Teaching an AI a brand-new task from just one labeled example is fast—but how sure can you be about its answers? CAOS (Conformal Aggregation of One‑Shot Predictors) is a new method from Maja Waldron that adds reliable confidence to one‑shot learning. Instead of trusting a single quick‑
AI
AI meets messy spreadsheets New research shows that large language models (LLMs) struggle when tables are subtly distorted—through small layout or labeling tweaks—even if the data is otherwise standard. * LLMs rarely notice or fix these distortions on their own. * Giving an explicit "watch out for table errors&
AI
LLMs with Retrieval-Augmented Generation (RAG) work better when they read lots of context — but long prompts are slow and pricey. Compressing the context helps, yet models often lose the thread and answer worse. ArcAligner is a lightweight module that helps models make sense of highly compressed context. It "aligns&
Agent-as-a-Judge: the next step in trustworthy AI evaluation As AI tasks get more complex, one-shot model graders fall short—biased, shallow, and unable to check the real world. This new survey traces the shift from LLM-as-a-Judge to agentic judges that plan, use tools, collaborate, and remember—making evaluations more robust
AI
Sometimes, vision-language AIs see what the prompt says, not what is in the picture. New research maps how that happens—and how to dial it down. Rudman and colleagues tested object counting with prompts that intentionally overstate what is in an image (for example, "describe four waterlilies" when
Robots learn from videos of their own actions—but collecting lots of diverse, multi-camera footage is slow and costly. RoboVIP is a generative approach that “films” new training scenes without touching the hardware. * Visual identity prompting: Instead of vague text, the model is guided by example images of the robot,
Why running LLMs is hard (and how hardware can help) Large language models don't just need fast math—they need to fetch and share enormous amounts of data, one token at a time. In inference, the autoregressive "decode" phase dominates, making memory and interconnect, not compute,
Meet Atlas 2 — a new family of pathology vision foundation models designed to close the gap to clinical deployment. Why it matters: Previous models often forced a trade-off between accuracy, robustness, and compute cost. Atlas 2, Atlas 2-B, and Atlas 2-S aim to balance all three. * Trained on 5.5
AI
As AI models start learning from their own outputs, they risk entering an echo chamber. A new study names this cycle the Self-Consuming Performative Loop (SCPL) and shows how it can warp model behavior over time. * What’s the loop? Deployed models shape what people ask and which data gets
AI
Personalized AI you can read, edit, and take anywhere Most AI systems hide your preferences in opaque vectors. This work makes them plain English instead. Your "profile" becomes a short text summary the model can use across apps and tasks. * Interpretable: you can see and revise what the
When robots know when to help Assistive robots shouldn't guess. They need to know what you plan to do, and how sure they are, before stepping in. Researchers Johannes A. Gaus, Winfried Ilg, and Daniel Haeufle show that raw AI "confidence" can be misleading. Their fix:
Meet DrivoR, a simple, efficient transformer for end-to-end autonomous driving. It builds on pretrained Vision Transformers and adds camera-aware "register tokens" that compress multi-camera views into a compact scene summary, cutting downstream compute without sacrificing accuracy. DrivoR then runs two lightweight decoders: one to propose driving paths, and