Comment by johnfn
7 days ago
I'm not sure what to say about calling someone a "liar" for stating that AI can work for hours unattended. I can prompt AI and have it run for an hour+ at a time and get good results out of it. I have no reason to lie; this is just a factual statement, sort of like saying that my test suite runs for an hour or something. Yes, you need to prompt it correctly and have the right environment and so forth, but it is absolutely not a "lie".
If you actually read the post you'll see the reasons to call him a liar:
1) faking benchmarks and lying about a model he profited from commercially (ie. fraud)
2) implying that only a few people (like himself) saw COVID coming. This is a lie: it was the New York Times that published a huge article on the coronavirus at the time indicated, and he, of course, didn't see it coming
3) he doesn't just fail to disclose his commercial interests in what he's peddling, he denies them
4) he confidently states that AI builds the next generation of AI, which he can't know, and has not been stated anywhere
The list goes on.
I did actually read the post -- or at least the first two pages, until the increasingly unhinged comments started to get a little redundant and I figured I had gotten the gist.
> implying that only a few people (like himself) saw COVID coming
Nowhere does the post imply this. The post says COVID was an exponential curve, and he thinks that AI is a similar curve. There is nothing in there saying that only he was the one to see this. The comment, and you, are responding to a sentiment that doesn't exist in the document.
> he confidently states that AI builds the next generation of AI, which he can't know
Anthropic reports 55% of engineers use Claude for debugging on a daily basis in December[1]. I am not sure how you come to the conclusion that "has not been stated anywhere".
I would respond to your other points but I feel like these are so thoroughly incorrect that I should probably stop here.
[1] https://www.anthropic.com/research/how-ai-is-transforming-wo...
And your answer for him faking benchmarks for financial gain? You conveniently left that one out.
1 reply →
Yes; and you can also find a bear that dances, if you visit a circus. Therefore saying bears can't dance is a lie.
I don't really understand what you are trying to say with this comment.
Something can be factually true; but in so rare a circumstance, that the claim is simultaneously true and so misleading it's practically a lie. Just like AIs that think for hours without guidance. That implies full automation is imminent, when the reality is it only works about 20-30% of the time correctly.
3 replies →