AI & Enterprise
Google DeepMind releases one-year results for AlphaEvolve; DNA variant detection errors cut 30 percent
Google DeepMind released one-year results for its generative AI tool AlphaEvolve, highlighting measurable performance gains across multiple fields. In life sciences, it cut DNA variant detection errors in the DeepConsensus model by 30 percent. In power grids, a model’s feasible-solution rate for AC optimal power flow rose from 14 percent to more than 88 percent. It also improved disaster-risk forecasts, proposed lower-error quantum circuits, and was applied to Google infrastructure and commercial use cases.
