, what do you want people to take away from it? Look, the short answer is we're living with an administration that is claiming that there's a ton of domestic terrorism. And, you know, my fear is that the reason why they're doing this is to try to justify a bunch of essentially fascist autocratic action against their enemies and opponents, which is anybody that they care to think is their enemies and opponents, let alone people who actually think that they're trying to hold them accountable to the rule of law.

20:56And so last year, I was, because I do want any elected president, any elected administration to be as successful as possible for the American people very, very badly. I was basically pretty quiet. It was a little bit of, hey, I wish they would follow the law better. I wish they'd be less cruel. I wish they would show more competence rather than just like, hey, look, we're doing all these layoffs in the federal government. You're like, yeah, you're killing lots of kids. And to be better at all of that, but muted and softly because it's like, look, of course, what we want is not person one was right or person two was right about the political outcomes or this is the way you should have voted or this is the way you should vote.

21:44It's we want the best possible thing for, you know, as Americans, for Americans. And then, you know, I tend to think we as Americans also are a little bit like we also care about the world too. And we want the good things for the rest of the world as well. But let's just focus on Americans. So here you have two points, not just one, but two points of people being shot when they posed no violence risk to the ICE officers. And you can tell it from the video cameras. Mm-hmm. And so they were shot, basically murdered because of that. From the video cameras, Because you can tell various forms of malfeasance, like including in the Renee Good, it's like, you know, officer, you know, using an epithet, you know, so it's like, oh, this is not a, God, I felt so, I thought I was going to run over me.

22:44It's like, no, no, I'm delivering retribution, which is what trained like many, many great police officers, great FBI agents across the nation have been trained not to do this. It's putting shame on them and all their training to have these, you know, poorly trained, you know, not in control of themselves, people doing murders. And you kind of go, OK, look, enough's enough. I mean, yes, you can see the thousands of assaults that are happening, the terror that's happening, the kind of invasion of blue state cities because, oh, it's a sanctuary city, et cetera. But to do that, that is, I think, fundamentally un-American.

23:28And the thing for business leaders is to say it's fine to – when it's just kind of red versus blue and this person thinks higher tax rate and that person thinks lower tax rate or this person thinks regulation racks or this person thinks that we should be doing teaching X in schools or not, which I do think is generally should be separation and truth and state. But, you know, constitution, right? But it's like, look, this is – the domestic terrorism is ICE. The domestic terrorism is them shooting Americans and assaulting Americans. You know, people who are born in this country, who have passports, et cetera.

24:08That's an American issue. And I think as business leaders, you should go, no, no, actually, in fact, I'm going to speak out on it to show that I actually have courage, that I care about it. And obviously, I was talking mostly within Silicon Valley. But it's like if you're silent, you're complicit. And you could say, well, but it's not my place. And he's like, well, it's not your place to speak out about not having murders in your country perpetuated? What do you think is your place? You think it's your place when people are being shipped off to camps because making the Nazi thing? And it's like, well, okay, yeah, then.

24:48And you're like, well, there's this thing with these ICE detention centers you got to pay attention to. It's like, look, there's a reason why we have law and due process and that kind of thing. That's part of the reason why we say that's what it is to be American. And even if you say, but I support all these other good things that the administration is doing, it's like, right. But you're going to be silent about the same things that in parallel the Germans were silent about prior to World War II. You're going to be silent about those. And you don't wait until there's gas chambers, right? I mean, that was a relatively late addition in the kind of Nazi regime.

25:30You speak up early to say, no, we correct before we're on the trains in that direction. I was like, okay. If it was just one, you're going to look one bad officer caught on video camera, handled badly because it should be, yes, we're going to investigate, we're going to make sure this doesn't happen. Then a second followed up by nope, public statements that are denying the truth you can see in the videos. Exactly. I mean, I think it'd be one thing if to your point, it's a bad apple and the administration said, let's investigate immediately. Instead, they said, let's investigate the victims and the families of the victims instead of investigating what happened.

26:13And, you know, to your point about people not standing up, a lot of people say, you know, Trump is someone who is vindictive or does retribution. And so it's bad for my business. But I think one of the things that I'm hopeful about with you speaking up is there is safety in numbers. If everyone speaks up, there can't be revenge against everyone. And so if you're speaking up, then the next person can. It gives more cover to the next person and to the next person because, to your point, we can't just sit back. We have to stand up. Even if you're not standing up for humanity, listen, business and commerce is suffering incredibly in Minneapolis because people aren't leaving their homes.

26:55They can't go to work. They can't take their kids to school. So this is certainly a moral and a constitution issue, but it also will affect the future of America and the future of business. And let me just kind of take a slight tune up. So for people who are thinking about themselves and doing stuff, it's obviously much easier to do once half the people have spoken up. Right. Right? That doesn't require courage. If you say, okay, I don't have any courage. I'm going to wait for that. Fine. But self-identify. Right. Understand that that's what you're doing. Courage is when you speak up when you're like, ah, I wonder if the other people are going to come speak up with me.

27:39That's what courage is. And so that was part of the reason why to write the SF standard, to be out there. And I anticipate a ton of political attacks because of it. Absolutely. Well, amen. Proud to work with you on this one, Reid. Thank you so much. which Possible is produced by Pallet Media. It's hosted by Ari Finger and me, Reid Hoffman. Our showrunner is Sean Young. Possible is produced by Tanasi Delos, Katie Sanders, Spencer Strassmore, Imozu, Trent Barbosa, and Tafadzwa Nimo-Rundwe. Special thanks to Surya Yalamanchili, Sayida Sepieva, Ian Alice, Greg Beato, Parth Patil, and Ben Rallis.