We need a Science of Evals — Apollo Research

apolloresearch.ai · 2,835 words · saved by 1 readers

bibtext citation: @misc{Hobbhahn2024ScienceOfEvals, author = {Marius Hobbhahn, Jeremy Scheurer}, title = {We need a Science of Evals}, year = {2024}, howpublished = {\url{https://www.apolloresearch.ai/blog/we-need-a-science-of-evals}}, note = {Accessed: 2024-03-06} } In this post, we argue that if AI model evaluations (evals) want to have meaningful real-world impact, we need a “Science of Evals”, i.e. the field needs rigorous scientific processes that provide more confidence in evals methodology and results. Model evaluations allow us to reduce uncertainty about properties of Neural Networks and thereby inform safety-related decisions. For example, evals underpin many Responsible Scaling Policies and future laws might directly link risk thresholds to specific evals. Thus, we need to ensure that we accurately measure the targeted property and we can trust the results from model evaluations. This is particularly important when a decision not to deploy the AI system could lead to signifi

We Need A ‘Science of Evals’ – Apollo Research January 22, 2024 We Need A ‘Science of Evals’ Contents In this post, we argue that if AI model evaluations (evals) want to have meaningful real-world impact, we need a “Science of Evals”, i.e. the field needs rigorous scientific processes that provide more confidence in evals methodology and results. Model evaluations allow us to reduce uncertainty about properties of Neural Networks and thereby inform safety-related decisions. For example, evals underpin many Responsible Scaling Policies and future laws might directly link risk thresholds t

We need a Science of Evals — Apollo Research

related reading