Recall, Not Encoding, Is the Bottleneck for LLM Factuality
Google Research's knowledge profiling framework shows frontier LLMs like Gemini3 and GPT-5 encode nearly all facts but often fail to recall them. Using WikiProfile, a 2,150-fact benchmark, they classify errors as recall failures ('lost keys') rather than encoding failures ('empty shelves'), suggesting post-training and inference-time methods could improve factuality.
Sources and evidence
Summary last validated Aug 31, 2026
Reader actions
Report an issue
Use this for an incorrect summary, wrong source, duplicate story, or wrong category. Submissions are private and do not change the story automatically.
Related coverage
- research
NIST Study Maps How Toxic Adulterants in Fentanyl Vary Across US
NIST researchers analyzed fentanyl samples nationwide and found that the toxic substances mixed into the drug vary by region and shift over time. The study aims to help first responders and law enforcement identify dangerous adulterants in local supplies, which the agency says can help save lives.
- research
NVIDIA teaches robots to assemble GB300 tester trays
NVIDIA describes how it taught robots to assemble tester trays for its Grace Blackwell GB300 superchip, which powers AI training and inference. The work required skilled physical labor in factories worldwide. NVIDIA says the project offered lessons in robot learning, mechanical intelligence, and traditional engineering.
- industry
Google launches Playground experimental gaming platform
Google introduced Playground, a new experimental gaming platform that lets users create, play, and share custom games. The announcement appeared on Google's official blog. No further details about the platform's features, availability, or underlying technology were provided in the source.
- research
Ai2's Bolmo byte-level language model technique published in Nature
Ai2 announced that the technique behind Bolmo, its fully open byte-level language models, has been published in Nature. New checkpoints show the approach generalizes beyond the Olmo model family to other model families, according to the Allen Institute for AI.
- research
Microsoft Researcher Jennifer Neville Discusses AI Surprising Failures
Microsoft Research podcast features Jennifer Neville, who discusses her unexpected path into computer science and her work identifying "surprising failures" that make it hard for AI to handle complexity. The episode explores how failure teaches lessons and led to her professional sweet spot in AI research.