Chinese artificial intelligence company DeepSeek has unveiled a design proposal aimed at making AI coding assistants more autonomous when handling multiple tasks, with the stated goal of reducing the amount of human coordination and monitoring required. According to a Sunday report by the South China Morning Post, the proposal comes from a paper published by DeepSeek’s Harness team, which focuses on the operational burden created by supervising several AI agents at once. The paper argues that the ceiling for single-task performance depends on model capability, while the ceiling for multitasking depends on human attention. It also says one person can supervise only three to five active AI agent sessions at the same time. Under the proposed system, tasks would be assigned across agents and their work coordinated so that human operators can spend more time reviewing outputs and making decisions. The proposal arrives as competition among AI developers continues, including U.S. companies Anthropic and OpenAI. Cointelegraph reported in October last year, citing CoinGlass data, that Chinese AI models outperformed U.S. peers in cryptocurrency trading tests, with DeepSeek posting the only positive unrealized return at 9.1% in one experiment.
Chinese artificial intelligence company DeepSeek has introduced a design proposal intended to make AI coding assistants more autonomous and less dependent on human intervention when carrying out multiple tasks.
According to a Sunday report by the South China Morning Post, the proposal comes from a paper published by DeepSeek’s Harness team. Its central aim is to reduce the human labor needed to coordinate and monitor multiple AI agents.
Paper points to human attention as the main bottleneck in multi-agent work
The DeepSeek Harness team said that managing multiple agents consumes human attention. The paper states, 「the upper limit for a single task depends on model capability, while the upper limit for multiple tasks depends on human attention.」
The paper also says that one person can supervise only three to five active AI agent sessions at the same time. Under the proposed system, tasks would be distributed across agents and their work coordinated, leaving humans to focus more on reviewing results and making decisions.
Proposal lands during intense AI developer competition
The proposal comes as competition among AI developers remains intense, including U.S. companies Anthropic and OpenAI.
Cointelegraph reported in October last year that data from blockchain analytics platform CoinGlass showed Chinese AI models outperforming U.S. counterparts in cryptocurrency trading. In one experiment, DeepSeek was the only AI model to post a positive unrealized return, at 9.1%, while its development cost was only a fraction of that of its U.S. rivals.
This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan. Disclaimer:
The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.
Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.
