Coffee With Scott Adams — Knowledge Archive May 24, 2026
Scott Adams Philosophy Archive
Search ideas
Episodes Episode #3048 Segments
MainContent Economics & Finance

Back to episode — Episode 3048 CWSA 12/20/25

Context —

eir prices so we had most favored nation prices? Did you see that? Do you think any other president could have threatened? The others didn't even have the tool. Trump created the tool, the tariffs, and then threatened big pharma unless they brought $150 billion dollars of manufacturing back to this country and lowered their prices as much as, you know, 80 or 90 percent that otherwise they would ge…

← Previous segment →

tom line than a worker by himself or the AI by itself, which is a very solid argument, I think, but we don't know if that's the way it'll turn out.

The other thing is that people can't connect the potential benefits of AI to how their life will be better, but they can tell that it looks like AI might use up their water with data centers, which is a hoax. So apparently the data centers do not use up the local water supply. That's just something people believe.

Also, what is this in all caps? I can't read it. Also if Trump demands that the big data centers and the AI centers, if he demands that they build their own power plants, you might end up with a lot of clean energy that you didn't have before and it wouldn't affect the locals because the data center would have its own power plant. I think that's what we've already heard some announcements about that, right? Was it Google who is building combination? Maybe it was Amazon, but they'll all have to do it. So if you're in the AI business, you will have to be in the nuclear power business, and that's the only way that any of it will work.

Really? You generated content that would have taken you a week manually? Huh? Well, there you go. If hiring you once made sense, but you were 10 times more productive with AI, why wouldn't they want two of you? Because the AI didn't do the work on its own, right? I'm trying to think in the real world, you know, if I put myself back in the cubicle, even if AI could do all the things that humans can't do fast enough or good enough, you still need a human to tell it to do it. You still need a human to say whether it's been done and you still need a human to report to their boss and say I could or could not do this and you know so I feel like there's no legitimate way where an AI can just do what the boss wants. The boss will have to explain it to a human, threaten them with firing and tell them to use AI as a tool, but that ultimately the human will be responsible. So I commented about the All-In Pod first. It was a great episode. One of the best podcasts I've ever seen actually. It was just so interesting. Those guys are so good at explaining complicated things.

My voice is stronger maybe. Yeah. So I wouldn't trust an AI to do almost anything. I saw a lawyer, some lawyer said that he spent a week preparing a legal argument or something and then he tried to duplicate it with AI and AI did as good or better but did it in like a minute. But you still need the lawyer because the AI isn't a lawyer, right? You know, you're not going to have a court case where the lawyer says, "I will send a robot to do the closing argument." Yeah.

No cough this morning. I had a pretty bad allergic reaction. I think it was last night. Give me about six hours of coffee, but when it stopped, it just stopped. Yep. AI information

Context —

delusion. I don't know what that is. You have to know how to prompt the AI. Exactly. If the skill of using AI still resides in the person who's giving it the prompts, how are you going to replace the human? Anyway, we'll see what else is happening. Come in. Hi. I'm on a live stream right now, but that's okay. Yeah, I'm gonna put my name on the board. Okay. Is going home, so I'll be your nurse tod…

Next segment → →