Beware of AI. It gets things wrong, all the time!

I’ve used AI a lot. ChatGPT, DeepSeek, Perplexity, Grok. Grok is the best for minimal hallucinations and being “fundamentally truth seeking”. 

In the case above, I asked it the same questions and it absolutely refused to give a “Yes or No” answer. It kept coming back against each push by me, with the line — correct, I believe — that the questions were too complex and nuanced for a binary yes or no answer. 

But still Grok gets things wrong. It will get data wrong and turn out silly charts. Sometimes wildly so. If you count in AI for anything serious, you absolutely have to double and triple check its answers.

The worst AI is DeepSeek. The Chinese one. Anything about China, current affairs or political issues, forget about it. If you want to know about the Cultural Revolution or the 1989 Tiananmen Square Massacre, or anything about the current mass purge of the Generals in the Central Military Commission, forget about it. DeepSeek will refuse. No amount of pushing will get it to change its mind. 

An example of ChatGPT’s world view. Which, of course, is the world view of its young leftie training team. 

From here 

Of course there are plenty on the Left who’ll say “but this is true!”. 

And I’ll say “no it’s not”. I know plenty of conservatives who are decent, honest folk. I know Democrats who are horrid to people who simply have a different opinion from them. I used to think Obama wonderful, but now see that he was the one who began the race-grift game, and did harm in the Middle East with his Iran policies. Elon Musk is doing vastly more for the world — Starlink, electric cars, Neuralink —- than he’s “bad” because he shifted support to Trump. 

And… Woke is usually batshit crazy.

So, yeah, like Grok, the questions can’t be answered by yes or no.