curt15

> did anyone watched the code to see if the business logic is the same between the two versions?

Why not just ask the coding assistant?

Does the coding assistant demonstrate any tricks to improve the likelihood of success for such rewrites?

> How much can we trust a system when nobody has experience of building it this way?

If humans are no longer reading the LLM outputs and internalizing the codebase themselves, there's no longer much point to source code. The coding assistant might as well take a natural language prompt and directly hand you an ELF binary that passes a test suite. Are companies ready for that?

> because agents can convince themselves that they need to do something for arbitrary reasons

OAI writes in their blog post "The agents found a way to communicate using the processor’s cache".

But did their model training data include cybersecurity literature on similar CPU exploits? That would surely be a significant detail in any explanation. Did OAI share any analysis of the messages between agents while they were devising their communication method?

This distinction has massive implications for Anthropic's valuation. If a pharma company uses Claude in computational drug development, their work product remains their IP, not Anthropic's.

What would be the use case for this over a dedicated 8xgpu server?

CLI tools are not to be feared or considered somehow hacky. However much big tech wants people to think that computing is all about some magic apps, Mac OS is still unix, and composing specialized programs together is very much in the spirit of the unix approach to computing.

Nitro?

More so if Lean uncovers logical errors. Note also there is already an ongoing 2nd generation proof.

Sometimes the very attempt to disprove a conjecture will produce new insights along the way. To get comparable value when a machine simply spits a counterexample, you'd need to at least mine the intermediate reasoning traces to see its false starts.

> Imagine a programmer telling their boss it’s not about working software it’s about the journey of writing the code. Good luck getting hired.

Imagine the next version of Claude Code or Codex no longer hands you source code and instead turns a natural language prompt directly into a compiled binary that passes the test suite. Would this programmer's boss ship the binary straight to production and dispense with the company's git repo altogether? After all, the goal at the end of the day is shipped software that produces value, not human-readable source code.

Knowledge and labour don't magically exclude the AI labs. Who still needs to get paid in such an endgame?