Jeff For Local Marketing (speaker) explains he frequently tests tasks through AI models to see capabilities. He reports that AI models now score above human PhD-levels on standard intelligence/math tests, cites hearing this from Peter Diamantes, and states that ChatGPT (version 4.5) scored 100 on a math test. The tone is conversational and anecdotal; the clip frames model performance as striking and broadly capable.
100
There's almost nothing today that I don't try to run it through one of the AI models. Why? Because I'm going to find out. It kind of knows everything. It's already reached those levels where it already, it already gets scores on all the tests about how smart are these things. It already gets higher scores than any human on, on the planet. I read that, I mean, I heard that last week from Peter Diamantes where they were talking about this. It's already, it's already scoring scores that, you know, they're, they're way beyond the PhD level. In fact, they were, they were reading part of this test and it was on the math test and they were reading questions and I'm like, I don't even know what that means. Just a question. I had no idea what this meant. And ChatGPT in this case, 4.5, scored a 100. So it's pretty cool.
