NavList:

This is off-topic, but since I asked ChatGPT for a description of NavList, I thought I should share its math skill "issues" here, too. Basically, it has trouble punching the keys on its calculator!

ChatGPT is like Soylent Green. It's made of people. No computer could make mistakes like this!

I asked it to do a basic physics calculation: computer the relativistic time dilation factor at 99.9999% of the speed of light. I posed this puzzle because the answer is very familiar to me, and because I think it's fun. Numerically, at 99% of the speed of light, the time dilation is very nearly a factor 7. For every "pair of nines" tagged onto that percentage, that 7 gets multiplied by 10 and gets ever closer to a multiple of 1/sqrt(2) or 0.70710678. At 99.9999% of c, the time dilation factor is a bit more than 707. It's not a difficult problem. All the thing had to do was calculate 1/sqrt(1-β²) with β (which is v/c or speed relative to the speed of light, c) in this case equal to 0.999999. Note: this is really basic, day one special relativity in a high school physics class. If it sounds tricky, recognize that it's no more difficult than something like calculating the tangent of 89.919° in navigation work. You punch some keys, and you get the result.

Well, ChatGPT did fine almost to the end. It understood what it had to work on. It did make an odd statement about the speed of light itself, suggesting it was reading from a rather old book, but it set things up properly and it reached a point where it said it needed to calculate 1/sqrt(1-0.999998...). And if it had done that, it would have found 707, just as expected. Instead, it made a typo. The most famous Artificial Intelligence engine on Earth made a typo! It dropped a "9" in there and calculated 1/sqrt(1-0.99998...) [that's four 9's before the 8 instead of five 9's]. And so it reported that the result is approximately 223.6.

Wow!

It's as if this "artificial intelligence" is sitting there in its parents' basement banging away on a cheap scientific calculator, and it skipped a 9. It made a typo! So therefore, I conclude that ChatGPT is People! It's just like the final scene in "Soylent Green" (released in 1973, and set, by the way, in the distant future of 2022) except that it's not artificial food that's made of people; it's artificial intelligence. Turns out there's no A.I. miracle here. It's just unemployed graduate students slaving away in some Silicon Valley sweatshop. 

Or not. ;)

Frank Reed

File: