I've tried LLM agents for coding. I find they they do an OK job at getting a small project off the ground. Afterwards they're useless for me. However, it really depends if you're writing code that's essential a copy & paste of common code on the internet or if you're writing something new. What I haven't tried is using agents only to unit test my code. That could possibly work out ok.

I've been writing software for 30 years. Honest code reviews take more mental effort than writing the code yourself, unless the changes are small and clearly and verifiably well tested. Proper design for unit testing is hard and beyond the capabilities of AI. Hence, you can't really do better than a human software engineer, yet.

You are correct. Meta (parent company of Facebook, Instagram, and WhatsApp) has an annual R&D spend exceeding US$40 billion. Five times more than the $7 billion (in today's money) quoted as the Bell Labs R&D budget.

Most people running Linux do not use it as a multiuser system. Most people have the root password. I don't picture dad giving their teenager a login to their Linux laptop. The teenager would have root on their own Linux laptop and could set any birthday they want.

Their other bot problem is that they frequently flag for TOS violations when you're saying the opposite of a violation. Like, if you say "I object to the notion that women shouldn't have equal rights" the bot will flag you for the phrase "women shouldn't have equal rights." So the only safe thing is to not even touch topics that relate to their TOS.

C'mon Elon, you could have used petamicrowatt units to justify calling it a petafab chip plant. I'm disappointed in your imagination for exaggeration

American industry doesn't have this kind of agility anymore

Nothing I do compares to idiot politicians starting wars and burning millions of gallons of oil, gas and tons of ordinance. I'll pick up trash in my neighborhood and make sure my car doesn't ruin my local air quality. Otherwise, nothing I do matters for global warming, even voting as the last blue guy blew up a gas pipeline too.

Question: "Open the Kill Switch Pod Bay doors Hal !"

One month ago Amazon announced huge AI related layoffs and reduction of red tape for decision making. Anyone could have predicted this outcome. There a new phenomenon, AI induced psychosis, which I think is much more widespread than people realize. It's almost unavoidable if you use AI frequently and will quietly creep into your thinking in whatever domain you use LLMs for.

However, they will ascend the ranks and their salaries will far out perform those who see through the BS. I consider it a different kind of intelligence, where people are really good at turning their brains off when it's beneficial to do so.

Legally they have to blame the computer. Those refunds are never happening

"I don't know what's wrong?!? I just can't issue the refund?!? Who's responsible here?!?!!!"

How dare they? Waaah!

This makes them our enemies. If you play with them, we won't play with you.

Which allows the AI to deepen the delusion