@ZachWeinersmith I spend an unfortunate amount of time helping new coders with ChatGPT-generated code. Repeatedly, the code is worse than nothing. It takes them down the wrong road (often a nonexistent road), making it harder for them to find the right approach. Code-reviewing AI code is very difficult because its mistakes are not like the mistakes human coders make.
You should not use an AI to write code you couldn’t have easily written yourself. You must be able to carefully judge its output.
What is your main browser (engine)?
Re-running a poll I saw on Mastodon a while ago, before various migrations. Just collecting numbers, not seeking advice or anything. Please boost for reach.
The myth that AI x risk concerns are 'orthogonal' to dealing with actually existing AI harms has to be challenged. We make the road by walking.
Dealing with existing harms now is the path no matter what your long term goal might be.
Those voluntary commitments from AI cos announced at the White House on July 21? Here's a look at what's missing and what's weirdly included -- and why self-regulation just isn't going to cut it.
Lesley Carhart 
Rob Napier
Ivan Sagalaev 
mia
Sasha Costanza-Chock
Berkubernetus
Paris Marx
Prof. Emily M. Bender(she/her)