Research
1962 Lab researches autonomous AI invention: systems that can originate, verify, refine and physically realise useful designs beyond those explicitly imagined by humans.
The research question
Can AI discover useful designs beyond human and algorithmic baselines, prove that they work, and know when the evidence is strong enough to make them real?
Today this is a research program, not a claim that autonomous invention or general physical delivery has been solved. Human intent is where we start; whether it is the ceiling is the open question.
We organise the work by capability layer rather than by product surface, because a layer can be studied long before it becomes a feature.
Turning a wish, a constraint, or an opportunity the system noticed itself into a goal that can actually be checked.
Generating genuinely different candidates — and changing method when the evidence says the current one is spent.
Deciding whether a candidate works using deterministic scorers, formal checks, test rigs or real physical measurement.
Whether the system knows when it is likely to be right, and when to continue, stop or change direction.
Handing a verified design to printing, suppliers or factories — and verifying what comes back.
A cheap, repeatable computational testbed for a question that has to be answered before an AI system can be trusted to run its own search: does the model's own confidence actually tell you anything about whether its idea will work?
What we found, and what we did not →We would like to hear from researchers and practitioners in:
Write to hello@1962lab.com . This is separate from the idea intake on the main site.
