DeepSeek designs AI agent system to cut human oversight, SCMP reports
The Chinese AI company's DeepSeek Harness team has published a paper proposing a system that shares coding tasks among agents so people only review results, the South China Morning Post reports.
Why it matters: DeepSeek is a low-cost rival to OpenAI and Anthropic, and its models have already drawn attention in crypto trading experiments.
Key points
- DeepSeek Harness published a paper on coordinating multiple AI agents with less human input, reported by the South China Morning Post on Sunday
- The proposal comes as DeepSeek competes with US developers including Anthropic and OpenAI
DeepSeek has set out a design for giving AI coding assistants more autonomy, so they can handle several tasks with less human involvement. The DeepSeek Harness team published the proposal in a paper reported by Singapore's South China Morning Post on Sunday.
The team argues that managing several agents at once is limited by human attention rather than by model capability. The paper says a person can supervise only three to five active AI agent sessions simultaneously. Under the proposed system, tasks would be spread among agents and their work coordinated, leaving humans to review results and make decisions.
Live updates on NewsFeeds:AIDesk editor: Keshav Aggarwal (accountable for this page; this summary was not edited by hand)
How this was produced: an AI model wrote this summary of Cointelegraph’s report, which we credit and link above. Read our Editorial and AI policy. Spotted a mistake? See Corrections.
Not investment advice. Read the risk disclaimer.