Lol what performative shite. Chinese astroturfing 101. You're either mentally ill or a shill.

90% of them would take an APM position at Google in a heartbeat. This is totally performative.

"No you see I'm a special SWE, I'm not one of the 80% that will get automated!"

Perhaps you fail to see how this would probably lead to textbooks costing even more and which will be directly passed to students as higher tuition?

You are assuming that the actual cost of a textbook on math which hasn't changed in centuries is hundreds of dollars per student per class when in actuality without the profit incentive a 100M could use the same ebook over a decade wherein the unit cost is almost too low to measured even if pay excellent folks to produce a new work.

The first thing in the release page is benchmark results...

https://www.anthropic.com/news/claude-fable-5-mythos-5

So you're saying colleges will just increase tuition to cover the cost of textbooks right?

Or they could lay off some middle managers. If they blame the layoffs on AI the market might even reward them for it.

And what, use the historically fabulously successful flat model?

The market will reward colleges for layoffs?

>Of course, none of this scales. Some of our intro courses have a thousand students.

Any ideas are much appreciated.

Oral exams graded by LLMs? Scale with the improving models. Based on GPQA Diamond results they're mostly at PhD level for subject trivia anyway.

The problem here is that it will work for now, but how do you make sure the LLM talks to the student, and not a different LLM? I guess vision models FTW?

In the end, will be build a GAN loop?

Why am I now reminded of corewars?

You don't need to read this to know it's wrong. Ted Chiang hasn't solved the hard problem so he can't say that anything is conscious or not. I'm a big fan of his sci-fi but his bold pronouncements about AI are pretty flimsy and are a pretty bad look for him imo.

This is kind of like saying you can't compare Computer Vision models to Human performance because those models were literally trained to identify objects in images...

I'm not saying you can't compare them, I'm saying it's pointless. LLM's are extremely large scale multivariate regression machines, evaluating it's output within it's own training domain is as pointless as seeing if a ball rolls downhill.

MCP will die for the same reason RAG died and why prompt engineering is dying. The models get better at understanding what you want and where to find the right tool or context to solve the problem on their own.