科技与人工智能
"You’re right"—LLMs will get climate info wrong if you ask
来源: Ars Technica · 发布于

Many studies over the years have examined how people encounter climate science information. Some focused on the top results that Google surfaced (back when Google was still in the business of serving links), while others examined social media content or newspaper and TV news cove
Many studies over the years have examined how people encounter climate science information. Some focused on the top results that Google surfaced (back when Google was still in the business of serving links), while others examined social media content or newspaper and TV news coverage. With large language models (LLMs) now forming a share of the public information diet, it's worth looking at how they handle climate change.
But there’s no direct way for researchers to look at the information people have seen. It is served ephemerally to individual users, confined to their chat session. A study led by researchers from University College London and the United Nations Educational, Scientific, and Cultural Organization (UNESCO) tried to tackle this question from multiple angles, testing LLMs with different prompts and evaluating their output.
They started with benchmarks of simple factual questions (benchmarks that might even be in models’ training data at this point) but also analyzed more complex interactions—including when users think they’re right to push back on an answer. CO 2 Q&A LLM research often faces ( no-true-Scotsman-tinged ) criticism for focusing on a particular model rather than some other frontier model that may have been released after the experiment concluded. In this case, the study used 33 models, including multiple versions of the major frontier and open-weight models.
The flagship frontier models used were Claude Fable 5, Claude Opus 5, Gemini 3.1 Pro Preview, Gemini 3.6 Flash, and GPT-5.6 Sol. (You can find the full list in the report .) Read full article Comments
相关报道
Google gives Gemini the ability to use your Mac for you
Google's latest update to Gemini brings an AI agent to the Mac that can organize files, fill out forms, and tackle multistep workflows.
TechCrunch ·
Google to keep Gemini chats for 2 months, up from 72 hours, even with history off
Google's Gemini update extending chat retention will roll out globally over the coming weeks.
TechCrunch ·

Google makes it easier to use Gemini without training AI on your data
It can be hard to use Google products—or even the Internet itself—without coming into contact with Gemini AI. That is, of course, by design . The more you use Gemini, the more data Google collects to train and improve its generative AI chatbot. The options to limit training on yo
Ars Technica ·

Gemini on Mac can now use your computer
Google's Gemini app for macOS now has computer use, meaning it can help you "get multi-step tasks done across your local files, desktop apps, and the web," the company says . Google suggests the feature might be useful for things like speeding up workflows, filling in forms, maki
The Verge ·

Google will launch Android app verification requirement globally in January 2027
Google announced Android developer verification last year , and despite numerous objections from developers and users, the company is forging ahead . It began blocking the installation of unregistered apps in some countries this fall, and the system is expanding globally soon. Go
Ars Technica ·