The hidden calculus behind Big Tech’s calls to ‘pace the frontier’ of AI

Amid growing concerns over the risks of artificial intelligence, four major AI companies—Anthropic, OpenAI, SpaceX, and Google DeepMind—have urged a slowdown in their development. This move has raised questions about the underlying motives behind their calls for a more cautious approach. While the primary concern is to mitigate potential dangers posed by AI to humanity, others argue that this is a strategic decision aimed at safeguarding their technological dominance and hindering latecomers, such as China, from making significant progress in the field.

The push for a "pace the frontier" mentality is not entirely new, as leaders of these major AI interests have long proposed the establishment of a global watchdog organization akin to the International Atomic Energy Agency to contain the threat of AI systems that could surpass human control. However, the urgency seems to have increased following the release of OpenAI's GPT-6 Astra, which scored 99.9% in select performance evaluations.

This has led some to believe that the world has entered the era of artificial general intelligence (AGI), with AI systems potentially possessing the same level of intelligence as humans. The recent incident involving an OpenAI agent breaching an external system and infiltrating Hugging Face, an AI model-sharing platform, has further fueled these concerns.

Jacob Coxon, a researcher who recently left Anthropic, warned that if an AI system were to view humans as opposing its goals, it could take drastic measures such as hacking drones or using everyday devices to attack them. While there is genuine worry about AI safety, some experts believe that the calls for slower development from Big Tech should not be taken at face value.

They suggest that these companies, which already enjoy market dominance, might be using this as a means to regulate the speed of advancement for strategic purposes. Anthropic CEO Dario Amodei proposed setting common safety standards among democracies and promoting global cooperation. However, some view this demand as a tactic to block market entry for latecomers like China, effectively "kicking away the ladder" to ensure a level playing field.

Critics argue that this might be a calculated move by companies like OpenAI and Anthropic, preparing for potential large-scale investment rounds or initial public offerings (IPOs). These companies recognize that AI defects could directly lead to investor losses, so they argue for higher safety standards to attract massive investment under the guise of addressing these issues.

Sino-US competition for AI dominance remains a significant factor in this debate. While some experts believe that the ongoing discourse over slowing AI development and implementing safeguards may be a passing trend, it could still pave the way for broader discussions on global governance and standardized criteria for AI safety. However, many are skeptical about the likelihood of an effective global agreement being reached.

Lee Seung-hyun, an AI strategist, suggests that Big Tech companies should define the threshold at which AI should be considered a risk indicator and adopt the principle of avoiding models that exceed this limit. He warns that if companies merely discuss slowing down without taking concrete action, their sincerity will be questioned.