A crypto project (@yesnoerror) has the opportunity to do something epic here which is audit the full body of scientific literature - 90m+ papers for errors This is an endeavor enabled uniquely by AI x crypto because: (a) AI allows for processing of immense data sets (eg scientific papers) at a scale infeasible via human intelligence. What would take o1 a few weeks/months to process would take humans 45,000 working years. So even if you had a team of 5,000 PhDs it would take almost a decade (while not even keeping up with new papers in that decade) and very conservatively $5.4 billion assuming $120k/PhD/year and zero other organizational/management/other costs. With an o1 tuned model you could get a more accurate & standardized audit for ~$30 million ($0.30 x 90m papers) - less than 1% the cost (b) Given that this endeavor is a public good, raising $30m for it would take some effort. Crypto markets make that a lot easier. A few times a week there is some meme or AI project that sprouts up with eventual $30m+ realized profits. The agent is currently finding errors in papers at a ~3-4% rate and has reviewed 1700+ papers. This will soon be scaled to multiple papers per minute. Out of 90 million papers, its very likely there are many important papers out there with consequential errors and when corrected can have real positive implications for the world. Long term, there is potential for a product of value to be built given the team is training a model to better detect errors and also holistically assess the quality of papers. Now its feasible to generate a standardized quality score for every paper taking into account methodology, logic, data integrity, etc. It will be a lot easier to discern trash papers from real papers and overall encourage better science. Imagine a leaderboard for universities and other research orgs that incorporates not only citations but also quality. An AI model tuned for auditing papers could also potentially revolutionize the peer review process. First becoming a standardized tool and eventually potentially replacing humans.