Anthropic just proved AI isn't getting better

Anthropic just proved AI isn't getting better(youtube.com)

9 points by kesor 11 days ago | 3 comments

dpoloncsak 11 days ago |

This whole thing just comes off like an ad for the course promoted at the end. "AI isn't going to get any better, which is why you need my course to learn how to run a business for yourself".

Wouldn't he have more to show for it than some 'Encrypted notes' app?

zamalek 11 days ago | |

That's a very empty statement: you're haven't mentioned and critiqued any of the points he made against LLMs at all.

dpoloncsak 11 days ago | | |

Because none or novel complaints.

>Context is expensive - Yeah. It's been the main point of contention for a while, right?

>It performs a quick search and then writes to those files, ignoring arch - The entire point of the Sol -> Terra loop is to solve this. Its not an inherent flaw, its a short coming of current models.

Also, all of these were worse have have been getting better over time. Context is getting bigger and cheaper. These are not proven to be inherent flaws in LLMs