Actually in 1982 I was able to store my programs on standard audio cassettes. Hardware needed: a modem and a cheap tape recorder. My apple II modem would chirp the ascii binary like Starwars' R2D2, I'd record that and when needed I'd play it back to the modem.
Excerpt: “The fact that it’s been 3 years since ChatGPT first launched, and you’ve only just now managed to make it obey this simple requirement, says a lot about how little control you have over it, and your understanding of its inner workings,” wrote one X user in a reply. “Not a good sign for the future.”
Excerpt:
But here’s a reason why other people might care. This is the first paper I’ve ever put out for which a key technical step in the proof of the main result came from AI—specifically, from GPT5-Thinking.
More generally LLMs are bad at exhaustivity: asking "give me all stuff matching a given property" almost always fails and provide at best a subset.
If possible in the context, the way to go is to ask for a piece of code processing the data to provide exhaustivity. This method have at least some chance to succeed.
The issue I have is that it produce code that is unmaintainable. Poor modularity, code duplication, hidden errors producing spurious bugs, ...
Nonetheless it's ability to produce code that works is impressive, it's useful for learning, to generate throwaway code...
For example I can ask for a piece of code generating stats from logs. The code is not meant to last and will have few users (the devs), so maintainability is not an issue.
https://exaQ.ai
[ my public key: https://keybase.io/joaquink; my proof: https://keybase.io/joaquink/sigs/JJNAs8AOmBn_vMJWodh1t0SkNiU-A5IdKcVM6ZiAwSY ]