#DeepSeek's groundbreaking efficiency innovations validate what we've been saying all along – our At-Memory compute architecture is the future of AI. Their focus on spatial networks and optimized data streaming? That's our specialty. https://
untether.ai/the-deepseek-m
oment-significance-motivation-and-industry-impact/
…
RESEARCH
-
DeepSeek’s Efficiency Innovations Validate At-Memory Compute Architecture
By
–
-
Personalized Neoantigen Vaccines Show Promise Cancer Treatment
By
–
The promise of a personalized (neoantigen) vaccine for treating cancer, as seen by immune response bolstering against renal cell carcinoma in a Phase 1 trial https://
nature.com/articles/s4158
6-024-08507-5
… https://
nature.com/articles/d4158
6-025-00308-8
… @DrChoueiri @Nature -

Gemini models dominate LMarena rankings
By
–

New Gemini family models are now taking 1st, 3rd and 9th spots on lmarena Gemini 2.0 Flash thinking exp 0121 is holding the top score
-
LLM Bias and Grounding Issues in AI Medicine Applications
By
–
That's right. We wrote about it here https://
nytimes.com/2025/02/02/opi
nion/ai-doctors-medicine.html
… bias (automation neglect) and lack of grounding with use of LLMs may contribute -

Brain cells regulating hunger discovered via spatial omics
By
–
Discovery of the specialized brain cells that convey and regulate hunger/satiety
The cells integrate input of food (visual, smell, texture) and gut hormone circuitry. Another jump forward through spatial omics. https://
cell.com/cell/fulltext/
S0092-8674(25)00047-9
… @CellCellPress @cpnoc @Columbia -
How AI Represents Knowledge Differently Than Humans
By
–
Epistemology—the study of knowledge—has always been a human endeavor. But AI represents knowledge in ways we barely understand. It doesn’t store facts like us. It doesn’t “reason” like us. Instead, it encodes vast statistical relationships in high-dimensional space. This
-
Mechanistic Interpretability: Understanding How LLMs Think
By
–
Fascinating insight from @tegmark on how LLMs ‘think.’ Because AI is a ‘black box,’ mechanistic interpretability—the emerging study of what goes on under the hood—is vital. One only wonders if our study of AI epistemology can keep up with AI itself. I don’t think it can.
-

Third Way in AI: Stanford Expert’s Davos 2025 Insights
By
–
A key theme at #Davos2025: the rise of the "third way." @StanfordHAI Center Fellow @alex_pentland shares his insights from the forum last month: https://
stanford.io/3Q5hXwI -

GPT-4 Enhances Physician Performance in Clinical Care Tasks
By
–
A randomized trial of GPT-4 vs 92 physicians with or without this #AI LLM for performance on patient care tasks.
AI improved physician performance, on par with AI alone (based on 5 clinical vignettes) https://
nature.com/articles/s4159
1-024-03456-y
… @NatureMedicine @AdamRodmanMD @jonc101x -
Next Generation Models for Tabular Data Problems
By
–
Big congrats, Frank, that's very exciting. Actually, I believe a huge portion of real-world problems are tabular data problems. And I feel like we were stuck there with XGBoost for a decade, it's exciting to finally see a next gen of models for tabular data problems!