July 2025

Multiagent AI for generating chain-of-thought training data Using ensembles of agents to generate and refine interactions annotated with chains of thought improves performance on a battery of benchmarks by an average of 29%. Conversational AI Charith Peris Tharindu Kumarage July 31, 05:18 PM October 10, 06:52 PM Chain-of-thought reasoning, in which a large language model (LLM) is asked not only to perform multistep actions but to explain its reasons for taking the steps it does, has been shown […]

Ver mais

Measuring the effectiveness of software development tools and practices New cost-to-serve-software metric that accounts for the full software development lifecycle helps determine which software development innovations provide quantifiable value. Economics Willie deButts July 29, 12:05 PM July 29, 01:18 PM At Amazon, we constantly seek ways to optimize software development tools, processes, and practices in order to improve outcomes and experiences for our customers. Internally, Amazon has the variety of businesses, team sizes, and technologies to enable research […]

Ver mais

Introduction I really like working on unsupervised learning problems. They offer a completely different challenge to a supervised learning problem — there’s much more room for experimenting with the data that I have. It’s no wonder that the majority of developments and breakthroughs in the machine learning space are happening in the unsupervised learning domain. And one of the most popular techniques in unsupervised learning is clustering. It’s a concept we typically learn early on in our machine learning journey and […]

Ver mais

Pushing the boundaries of secure AI: Winners of the Amazon Nova AI Challenge University teams battle to harden and hack AI coding assistants in head-to-head tournament  Conversational AI Staff writer July 23, 08:31 PM August 04, 07:32 AM Since January 2025, ten elite university teams from around the world have taken part in the first-ever Amazon Nova AI Challenge, the Amazon Nova AI Challenge Trusted AI. Today, were proud to announce the winners and runners up of this […]

Ver mais

Mitra: Mixed synthetic priors for enhancing tabular foundation models Generating diverse synthetic prior distributions leads to a tabular foundation model that outperforms task-specific baselines. Machine learning Xiyuan Zhang Danielle Maddix Robinson July 22, 01:40 PM February 19, 12:55 PM Tabular data powers critical decisions across domains such as healthcare, finance, e-commerce, and the sciences. The machine learning methods traditionally used for tabular data, however such as random forests and XGBoost typically result in models tailored to individual datasets, […]

Ver mais

Pruning network nodes on the fly to improve LLM efficiency Language models inspired by specialized processing regions in the brain offer significant time and cost savings. Conversational AI Jing Liu Grant Strimel July 21, 01:52 PM July 21, 01:52 PM Foundation models (FMs) such as large language models and vision-language models are growing in popularity, but their energy inefficiency and computational cost remain an obstacle to broader deployment. To address those challenges, we propose a new architecture that, […]

Ver mais

I am teaching CS 2881: AI Safety this fall at Harvard. This blog is primarily aimed at students at Harvard or MIT (where we have a cross-registering agreement) who are considering taking the course. However, it may be of interest to others as well. For more of my thoughts on AI safety, see the blogs Six Thoughts on AI safety and Machines of Faithful Obedience.  IMPORTANT: At the end of this blog is Homework Zero. If you are […]

Ver mais

FalseReject: Reducing overcautiousness in LLMs through reasoning-aware safety evaluation Novel graph-based, adversarial, agentic method for generating training examples helps identify  and mitigate “overrefusal”. Conversational AI Zhehao Zhang Weijie Xu July 18, 01:51 PM September 04, 08:35 AM Large language models (LLMs) have come a long way in enforcing responsible-AI standards through robust safety mechanisms. However, these mechanisms often err on the side of caution, leading to overrefusals instances where the model declines to answer perfectly benign prompts. This […]

Ver mais

Unsupervised, generalizable method for doing anomaly detection An ensemble of models, weighted according to their reluctance to flag anomalies, outperforms its predecessors. Machine learning Abishek Sankararaman July 15, 12:17 PM July 15, 12:17 PM In many of todays industrial and online applications, identifying anomalies rare, unexpected events in real-time data streams is essential. Anomalies can indicate manufacturing defects, system failures, security breaches, or other significant events. The typical machine-learning-based anomaly detection system is trained in a supervised manner, […]

Ver mais