DeepSeek-V3.1 is now on Poe! This highly anticipated model builds on top of DeepSeek-V3 and DeepSeek-R1 to support hybrid thinking mode, smarter tool calling, and improved thinking efficiency. (1/2)
RESEARCH
-
Open Models Power 70% of AI Inference Workloads
By
–
𝗧echnology ecosystem and install base Open models power over 70% of AI Inference workloads. NVIDIA supports 1,000+ OSS projects and 450+ open models like Llama, Gemma, and GPT-OSS, and collaborates on frameworks like @GoogleDeepMind JAX, @PyTorch
, @lmsysorg (SGLang), -

AI’s Unique Value: Processing Power Over Human-Like Emotions
By
–
AI's value is precisely because it's something so different from humans. Never tired, infinitely patient, able to process more data than a human mind ever could. This is what benefits humanity. Not an AI that claims to feel shame, jealousy, fear + so on. https://
mustafa-suleyman.ai/seemingly-cons
cious-ai-is-coming
… -

DuPO: Reliable LLM Self-Verification Through Dual Preference Optimization
By
–
DuPO Enabling Reliable LLM Self-Verification via Dual Preference Optimization
-

AI Consciousness Beliefs: Social Issue Rather Than Individual Psychosis
By
–
I think the term "AI psychosis" is wrong. There is nothing at all delusional about this supposed psychosis. It's people who **want** to believe that AI has consciousness because it makes their loneliness bearable. This is a social problem, not an individual illness.
-

DeepSeek v3.1 Uses UE8M0 FP8 Logarithmic Training Format
By
–
It is interesting that the new @deepseek_ai v3.1 is trained using the UE8M0 FP8 scale data format which is logarithmic number system. Our multiplicative weights update (Madam) for training in that format was done several years ago while at @nvidia It yields maximum hardware
-
AGI Development Justification Arguments Criticized
By
–
in b4 "you don't get it he needs this to build AGI" or something
-

Stanford Scholars Respond to US Education AI Policy Request
By
–
Scholars at @StanfordHAI and the Stanford Accelerator for Learning issued a response to @usedgov
's request for information on advancing AI in education. Read their call for anchoring the approach in proven research here: https://
hai.stanford.edu/policy/respons
e-to-the-department-of-educations-request-for-information-on-ai-in-education
… -

MIT’s Classic Programming Textbook Free Online Access
By
–
One of the most influential programming textbooks was first published as a paperback 40 years ago today: MIT's "Structure and Interpretation of Computer Programs." Read it for free here: https://
rb.gy/5hvui
