Coffee With Scott Adams — Knowledge Archive May 24, 2026
Scott Adams Philosophy Archive
Search ideas

Context —

it could be a calculated risk as in, you know, something could happen to Trump along the way so that wouldn't be stupid. But there's no way you could win in a head-to-head I don't think that could happen. All right, I spent the morning arguing with Google's new AI called Bard, and it definitely argues like a Democrat. So I went after it on the fine people hoax, trying to make it admit what he sai…

← Previous segment →

back and forth with me to do anything it could to avoid admitting that that happened. Do you think it doesn't know? I think it has a full transcript. I feel like it knows, but it acted like it didn't. It was just sort of agreeing with the dominant coverage, the news coverage. All right, it actually was changing the topic on me. I mean it was actually doing the human thing where you change the topic when you're losing the argument badly. And they also asked about the drinking bleach hoax and it made no mention that the disinfectant was light until I pointed it out. And then it was, oh yes, disinfectant was light, but people disagree about what he was talking about. So I said but you know he mentioned light before and after the comments that you're looking at. Immediately before he said light, light, light, and then immediately after he said and light. So that you knew that what was in between was always intended to be light as well. And it couldn't handle that. It couldn't handle the full transcript. It was like it didn't know it or just wouldn't admit it. It acted the way it acted was exactly like a human with cognitive dissonance. Which makes me wonder if AI can have cognitive dissonance, right? Because cognitive dissonance is when you know, you're sure you know what you're talking about, but then you find out you don't, and so you just immediately hallucinate that really you were right all along. And that's what the AI was doing. I was very clearly showing it with facts that it could validate. It could check the facts easily and it did. I mean it was checking facts as I was talking, but it could check the facts so it knows that it was wrong and it wouldn't admit it. It just wouldn't admit it. Cognitive dissonance.

All right, so I asked Bard what my IQ was based on public reporting, and here's what it said. It said Scott Adams stated that he has an IQ of 180. He has also said that he is a member of Mensa, an organization for high IQ people. Are either of those two things true? Have I stated that I have an IQ of 180? No. I stated that it would be a funny prank to see if I could convince the internet that I do. I've never said I had an IQ of 180. But Bard thinks I did. They also said according to Bard I've said that I'm a member of Mensa. Nope, nope. I'm not a member of Mensa. I used to be, but that was 25 years ago. So the first two sentences it gets factually wrong. And then it says well it's possible that Adams has a high IQ because I wrote some books. And then it says ultimately the only way to know for sure what Adams' IQ is would be to have him take an IQ test. However even then there is no guarantee that the results would be accurate. IQ tests are not always reliable and they can be affected by a number of factors including stress, fatigue, and anxiety. So if I took an IQ test and I scored a 180, it could be not because I'm smart. It could be because I'm stressed, fatigued, or anxious. So there are lots of reasons that the answers could be wrong. So that's the way it works. No, that does

Context —

n't work that way. Okay. And so it ended this way. So while it is possible this kind of has a high IQ, there's no way to know for sure. Really? There's no way to know for sure except you could ask me to take an IQ test. All right, I've got a question for you. So the news today was that Bard, the Google AI, is connected to the internet, and therefore if it's connected to the internet you could ask…

Next segment → →