One pattern I’ve been using in the newest OpenAI model Astra is asking it upfron
By Allie K. Miller · September 12, 2026 · Curated by George's Blog
One pattern I’ve been using in the newest OpenAI model Astra is asking it upfront for hypotheses.
And I can see Ethan Mollick is doing it as well in his prompts.
Then, after I get its hypotheses, I ask it to spin up dozens to hundreds of agents to go off and research and validate and discover, each with different context windows.
If we believe AI will start to have scientific breakthroughs in the next 18 months (which many experts do, including myself), then running more hypotheses testing could prove to be more helpful.
Said another way, is AI just hammering a million options and randomly finding a good one (like the CoastRunners reinforcement learning example) or is AI actually ideating new theories and solutions? And does it matter?