Credit: Unsplash/CC0 Public Domain When ChatGPT and other generative artificial intelligence can produce scientific articles that look real—especially to someone...

An artistic representation of the large language model trained on the DNA sequences. Credit: Magdalena Gonciarz, generated with Dall-E3 DNA...

Some examples of how words connect to each other in a text, forming a network. While words such as "vertex"...

An illustration of ripple effects in LLM knowledge editing. Our work empirically demonstrates the positive correlation between gradient similarity explains...

Credit: Pixabay/CC0 Public Domain Whether it's the Oscars, the Tonys, or the Grammys, observers annually make predictions as to which...

Thermometer performance vs. number of training tasks. Thermometer's calibration performance (average ECE over fifty seven MMLU tasks) improves as the...

An illustration of the SELF-REASONING framework for improving the RALMs. The upper is the basic LLMs which answer the question...

Generative artificial intelligence (AI) models trained on synthetic data generate outputs that are progressively marred by artifacts. In this example,...

Anomaly detection poses several challenges that are not seen in traditional AI problems. Credit: Trevor Vannoy on Unsplash An algorithm...

Density of Predictions by Point Estimate pˆ(xi) & Uncertainty Metric (Darker Colors = Higher Density). Credit: arXiv (2024). DOI: 10.48550/arxiv.2404.08592...

MAIA framework. MAIA autonomously conducts experiments on other systems to explain their behavior. Credit: arXiv (2024). DOI: 10.48550/arxiv.2404.14394 As artificial...

Strong fluctuations across AIW problem variations. Also for higher performers, eg GPT-4o, GPT-4 and Claude Opus 3, correct response rates...

Credit: Chen et al. Recent advances in the field of artificial intelligence (AI) and computing have enabled the development of...

Credit: CC0 Public Domain Researchers at USC have developed a new method to accurately predict wildfire spread. By combining satellite...

A predictive coding neural network explores a virtual environment. In predictive coding, a model predicts observations and updates its parameters...

Schematic representation of scissor-type deployable mast system. Credit: International Journal of Mechanical System Dynamics (2024). DOI: 10.1002/msd2.12102 Dynamical systems describe...

Original representation and proposed temporal shift . Credit: Intelligent Computing (2024). DOI: 10.34133/icomputing.0073 Humans can guess how someone on the...

VLMs cannot reliably count the intersections between the blue and red plots. Credit: arXiv (2024). DOI: 10.48550/arxiv.2407.06581 A trio of...

MIT researchers examined how LLMs fare with variations of different tasks, putting their memorization and reasoning skills to the test....

LlavaGuard judges images for safety alignment with a policy providing a safety rating, category, and rationale. Credit: arXiv (2024). DOI:...

Path planning and motion control PlaMo animates a humanoid in a complex 3D scene. The input consists of a physically...

Left to right: Researchers Austin Youngren, Carlos Mabrey, and Mariah Bradford from the Department of Computer Science work together to...

Overview of GenSQL. Credit: Proceedings of the ACM on Programming Languages (2024). DOI: 10.1145/3656409 A new tool makes it easier...

An example of ChatGPT's analysis of deepfake images. The large language model was less accurate than state-of-the-art deepfake detectors, but...

Pictured: first column: Video frames taken from YouTube and fake videos generated from Sora by OpenAI; second column: frames reconstructed...