Yes, we should create a comprehensive estimate of all the sources of change in consumer surplus. We would likely find that much of the welfare increase created recently is not from water, AC, haircuts or cars, but from new goods that have low or zero price, like AI. Even if
RESEARCH
-
Social Groups Shape Technology Path: Historical Lessons
By
–
Even understanding the history of technology can help. For example, Pinch & Bijker’s studies of the early showed how tech is often not deterministic, its path is shaped by the social groups that use it & their view of its problems. Some useful lessons. https://
eg.bucknell.edu/~koutslts/FS09
8/Articles/JW_8.pdf
… -
AI’s Cross-disciplinary Impact and Research Application
By
–
One of the most interesting things about AI is how widely the effects will be felt and how relevant research across many fields is to its application. We don’t need to start from first principles in every use case.
-
Mentorship Models and Adoption Frameworks for Advanced AI Systems
By
–
For example, we actually know a decent amount about how mentors, tutors, coaches and teachers operate that would help push past “friendly assistant.” And we have models of technological adoption and social change that we help make more nuanced predictions about the AI future.
-
AI Research Needs Expert Dialogue Across Disciplines
By
–
The AI conversation on X can be frustrating as researchers keep bumping into well-understood problems in economics, sociology, history, & psychology that would be useful to know but are hurt by the lack of dialogue with expets (both as they left X & they aren’t part of AI talk).
-
DeepSpeed: Distributed Fine-Tuning Framework for Large Language Models
By
–
3. DeepSpeed • Built for large-scale distributed fine-tuning with ZeRO and FSDP
• Optimized for multi-GPU and multi-node training with advanced memory management
• Trusted in production environments for scalable LLM training GitHub repo: -

Stanford Researchers Decode Brain Activity During Imagined Speech
By
–
With the help of an AI model, sentences from a vocabulary of up to 125,000 words were recognized with up to 74% accuracy. Researchers at Stanford have decoded for the first time the brain activity that occurs when people imagine themselves speaking (known as “inner speech”).
-

GPT-5 Personality: Sandwich Feedback Better Than GPT-4o
By
–
The new GPT-5 personality likes giving sandwich feedback (you are great- suggestion for improvement – you are great). In general, better than GPT-4o at pushing back while being a bit syncophantic. (It would be good for the AI labs to look at the research on giving good feedback)
-
Unverified AGI Claims and LLM Hallucination Concerns
By
–
Por favor que nadie le pida pruebas de cuándo yo he dicho que "X modelo es AGI"… no vayamos a descubrir que los LLMs alucinan, pero que devruso directamente miente para farmear unos pocos likes
-

ChatGPT-5 API Tops LM Arena But Chat Version Ranks Fifth
By
–
Un autre gros problème de ChatGPT-5 qui vient d’être révélé. La version disponible dans le Chat est classée seulement 5ᵉ sur LM Arena, loin derrière Gemini et Claude. La véritable version top 1, c’est celle de l’API. Mais OpenAI n’avait rien dit jusqu’à aujourd’hui, ce qui