Ajeya is a Senior Research Analyst at Open Philanthropy. Ayeya has also written the first work on eliciting latent knowledge on language models. She’s currently thinking about the likelihood that powerful AI systems may be misaligned, and what technical work may help to reduce that risk. Previously, she worked on estimating when transformative AI might be developed and thinking about how worldview diversification could be implemented in budget allocation. She joined Open Philanthropy in July 2016 as a Research Analyst. Ajeya has a B.S. in Electrical Engineering and Computer Science from UC Berkeley, where she founded the Effective Altruists of Berkeley student group and taught a course on effective altruism.
People have been surprised by the pace of progress in AI before, and they could get surprised again, and it’s very important to try to create benchmarks to measure and forecast possible dangerous capabilities in models. In this talk, we’ll examine the speed of this progress and the solutions and policy recommendations to keep up.
Add comment