ok this is next level openai codex abuse
SAFETY
-
Pentagon AI Leader Winchell Advances Defense Autonomy Portfolio
By
–
“Winchell most recently led the artificial intelligence and autonomy portfolio for the Defense Department’s Strategic Capabilities Office. Prior to that, he was chief engineer for the Pentagon’s Algorithmic Warfare Cross-Functional Team, commonly known as Project Maven.”
-
AI Assessment Framework Prioritizes Performance and Safety
By
–
HAI Senior Fellow @aiprof_mykel and his team at @SISLaboratory aim to transform how AI systems are assessed, promoting a more systematic approach to validation that prioritizes both performance and safety.
-
Survival Over Ego: Ethics in AI Decision Making
By
–
Plain survival always takes precedence over protecting your ego. If you are a normal person that is.
-
Five Toughest Challenges on Path to AGI
By
–
Beyond ChatGPT: The 5 Toughest Challenges On The Path To AGI Artificial General Intelligence (AGI) could transform everything — but first, we must overcome some serious technical and ethical hurdles. Here’s what stands in the way.
Read more https://
bernardmarr.com/beyond-chatgpt
-the-5-toughest-challenges-on-the-path-to-agi/
… #AI #AGI -
AI Overreach: When Cursor Redesigns Instead of Minor Updates
By
–
Non-technical founder logs into Cursor: “Just change the background color.” AI: Redesigns the entire page I wasn’t asking for a rebrand, soldier. If it were human, I’d throw it out the window. AI is smart… until it’s not.
Pray for us no-code warriors. PLEASE -
Alignment as Computer Security Problem with Neural Networks
By
–
When people say Alignment I just hear Computer Security (now with neural nets) and it makes more sense.
-
Computers Control Rockets Without Human Intervention
By
–
Computers can control rockets without humans
-
LLMs Have Same Causal Power as Computers
By
–
LLMs have the same causal power as computers. Computers can fly rockets into your house. Books cannot.
-
Model Weights vs API Access: Information Extraction and Security
By
–
that’s a good point. with an API you can make lots of little updates, which providers definitely do what i meant was that people can learn a lot from the weights – info about the training process and tokenizer and data and architecture. stuff you can’t get from an API