Chinese artificial intelligence company DeepSeek has outlined a design aimed at making AI coding assistants more autonomous. The proposed system would enable multiple agents to carry out tasks and coordinate their work while reducing the amount of human supervision required.
According to a paper published by DeepSeek’s Harness team and reported by the South China Morning Post, the goal is to limit the human effort needed to manage and monitor several AI agents at once. The team said that the number of tasks a model can handle depends on its capabilities, while the number of simultaneous tasks a person can supervise is constrained by human attention.
The paper estimates that one person can oversee only three to five active AI agent sessions simultaneously. DeepSeek’s proposal would distribute work among agents and manage their collaboration, allowing people to focus on reviewing results and making decisions.
The proposal emerges as AI developers, including US companies Anthropic and OpenAI, compete to advance increasingly autonomous systems, the South China Morning Post reported.
DeepSeek has also drawn attention in cryptocurrency-related AI tests. Cointelegraph reported last October that Chinese AI models outperformed US counterparts in a crypto trading experiment based on data from blockchain analytics platform CoinGlass. DeepSeek was the only model in that test to produce a positive unrealized return, at 9.1%.
Sources: Cointelegraph.