Everyone’s betting big on AI, whether it’s video game companies like Ubisoft and Unity or, perhaps foremost, Nvidia, which has seen its stock price skyrocket due to AI-accelerated data centers’ use of its graphics cards. But amidst all the hype, there’s been little discussion of what comes after. What happens if AI doesn’t take all of our jobs, and the bubble bursts? On this week’s Aftermath Hours, we talk about that.
This week, Chris and I are joined by writer, podcaster, PR guy, and former games journalist Ed Zitron to discuss, well, a lot of stuff. We begin by talking about “media being destroyed by idiots,” as Ed puts it, before addressing Nvidia’s pivot to AI and how it could crash and burn, taking countless jobs with it.
Then we discuss AI NPCs in video games, which I got to try out several different flavors of at GDC. The verdict: They can be fun to mess with, but the novelty wears off quickly, and then what’s the point of them? How does being able to talk to NPCs – something players often try to avoid – improve video games? After that we move on to BioShock creator Ken Levine’s long-awaited new game, Judas, which got a big reveal this week. After all that time, it looks… like another BioShock. Somehow, we still find time to then talk about Final Fantasy VII Rebirth and Dragon’s Dogma 2, as well.
You can find this week's episode below and on Spotify, Apple, or wherever else you prefer to listen to podcasts. If you like what you hear, make sure to leave a review so that we can hire everybody who loses their jobs in the great AI crash to write blogs.
Here’s an excerpt from our conversation:
Chris: Watching that [Nvidia AI presentation that has almost 30 million views on YouTube] just took something out of me. This is in the service of what? Also, when a line goes up that fast, I'm really scared of it going down equally fast.
Ed: My biggest worry is that they're spinning up hectares of data centers and buying all this new shit -- the new Blackwell GPUs, they're sold out through 2025 now -- and what happens if there's not enough AI to use them by then? What happens if we realize at this point that AI is not that useful? What then? Shit's just gonna sit there? People are gonna lose their jobs.
But also, there's a very real chance that the tech industry takes this massive haircut over this. The good news is that of the startups I'm seeing, there are plenty that aren't playing in AI or are just barely using it. I think that's the scariest thing: It's not that nobody uses it. The problem is, what if this is only worth a few percentage points of profit? What if this is not revolutionary, but it's evolutionary? What if it kinda helps? And if that happens, what now?
I talk to regular people. I try to, because I realize I'm depressingly online. People are aware that this stuff is bullshit. Most people are like "I don't know what ChatGPT is." I said in my newsletter the other day "Oh yeah, I don't really know anyone who uses ChatGPT," and someone responded with "Uh sir, I'll have you know that I use ChatGPT in everything I do," and I was like "I don't fucking care, mate." And he keeps sending me blogs of all the way he uses it, and it's like "OK, well first of all, I had to code this specific thing." I messaged him and said, "OK, the beginning of your blog is disqualifying. You can't tell me you use this in your day-to-day life and then describe the Rube Goldberg machine you had to build to make it useful."
It's frustrating because [of] crypto, metaverse, and AI, AI is better in the sense that there's a product, but it's worse as far as the outcomes will be. When this falls apart, Nvidia is gonna drop like a fucking hot turd. It's gonna zoop out of you like you've been eating them psyllium husks. Is that how you say that?
Chris and Ed: [Nearly unintelligible but extremely agreeable crosstalk about psyllium husks.]
Ed: When this drops, a lot of people are gonna be very embarrassed. But what's scary about this is, all of the money with the metaverse stuff really went into private companies. The investment rounds were in private firms. The stock market didn't really rush to the metaverse. It gave a bit of a boost, but even then, it kind of punished Meta. The street was like "This is bullshit. What are you talking about, Mark, you fucking idiot?"
AI, everyone's like "Oooooooo." And this whole industry is based on letting people fill in the gaps, letting people say, "If I could automate everything, I could..." without thinking "Could I actually do this? When will I be able to do this? Is it possible to do this?" We can automate workers? No you can't – not if there's a single fucking hallucination. If you go up to a Wendy's operator, and they're like "I'll get you a hamburger and also shoot you with a gun." Is that OK?
Chris: Also, I'm just gonna say it: the graphics cards are mid and overpriced now.
Nathan: They're super overpriced. They're hilariously overpriced.
Ed: Also, AMD has come back. It's interesting watching that turnaround. It's also very depressing: [Nvidia CEO] Jensen Huang isn't a bad CEO generally. But he is now.
Chris: I don't think he is. But if you watch the presentation, you can see that this is not his natural environment in how he's doing these things. It's like when a nerd gets popular very quickly and doesn't know how to behave. They don't know how to read the room.
I'm worried about the breaking containment part -- and also normal people stuff. TSMC, [a Taiwanese chipmaker benefitting from the AI boom], when people are like "TSMC is worth money," it's like of course. They actually make shit.
Ed: That's the worrying thing with Nvidia. They make shit, but now they're funneling billions and billions of dollars into something to do nothing. When is the next stage of AI? When will AI be good? When can I rely on it? Because I wouldn't rely on it right now. I wouldn't rely on it for a goddamn second. There are situations where it works, but it doesn't work every time, and it doesn't work regularly enough.
I make the joke about the Wendy's generative AI telling you it has a gun. That might be bad. However, the more likely thing is it goes "Yeah, I can serve you a McDouble." That's actually the kind of shit that would get a company in trouble -- if Wendy's AI started offering you a Big Mac. That's what would happen if they didn't check the model, if they just plugged and played this shit. But you're not going to see it. I think that companies are not jumping in.
There was this report in The Information about KPMG. They bought about 45,000 accounts to Microsoft's Copilot AI. Crazy thing is, when pushed on this by a reporter, they went "We don't really know how we're gonna use it, but we want to make sure that our people can answer questions on it."
This is where the money is going. It's not going into doing shit. Don't you think that if AI was capable of replacing jobs, they'd be replacing them already? You think these companies are like "Well, we want to slow roll this. We don't want to fuck people over"? God no! Look at what they did in covid. Look at what they're doing now. They would replace us in two seconds. Every single "We'll replace them with AI" thing fails. If it can't replace humans, then what the hell is it good for?
Chris: It's like self-checkout. There's a guy who's like "Oh yeah, you fucked that up. Hold on. Let me take care of it." It didn't replace people. Now you just need somebody to babysit the robot.
