§ feed · storyline
Welcome to May 13, 2026 - Dr. Alex Wissner-Gross
ProgramBench launches as a benchmark evaluating whether language models can reconstruct programs, framing test-takers as test-makers in a Singularity-adjacent assessment.
ProgramBench is an evaluation measuring whether language models can rebuild programs, relating to the Singularity concept where test-takers become test-makers.
§ sources1 publication · timeline below