Refine Wins Top Harvard Venture Prize For AI Audit Tools
#1447Ben Golub@BEN_GOLUBcheck out my soundcloud
(the very same model weights used in the proof of the conjecture are helping us do this)
Ben Golub@ben_golubEarlier this month @RefineDotInk, a startup making tools for AI research audits, won a top prize in Harvard's flagship venture competition. 🎉 So proud of the amazing team that made this possible. We wanted to take the chance to sketch the larger vision behind the company. 1/
Earlier this month @RefineDotInk, a startup making tools for AI research audits, won a top prize in Harvard's flagship venture competition. 🎉
So proud of the amazing team that made this possible.
We wanted to take the chance to sketch the larger vision behind the company.
1/
Research underlies decisions from drug approvals to monetary policy to anti-poverty interventions.
The rate at which errors are found in published work shows existing quality control isn't strong enough.
2/
Ben Golub@ben_golubEarlier this month @RefineDotInk, a startup making tools for AI research audits, won a top prize in Harvard's flagship venture competition. 🎉 So proud of the amazing team that made this possible. We wanted to take the chance to sketch the larger vision behind the company. 1/
Our larger vision is that there should be a certification layer for technical work, building the best of AI capabilities deep into the research review process:
infrastructure for research to credibly earn the trust of those who rely on it.
5/
Ben Golub@ben_golubRefine's current product is already making a difference: When Refine processes a paper, it inspects every claim with a tireless obsessiveness. Our reports are relied upon by users from Nobel laureates to top journals to flag key errors before they do damage. 4/
We're hiring, by the way! If our vision excites you, please reach out here or at yann@refine.ink
Here's CEO Yann Calvó López's elevator pitch, delivered at the Harvard competition: 👇
Ben Golub@ben_golubOur larger vision is that there should be a certification layer for technical work, building the best of AI capabilities deep into the research review process: infrastructure for research to credibly earn the trust of those who rely on it. 5/