Can AI Spot Nonsense in Pictures? Meet UAIT

Kari Jaaskelainen

13 Jan 2026 — 1 min read

Can AI spot nonsense in pictures?

Humans can tell that “A sandwich cuts a chef” is grammatically fine but semantically absurd. Many vision-language models (VLMs) can’t. UAIT (Uncommon-sense Action Image-Text) is a new benchmark that stress-tests whether VLMs truly understand who is doing what to whom—and what’s physically possible.

Synthesized uncommon action scenes via large language models, prompt engineering, and text-to-image tools.
Each image–caption pair comes with a carefully designed multiple-choice question for fine-grained reasoning.
Evaluated state-of-the-art VLMs against humans and contrastive baselines.

Findings: All models trail humans by a wide margin, especially when distinguishing grammatical correctness from real-world plausibility. Yet even lightweight models improve notably with targeted fine-tuning—showing strong potential for directional adaptation.

Why it matters: Beyond pattern-matching, next-gen AI must build real visual semantic reasoning. UAIT is both a diagnostic tool and a roadmap for training more robust, commonsense-aware systems.

Paper: https://arxiv.org/abs/2601.07737v1 — by Chen Ling and Nai Ding.

Paper: https://arxiv.org/abs/2601.07737v1

Register: https://www.AiFeta.com

AI ComputerVision VLM MachineLearning Commonsense Dataset Benchmark Research

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

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 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

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

Can AI spot nonsense in pictures?

Read more

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

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

AI that explains itself—by following the science

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