#исследования
-
Left to their own devices, AI agents started playing
MIT researchers dropped 13 AI agents onto a virtual desert island with no task at all — and they began inventing their own games. A side effect: the ones that played ended up better at their jobs.
-
OpenAI turns to elite mathematicians to avoid fumbling again
After a string of headline mathematical results curdled into a reputational crisis, OpenAI has convened an independent panel of nine prominent mathematicians to advise it on how to publish discoveries and talk to the research community.
-
Language models develop new social biases through adaptive exploration
A new paper, published on OpenReview, argues that large language models are capable of forming their own social biases — not inherited from data, but arising in the course of adaptive behavior.
-
AI as a Cognitive Virus: An Epidemic Model of LLM Dependence
An international team of researchers proposes describing the spread of large language models by analogy with a viral infection — complete with points of no return and the risk of irreversibly losing cognitive independence.
-
Anthropic shows off AI that improves its own “alignment”
Anthropic researchers have described a system in which automated agents, with no human involvement, boost a model's scores across all ten tests for undesirable behavior — faster and cheaper than people.
-
AI hits newcomers hardest: youth employment in affected occupations down 19%
An updated study by Stanford economists shows that artificial intelligence is markedly shrinking opportunities for entry-level workers, while barely touching experienced ones.
-
A Model That Knows the World No Further Than Fifth Grade
Researchers trained language models on a corpus limited to the elementary-school curriculum in order to separate the acquisition of knowledge from its retrieval. The takeaway: the filter on the training data sets a ceiling on capability.
-
AI Tips Made People Three Times Less Accurate but Twice as Confident
Researchers from French and Italian universities have found that access to AI advice all but strips people of their ability to admit “I don't know” — and sharply lowers the accuracy of their answers to boot.
-
Study: AI accelerates scientists' careers but impoverishes scientific discovery
The use of artificial intelligence in science fosters researchers' career growth, yet at the same time it reduces the diversity and depth of scientific discoveries. That is the conclusion reached by the authors of a new study.