Hi,
I am the author of this blog post (thanks for the share).
Just to be aware, this blog post does not target anyone in particular. Rakyll is just an example among many (many many) other people on social networks.
Do not hesitate to comment, will try to check all comments and answer if necessary (when I have time).
1) They are all incentivised to increase their stock holdings / options (Google has a 14% stake in Anthropic) and several researchers + VCs on the timeline also have shares in AI companies and will say anything to hype each other up.
So there is more bias in what they are promoting than reality.
2) Many of them are so tempted to make up numbers to game the evals leaderboard using superfluous "benchmarks" and evals (Meta was caught faking their evals for Llama).
3) Mentioning limitations or disclaimers will not get any attention. Only if you hype up AI capabilities, doomsaying, over-preaching about AI accelerating because of (1).
Once they "leave", their shares have already been vested and on to the next "AI startup" scam until it all crashes.
Hi, I am the author of this blog post (thanks for the share).
Just to be aware, this blog post does not target anyone in particular. Rakyll is just an example among many (many many) other people on social networks.
Do not hesitate to comment, will try to check all comments and answer if necessary (when I have time).
I usually ignore their Tweets simply because:
1) They are all incentivised to increase their stock holdings / options (Google has a 14% stake in Anthropic) and several researchers + VCs on the timeline also have shares in AI companies and will say anything to hype each other up.
So there is more bias in what they are promoting than reality.
2) Many of them are so tempted to make up numbers to game the evals leaderboard using superfluous "benchmarks" and evals (Meta was caught faking their evals for Llama).
3) Mentioning limitations or disclaimers will not get any attention. Only if you hype up AI capabilities, doomsaying, over-preaching about AI accelerating because of (1).
Once they "leave", their shares have already been vested and on to the next "AI startup" scam until it all crashes.