I only say partially since we can measure the impact of AI improvement more directly on meaningful tasks: better coding, fewer hallucinations in summarizing a complex medical case, etc. In these situations, experts can see meaningful small differences among models as they improve
RESEARCH
-
Speech and Language Processing August 2025 Release Available
By
–
Now that school is starting for lots of folks, it's time for a new release of Speech and Language Processing! Jim and I added all sorts of material for the August 2025 release! With slides to match! Check it out here: web.stanford.edu/~jurafsky/s…
-
AI Quality Depends on Data: What You Can Do
By
–
AI Is Only As Good As The Data Behind It—And What You Can Do About It
#AI #AIio #AIInnovation #ML #DataScience #Futureofwork @fabiomoioli @pascal_bornet @alliekmiller @mattshumer_ @OfficialLoganK @jeremyphoward @GaryMarcus -

Stanford releases free LLM and Transformer learning cheatsheets
By
–
Stanford University released the best cheatsheets you'll ever find to learn LLMs & Transformers! These cheatsheets cover: • Self-attention, Flash Attention, LoRA, SFT
• Mixture of Experts, Distillation, Quantization
• RAG, Agents, LLM-as-a-judge 100% Free and Open Source -
Large Foundation Models Depreciate Faster Than Any Asset
By
–
The high cost required to train large foundation models makes them the fastest depreciating asset in human history.
-
AI latency now crucial for human-like cognitive modeling and behavioral prediction.
By
–
Behavioral prediction at human-like reaction speeds? We’re entering a phase where latency itself starts to matter in modeling cognition.
-
Google Unveils SIMA Agent Learning Inside Genie 3
By
–
Je n’ai pas les mots… Ce nouvel exemple me laisse totalement sans voix. Google dévoile donc L’agent "SIMA" qui apprend à l’intérieur de génie 3.
— VISION IA (@vision_ia) 24 août 2025
Regardez ce nouvel exemple dévoilé par les chercheurs de Google : on dirait le marché de ma ville un samedi matin ! La qualité est… pic.twitter.com/KZJroj8BfcJe n’ai pas les mots… Ce nouvel exemple me laisse totalement sans voix. Google dévoile donc L’agent "SIMA" qui apprend à l’intérieur de génie 3. Regardez ce nouvel exemple dévoilé par les chercheurs de Google : on dirait le marché de ma ville un samedi matin ! La qualité est
-
Waterloo Co-op Experience Led to Google and Language Models
By
–
Happy to chat. Waterloo coop was the reason I got into Google in 2005 working on Language Models.
-
Training Models Beyond Markdown Documentation Misconceptions
By
–
Many people now think training a model is writing a large markdown doc…
-

Meta Meeting Room Named After Influential AI Research Paper
By
–
My previous meeting room at Meta was named after the title of this paper.