85 AI Terms Explained
→ View original post on X — @marcusborba, 2025-11-29 05:06 UTC

By
–
85 AI Terms Explained
→ View original post on X — @marcusborba, 2025-11-29 05:06 UTC

By
–
What if LLMs could learn tool use faster by fixing their own training data gaps? SJTU & Xiaohongshu just built LoopTool for this. It closes the loop between data synthesis and model training, using the model’s own weaknesses to refine datasets adaptively. It probes what the

By
–
Ever wondered how to fix the weird artifact tokens that mess up Vision Transformers’ fine-grained tasks—without retraining those massive models from scratch? HKU, Zhejiang & NTU introduce PH-Reg, a post-hoc method that adds register tokens to pre-trained ViTs via

By
–
From Agent-1 to Superintelligence: The AI 2027 Scenario The AI 2027 Report outlines one of the most thought-provoking trajectories for artificial intelligence: a rapid evolution from simple assistants (Agent-1) to autonomous, adversarially misaligned systems (Agent-4) and
By
–
Extremely excited about our internal evals! Abacus AI Desktop is topping another very important bench We will announce shortly

By
–
One point I made that didn’t come across:
– Scaling the current thing will keep leading to improvements. In particular, it won’t stall.
– But something important will continue to be missing. x.com/slow_developer…
By
–
A study published in The Lancet Digital Health found that the accuracy of all eight AI algorithms in detecting diabetic eye disease that may require clinical intervention ranged from 83.7% to 98.7%. https://
bit.ly/4itRu9p
#AI #Health #TheLancet

By
–
AI now exceeds human performance in most technical tasks — but humans still lead in multimodal reasoning The next frontier isn’t about replacing humans, but amplifying how we think across data, images & language. Source: Stanford AI Index 2025
#AI #Innovation #FutureOfWork

By
–
What if we could model human visual cortex responses without expensive, time-consuming fMRI data for every new subject? This study introduces BraInCoRL, a transformer that uses in-context learning to predict voxelwise neural activity from just a few examples—no extra finetuning