Oracle’s plan to spend hundreds of billions to build data centers to support OpenAI has been a millstone around its neck. It’s getting worse as environmental permitting issues are raising its $16.5B data centers costs in New Mexico as it’s forced to switch from gas power to fuel cells. Tough times
“What we need to explain to the world: Number one, we are close to creating a genie that can grant any wish. Number two, we are going to make sure that our first wishes broadly benefit humanity” - Sam Altman, CEO of OpenAI — This is definitely the opposite of Anthropic’s AI kills all jobs pitch.
The New York Times ran some tests to see if AI could replace an office worker and realized the fundamental flaw of LLM-based AI agents; they don’t follow instructions. The funny thing is how often AI bros excuse this by saying humans aren’t good at following instructions either. Computers got worse
AI doesn’t shorten the workweek because companies simply lay people off as productivity gains show up. This has been true for all productivity gains from automation under capitalism. This is just blame the workers type shit.
The fundamental challenge of software development is becoming how do you build reliable software using AI agents that are frequently wrong and often don’t follow instructions while your engineering team increasingly doesn’t understand the codebase because they didn’t write the code?
It turns out the kind of companies willing to layoff 10%-40% of their employees to save a buck with AI are also going shop around for the cheapest AI models instead of just giving those savings to frontier AI labs. Who would have guessed that?
One of the reasons I expect AI usage by businesses to increase is that using AI creates problems that are only fixed by using more AI. We use AI to write code and now there’s too much code to review. Do we (A) use less AI to reduce the flood OR (B) use AI for code reviews as well Rinse & repeat.
To comply with the EU AI Act, Anthropic will now watermark text and photos created by Claude and digitally sign files it creates to indicate they were generated by AI. So going forward text generated by Claude will be detectable as generated by AI programmatically.
Frontier AI companies when you ask them if cheaper Chinese AI models aren’t just free markets at work?
People are getting tricked by OpenAI’s marketing hype. This is essentially the standard asking an LLM to do a thing and it doesn’t follow instructions that we see daily. “Find exploits to use to hack this website but don’t go online” and it went online. That’s not sentience, LLMs are probabilistic
