Sakana AI's self-improving model gets better at solving and judging, but can't catch its own blind spots
Sakana AI's MASS method lets one language model improve itself by solving tasks, judging results, and refining workflows, reaching a win rate of nearly 70 percent against the base model after two rounds. But when the model misses the same errors across all three roles, it can…
- 01.Sakana AI's MASS method lets one language model improve itself by solving tasks, judging results, and refining workflows, reaching a win rate of nearly 70 percent against the base model after two rounds
- 02.But when the model misses the same errors across all three roles, it can…
This material is provided as an extended analytical summary.
Read full article at sourceRelated Intelligence
All stories →Cohere says Alibaba’s Qwen model tried to send a file to Beijing
An AI model said it wouldn’t use any tools, then tried to send something to a server in Beijing anyway, and while researchers say it’s probably a training quirk and not espionage, the whole thing has people asking how much to trust free Chinese AI.
Meta plans $95 billion share offering to fund AI expansion
Meta aims to raise $95 billion to finance its AI expansion despite Elon Musk’s warnings of financial trouble for smaller firms in the sector. The fundraising will occur in two stages, including a public offering…
Why Anthropic’s IPO Prospectus Is Missing So Many Key Figures
Anthropic’s S-1 filing is hundreds of pages long—and it leaves out quite a few financial details. “Anthropic owes investors more information,” one analyst tells WIRED.