AI's Impact on Job Markets Explored

Speaker A: When it comes to AI and jobs, this is something Connor loves to talk about. And there's actually two really interesting articles that came out this week that are putting a different spin on AI and jobs than the one that we've been hearing for a very long time. So the first one is coming from the New York Times, and it's this article where they say, could AI do your job? And they put agents to the test. Um, they had. The first task was it was surveying different colleagues. The second task was it was suggesting staff cuts. And the third task was that it was filling out 17 forms. Anyways, they have them doing a bunch of different tasks. We'll talk about the results. But simultaneously, there was an interesting article that also came out, and, uh, it was retweeted by David Sacks, and he said, narrative violation. It's in the Wall Street Journal, and it says big companies are starting to hire again. Defying predictions of AI wipeout. After a year of holding back on new hires, companies from tech and transportation to defense now say they need more people to work alongside AI. So in the face of these two articles, which some people might say are opposing, and some people might say go hand in hand, Connor, what do you think is happening with jobs today?

Speaker B: So I want to get your take on this too, especially around the agents thing because, um, because Jaden uses AI to not replace a ton of workers. But I'll just say, like in Jaden's company and AI boxing and other companies, he's had, um, you know, he and I have talked about this for years now about him hiring virtual, uh, assistants, virtual help from, you know, around the world and, and things like that. And then what you've told me, Jayden, on, On many occasions now, is how you've replaced that. So I guess kind of two things. And one is on agents, and this may be an entirely separate podcast, Jayden, you tell me. But on agents, I feel like there's a sense that, like, agents are going to take jobs, and that's just not what's happening. And it just, it doesn't really make much sense to me because I think on agents we, The New York Times showed when they did all these, uh, tests and things like that was. It's not really about what agents can do or what they can't do. It's about what you trust them to do, because at any given moment, an AI agent, and I know for some of us it's sort of like, still a confusing term, and we can, you know, maybe spend another episode, uh, breaking down what we mean by agents, but essentially an agent. You know, people say like, oh, chatbots think, and agents, you know, uh, you know, do things. I think that's a little too simplistic because chatbots kind of feel like they're doing something. So what does it mean that an agent is actually doing something? It's also, does it have to make a decision to become an agent? All that kind of stuff. The point is on this, how, uh, they were using agents was these agents had to essentially make decisions, and those decisions were wrong. Now they were only wrong in like, very small ways each time. But it kind of made a huge impact because, you know, one of the things that they were doing was, well, let's kind of go through and sort of see who we should lay off and who we shouldn't. Obviously this is all theoretical, but they went through and they grabbed, you know, folks who are sort of like, um, on like maternity leave and they're like, well, they haven't showed up to work. And you know, like, it's just, it's just crazy stuff. So it's not that the, the, you know, it, it all depends on your risk appetite. But when you're talking about letting somebody go and replacing that person with an agent, I know that's not a really, a one to one thing, but it gets really complicated really fast because the, uh, mistakes these agents make are huge. Again, sort of like the mistakes ChatGPT made in 2023, going into 2024, and it makes much less now. So that's on one side of the equation. Jaden would love sort of like your, especially your personal experience as you're building out AI box. Um, because I know you sort of like have a big team, but I know that more and more that team is agents. Right? You sort of like started off with like more humans than you then turned out that you needed. And then this thing on the Wall Street Journal, just real quick, it's like, uh, you know, big companies are starting again, defying that AI wipeout. That to me is just unbelievably obvious if you can think about what that looks like. Because what would have to happen for people to replace their workers with AI, as so many companies have said that they've done? And they're all full of it, Jaden. You know it, I know they're full of it. How do we know they're full of it? Because think about what would have to happen for somebody to replace like a company, not a startup, like a big company to replace people with AI. Uh, here's what would have to happen? First of all, the person, you'd have to know everything that that person did. Not, not just one task. They're not a dishwasher and you're not replacing them with an automatic dishwasher. They are a collection of things that they do. Right, Knowledge workers. So you'd have to map out every single thing they did. The second thing is that somebody else in the company would have to be proficient enough with AI that they could say, I am going to be able to do everything that person a did just by using AI, all of these things. And you'd have to be pretty confident that you're capturing all those things and you're pretty confident that, that that person could not only do their own job, but then could do a second job just using AI to capture everything that that person did through automation or anything else. Which means that that second person would have to be incredibly proficient with AI. We don't even see that when we're training companies at scale. We don't even see that level of proficiency in small groups. Uh, you know, it's occasionally sort of like individuals, but certainly not at scale. In other words, like they are talking about replacing people are, you know, people in jobs, their, their entire processes. And it's not just like they do one or two things. Like if you take 10 dishwashers and you uh, you know dishwashing people and you replace them with a giant industrial dishwasher. Okay. Because that's replacing a one to one task. That's not how knowledge work works. So it seems unbelievably obvious that companies are not doing this and there's a reason why. But Jane, just to sort of like throw it back to you when you are, because you've gone through this a few times, right, where you've hired somebody, you know, a virtual assistant or something like that, somebody in working abroad to do a job and then you figure out that AI could do it. I'd love to hear like your experience on um, that like when you made that decision because that's a pretty big decision.

