Teach Your Phone to Fix AI Mistakes in One Shot

Kari Jaaskelainen

27 Nov 2025 — 1 min read

AI on our phones still makes embarrassing mistakes. This research shows a simple fix: let users correct misclassifications with just one or two examples—no cloud retraining, no heavy downloads.

Server side: big foundation models teach (distill) strong features into a compact, device-friendly network.
On device: the app classifies using lightweight "prototypes" and updates those prototypes from your feedback—rather than retraining the whole model.

In tests on image classification and object detection, the system corrected over 50% of errors after a single example on Food-101 and Flowers-102, with minimal forgetting (<0.02%) and negligible compute cost. An Android demo app shows it working in practice.

Why it matters: faster fixes, better privacy, longer battery—and AI that adapts to you.

Paper: https://arxiv.org/abs/2511.21652v1

Paper: https://arxiv.org/abs/2511.21652v1

Register: https://www.AiFeta.com

#AI #EdgeAI #OnDevice #Mobile #MachineLearning #ComputerVision #FewShot #Privacy #UX

Read more

Meet GRACE: a moral governor for safer, more transparent AI

Meet GRACE: a moral governor for safer, more transparent AI

AI agents are getting powerful—so how do we make sure they do the right thing, not just the effective thing? Meet GRACE, a reason-based moral governor that keeps AI behavior aligned with human norms by separating moral reasoning from goal-driven decision-making. * Moral Module: uses deontic logic and explicit reasons

Why some fine-tuned LLMs miss phishing—and how to fix it

Why some fine-tuned LLMs miss phishing—and how to fix it

Not all fine-tuned LLMs spot phishing equally. A new study tests Llama 3.1 8B, Gemma 2 9B, and Mistral on high-stakes phishing detection—and uses SHAP and mechanistic interpretability to reveal why models do (or don’t) generalize. * Architecture × data diversity matters: Gemma 2 9B hits state-of-the-art performance (F1

AI that explains itself—by following the science

AI that explains itself—by following the science

AI that explains itself—by following the science Black-box AI can be powerful, but it often can’t tell us why it made a decision. Concept Bottleneck Models (CBMs) try to fix that by predicting human-understandable concepts first, then the final answer. The catch: standard CBMs ignore domain-specific cause-and-effect and

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A

Meet GenomAgent: A Team of AI Specialists for Smarter Genomics Q&A

TL;DR: Finding reliable answers in genomics is hard. GenomAgent turns one big AI into a coordinating team of specialists—and beats the current leader by 12% on a key benchmark. Genomic facts live across many databases. Standard chatbots struggle because they can’t flexibly query those sources. GeneGPT added