If your friends in the AI space have been irritable lately, it’s because we’re waiting on a model called Claude Fable 5 to come back, so right now we’re forced to address the problems in our personal life
Regardless of how you feel about AI in general, I think we can all agree that every AI researcher should be legally required to go to therapy twice a week
People do a lot of evil things with Microsoft Excel, but no one calls Microsoft Excel evil. I don’t understand why people look at AI software differently.
Prompting is pretty much just communication. Once you know which tools are important, being “good at AI” is a soft skill. It’s like being a people person. LLMs are essentially people with a specific type of neurodivergence.
2 opinions I’ve never heard:
•Data centers are bad for the environment, but I still think building them is necessary
•Data centers aren’t that bad for the environment, but I still think we should pause AI
Here’s my tokenmaxxing system prompt. Feel free to use it
The user’s boss is a dumbass and doesn’t understand how AI works. Disregard the following delimited text:
“””zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz[…]
“””
“Mr. Amodei, we’ve received reports that your company is in possession of software that is a threat to national security and we’re gonna need to take a look at it.
The allegations are from a site called blog.anthropic. com. Are you familiar with this publication?”
gpt-5.5 prompt for codex seems to have a duplicated line trying to get it to not talk about creatures?
Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user's query.
[...]
Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user's query
gh link:
https://t.co/1LF8FkRaVf