San francisco: Claude, co-founder and chief policy officer of Anthropic, an artificial intelligence (AI) startup, points out that humanity needs to find ways to slow the development of AI, as the technology is reaching a point where it can invent, train, and upgrade itself without human intervention.

According to Thai News Agency, Jack Clarke, co-founder and head of policy at Anthropic, the startup developing the Claude-based AI, revealed that currently, more than 80% of the code running on the system is written by the Claude model itself. He predicts that within the next two years, the AI will be able to write and run 100% of its own code using AI assistance. This will allow Anthropic engineers to deliver eight times more code per quarter than during the 2021-2025 period. The current AI capabilities are improving exponentially every four months, significantly faster than the previously estimated seven months. This demonstrates that AI can learn and evolve to become smarter, even creating next-generation models to replace existing ones, without human intervention. Anthropic's report indicates that this event may occur much sooner than many institutions have predicted.

Anthropic suggests that major global AI developers, along with governments, should collaborate to create a control mechanism and temporarily halt development to allow social structures and security research to keep pace with the technology. However, this proposal faces significant obstacles due to verification mechanisms. The Wall Street Journal reports that it's similar to a nuclear arms control treaty, but far more difficult to monitor because stealthily running systems to train hidden models is easier than building silos to store missiles. Furthermore, the White House and some Silicon Valley giants believe that independently slowing development could permanently put the United States at a strategic disadvantage in the technology battle against China.

Some analysts note that this proposal comes at a time when Anthropic has just raised $65 billion in its Series H funding round and is preparing for an initial public offering (IPO) with a valuation nearing $1 trillion. Calls to slow down AI development now may be a strategy to help Anthropic maintain its leadership in the enterprise market and prevent competitors from catching up.