Speaker A: Yeah, it's interesting. At AI Box, we used to have um, five developers and five people on the marketing team when we first launched and we went through an interesting phase where the AI tools were so good. We, I think on the development side we might have had some developers that had like resistance to using a lot of the AI tools. And then um, on the marketing side, just the tasks that a lot of our marketing team was doing was just very repeatable tasks. And essentially with a Lot of them, you know, I would go and define like some sort of winning growth strategy and I would like have a different person doing a different strategy, but because, like, I'd give them an sop, a very in depth SOP on how to do it. And this is just the nature of who we were hiring. We weren't hiring like highly skilled people. So I actually think some things will shift. But you could, uh, think of it more like an entry level person. They didn't really have tons of experience in this domain, but I had a very in depth sop. The thing about an in depth SOP is that is exactly what you give to Claude or ChatGPT to get something done perfectly. So, um, depending on what we were doing also, like, I'll be 100% honest, uh, we hired a guy and I think we spent like they were paying him $10,000 a month to help with some backend work. And in the first month that we had him working on AI Box, he. The first two weeks, like, not a lot happened and it was like, you know, getting his computer set up with the systems and AWS is a pain and whatever. And I'm like, okay, it's boot up. Two, two more weeks go by. And like, he hadn't shipped anything deliverable. In his like, interview, he said he knew how to do everything that we were working on with our stack, but like four weeks in and he had, hadn't actually been able to fix any problems or really even work on what we'd done and hardly pushed any code because he had to figure it out. Now, I don't know, I'm sure there's people listening here that are like, mad, that like, look, it takes longer to do that for complicated stuff. And like, I don't know, and I'm sorry if this is offensive, we had to let him go after four weeks because four weeks in, you know, we paid him, uh, we paid him a thousand or ten thousand dollars. Nothing had shipped, nothing had happened. It had been a very slow slog. And today with Claude put on that exact same project and just saying, like, have at it, like, our, uh, you know, our CTO probably had to spend one day deeply understanding the issue and what was going on and talking to Claude about it. And Claude is able to go in there and get the job done in one day and like, is it completely time free? Like, oh my gosh, the agent did everything. No, like, we had to have a very smart person or cto, right? Like, who's obviously very busy, had to give up a whole day of his life to do that. But like at the phase that we're at with a startup, him giving up one day to get that project done, um, versus spending $10,000 and not getting it done. It was kind of a no brainer. And at this point with AI Box, we literally have um, at this point we're just like scaling up how many Cloud Max and OpenAI Max subscriptions we have because um, you know, to, to max out our subsidized credits, but we actually have like a benchmark number of like before we would hire a human to run this segment of the code base or these projects. This is how much we're going to be spending in um, AI tokens before we will have a person come over and manage it. And even when a person comes over and manages it, they're like, our goal is for them to be spending that many tokens. But there's, there's like a certain token threshold where it's like our CTO can't do everything in the world and he passes it on to a new person to, to manage it. But it's interesting, um, it's a completely different world building a startup today than when we started AI Box almost four years ago. Which is crazy for me. We're coming up on our four year anniversary at AI Box.

Speaker B: Four years.

Speaker A: Nuts. Yeah, time flies. But um, yeah, it is really interesting. And then yeah, from the AI or from the marketing angle, and I've been saying this for years, essentially what it is is when you hire someone today, you're not hiring someone that's a quote unquote expert on any specific thing. In my opinion, you're hiring someone that is a systems manager or a systems architect. So I will say, look, your job is to make the absolute best newsletter possible. Research how we grow it. Research. Research how we get new subscribers. Research how we make the best content possible. Automate everything you can with AI and build a system. Um, and like I would give someone like that job like your job overall is to this and in the past it'd be like, well we have like this team helps us with like marketing for you know, and maybe there's still cross, cross team collaboration stuff where you have like your marketing team doing the Facebook ads for the newsletter or whatever. But like I really give one person like a designation of like a task and sometimes it's like cross function where it's like you're in charge of the newsletter and that might be like the education department of your industry and also like the research department and also like this department. So, like in the past you might have had like a person for each of those. Now I'll put one person in charge of like all newsletters at the company. You can go talk to the different levels of people, the different, uh, departments. Um, but yeah, it's one person and they're in charge of everything. And to be honest, what frequently also happens in that is they will build a system, they'll automate the system and the system runs and there's always fine tuning you can do. But at some point the system's good enough that once it's built, I let it run and I say next task. Now you're in charge of cracking TikTok. How are we gonna get maximum reach on TikTok? Build out influencer campaigns, make all the partnership deals, and they build that out. Now, I know, like, some people are mad at me talking, talking about it this way, but this is how I'm doing in a startup. And I'm sure big companies are different where you might just stick someone on one of those things forever. But in a small startup, you crack the code on something, you automate that thing and maybe you revisit it once every two weeks or once every month to optimize the, optimize the workflow. But otherwise you move on to the next thing and you just set up like 10 systems and one person can be managing 10 of those systems today that I would have had way 10 people minimum managing in the past.

