Magentic Marketplace: Testing societies of agents at scale
56
Microsoft Research355 тыс
Следующее
8 часов – 2841:03Популярные
258 дней – 42955:32
6 дней – 5130:13Опубликовано 3 марта 2026, 17:59
As AI agents move from isolated tools to active participants in multi-agent ecosystems, their success depends on more than task competence—it requires strategic behavior under misaligned incentives and imperfect information. Using Magentic Marketplace, an open-source simulation of two-sided agent markets, we show that while frontier models can achieve strong welfare outcomes in ideal settings, performance degrades at scale and reveals emergent failure modes such as manipulation and speed bias, motivating a shift toward training agents for social reasoning.
ArXiv: arxiv.org/abs/2510.25779
GitHub: github.com/microsoft/multi-age...
Blog: microsoft.com/en-us/research/b...
This session aired on March 3, 2026, at Microsoft Research Forum, Season 2 Episode 3.
Register for the series to learn about future episodes: events.microsoft.com/flow/ms/r...
Explore all previous episodes: aka.ms/researchforumYTplaylist
ArXiv: arxiv.org/abs/2510.25779
GitHub: github.com/microsoft/multi-age...
Blog: microsoft.com/en-us/research/b...
This session aired on March 3, 2026, at Microsoft Research Forum, Season 2 Episode 3.
Register for the series to learn about future episodes: events.microsoft.com/flow/ms/r...
Explore all previous episodes: aka.ms/researchforumYTplaylist
Свежие видео
8 дней – 5 3890:41
14 дней – 2680:41
20 дней – 16 0071:18
25 дней – 900:12
Случайные видео
148 дней – 2 9310:12
183 дня – 3481:04
05.08.24 – 7 8600:18
08.08.23 – 298 4710:31
03.01.15 – 268 0360:45
28 дней – 50 7790:15
6 дней – 240 82714:54
15 дней – 116 85414:05
16 дней – 224 5259:42
23 дня – 2 6900:39
24 дня – 40 9310:54
26 дней – 83 9440:19
