I left a PhD program at Berkeley to build this. I know what I'm asking. The security of tenure, the respect of the academy, the intellectual freedom to pursue questions without immediate commercial pressure.
But I also know what you're not getting in academia: the ability to see your ideas deployed at scale within weeks rather than years. The resources to hire a team and attack problems from multiple angles. The satisfaction of building something that works instead of something that gets reviewed.
We're expanding our technical team. I need someone who understands both the theoretical foundations and the engineering realities. Someone who can publish if they want to, but whose primary metric is whether the system actually prevents harm.
Coffee? I'll come to campus.
Jian Li COO, Inferthermic
Inferthermic: Intelligence to buffer the entropy Empirically proven 98% Risky Interaction Reduction (RIR)
Email 13 (from Alex):
Subject: The Alignment Problem Needs Builders
Professor Okonkwo,
Your critique of current AI safety methodologies at the last AAAI panel - you were right. Everyone in that room knew you were right. And then they went back to their labs and kept doing the same incremental research because that's what the incentive structure rewards.
I'm building something different at Inferthermic. We're not trying to solve alignment from first principles. We're solving it practically, through post-processing entropy reduction, and it's working. Enterprise customers. Real revenue. Actual reduction in harmful outputs.