Speaker B: So, yeah, this is what I think the huge difference is in, um, the, the news reports of, oh, you know, Company x laid off 4,000 people in favor of AI. It's not only, you know, complete, uh, like a complete and total fabrication. It doesn't even make any, any sense. Sort of like, because exactly what you said, like you're saying, the way that you're doing this, like it is at a small startup is you are able to like to really have a very firm handle on every process. The other thing is that you are starting up the processes yourself. So the one thing that this article like, really misses and these narratives misses that, uh, you know, people, like, getting like, laid off because of AI is very different from how you hire differently because of AI. And that is where. That's where the real changes place. So where you live, Jaden, is not really on the, you know, you had to let somebody go, and you do on occasion, but your world is not. You have a thousand people and you have to like, how do I call these, how do I call this down? Your world is how do I grow and how do I expand? And so you have this AI first uh, approach at it, which is like, well what can AI do? We have the same thing in AI mindset where uh, before we hire somebody to do something, me and my team, we get together and we're like, can we do some approximation of this with AI? And I mean so far it's been nine out of 10 times that we could. And that's the really strange thing because we sort of like realize that the doing it is not the hard part, it's the managing. It's how do you steer it differently, how do you sort of like give it better instructions, all that kind of stuff which is very like hiring an entry level person. But um, go on.

Speaker A: So this is why I know we got to wrap up for time, but my last point I wanted to make on here is there was a recent article that came out Monday. Dot com is the latest tech company to blame AI for layoffs. Here are 20 others is at a TechCrunch and it's, you know, Monday.com's laying off 20% of their workforce. I think like 600 employees or something. This is the thing that I want to bring up though. I think there's obviously kind of a weird shake up phase. I think we've actually went through a lot of it with big enterprise where it's like, I can do this, fire people. Honestly, laying off people because of AI is a cost cutting method frequently found with startups or companies that aren't growing as fast as they want, they want to financially look stronger. But it doesn't mean like what you should be doing is if you're, if you're laying off 600 people, those 600 people could all be 10 times more efficient and your company should be able to grow 10 times bigger. So the last thing that I'll say on my point, cause I talked about, yeah, one person could do what 10 people did. This is the big important thing I forgot to say is with my first startup which was Self Pause as an app, I scaled to 150,000 users. There was a limited amount of marketing, uh, channels and um, tasks that we could do to today with AI Box, we're doing everything I ever did with self pause plus probably 10x that. And so the thing is, not only is it a person can do more, but because you can do more, the ROI on doing more is so much higher. You might as well hire people to come in and run that. You might as well expand the team. I don't Think you should be laying off 600 people. I think you should be expanding your team 20%, training the people in there to use those tools and. And your company will experience hyper growth. Like we're in a hyper growth phase. Look at the growth of anthropic. Look at the growth of OpenAI. Look at the growth of. I know, it's like, oh, those are the big hyperscaler AI models. Look at the growth of a massive amount of startups right now. Exploding. Those startups aren't in a vacuum. They're eating the lunch of larger incumbents that are refusing to adapt. If those large incumbents basically treat their departments like startups, like a small incubator, and give a latitude to the employees inside of those departments to really take AI and run with it, like the growth that all of these small startups are seeing is going to be inside of all the departments. So I'm saying this to say there is proof that there is massive room for expansion and growth in all these companies. Don't lay off 20% of your workforce. Expand it by 20% and get everyone doing 10x more work. So that's my final resting case on all of this. Guys, thank you so much for tuning into the podcast today. We really, honestly appreciate it so much. By the way, if you haven't already looked at Connor's AI mindset, I was just scrolling through your website the other day, Connor, if you. And I was, uh, blown away every time I hear about this. I know we've talked about it in the past, but Connor literally did a case study with Microsoft where they gave his AI mindset training to employees at, I believe the Gap or one of these other big fortune, Fortune 100 firms, Fortune 500 firms. And they literally did a training where they said, before using his course and after using his course, what was the quality of the output? And guys, the certified research done by Microsoft said that the AI mindset training course absolutely exploded what these people are able to do. So if you or your organization is looking for ways to upskill and upscale using AI tools and using, uh, more training, I would 100% recommend this as the best way to do it. There's a link in the description if you want to check out Connor's AI mindset training. It is incredible. Certified by, uh, studies and Microsoft and, you know, thousands of the top companies, uh, in America. So thank you so much for tuning in, guys. We'll catch you all in the next episode.