Re:Building an early warning system for LLM-aided biological...
by aftbit on Wednesday January 31, 2024 @02:18PM
(#39208050)
Well ironically this study shows that GPT-4 isn't actually very good:
>However, the obtained effect sizes were not large enough to be statistically significant, and our study highlighted the need for more research around what performance thresholds indicate a meaningful increase in risk.
A "mild uplift" in capabilities that isn't statistically significant doesn't really sound like overestimation.
>However, the obtained effect sizes were not large enough to be statistically significant, and our study highlighted the need for more research around what performance thresholds indicate a meaningful increase in risk.
A "mild uplift" in capabilities that isn't statistically significant doesn't really sound like overestimation.