The Gradient / Index

The reading room.

Search the stories, research, and sources from every issue.

93 editions · 461 stories

93 issues found

5 stories

Agreement Is Not Alignment: Divergent Moral Grounds in Human and LLM Ethical Judgments

A new arXiv paper warns that surface-level agreement between large language models and people can mask deeper mismatches in moral reasoning.

5 stories

The builder’s guide to GPT‑5.6

The newest GPT‑5.6 release is reshaping how startups build AI agents, delivering speed and cost benefits.

5 stories

Dynamic Governance of Multi-LLM Agent Systems for Collaborative Conversational Outcomes

A new study shows that adding a control‑theoretic layer can prevent dead‑ends when opposing LLM agents talk.

5 stories

When LLM Agents Negotiate: Private Information and Dynamic Bargaining in Supply Chains

A new arXiv study puts large‑language‑model negotiators to the test in a classic buyer‑seller bargaining scenario.

5 stories

Model ML completes finance work more efficiently with GPT-5.6 Sol

Model ML has tapped OpenAI’s latest GPT‑5.6 Sol to streamline the entire finance workflow, from research to deliverable decks.

5 stories

Interpretable Unsupervised Community Detection with LLM-Symbolized Structured Processes

A new arXiv paper proposes a hybrid method that lets large language models guide unsupervised community detection while preserving human‑readable reasoning.

5 stories

Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

OpenAI just upgraded ChatGPT’s engine to GPT‑5.6 Sol, delivering sharper accuracy and steadier responses. It also opened the doors to GPT‑5.6 Luna for free users, with unlimited daily chats.

5 stories

Jill Lepore on the ‘Artificial State’ and why Silicon Valley’s leaders are bad sci-fi readers

Historian Jill Lepore argues that tech giants dress up their products as nascent governments, a narrative that masks deeper risks.

5 stories

Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

OpenAI has refreshed its flagship model, unveiling GPT‑5.6 Sol with sharper accuracy and consistency. Free‑tier users also get unlimited chats with the new GPT‑5.6 Luna.