[Anchor]

Debates are heating up over claims that the pace of artificial intelligence development should be slowed down. The chief executives of leading AI companies, Anthropic and Nvidia, have also clashed by expressing contrasting views.

Reporter Choi Seung-hun analyzes the complex motives hidden behind this debate.

[Reporter]

[Dario Amodei / CEO of Anthropic: We must examine ourselves first. We need to improve practices, enhance transparency, and invest more in safety.]

[Jensen Huang / CEO of Nvidia: Safety is an engineering problem to be solved. We don't need new laws or new regulations.]

Anthropic CEO Dario Amodei has voiced concerns over the steep development speed of AI and argued that safety measures are necessary, while Nvidia CEO Jensen Huang maintains that matters should be left to the autonomy of businesses and the market.

The debate was ignited by a warning from a researcher who recently left Anthropic.

The researcher claimed that the industry is racing toward "self-improving superintelligence" where AI enhances its own performance, stating, "Developers truly believe this technology could kill all of us within 10 years."

Anthropic's head of safety research also stated that the probability of human extinction within 10 years is seen as over 10%.

Following this, Amodei stepped up to advocate for adjusting development speeds, and subsequently, chiefs of rival companies such as OpenAI and Google DeepMind chimed in as well.

Experts remain sharply divided on just how dangerous AI actually is.

While some experts put the possibility of human extinction in the double digits, other researchers view it as lower than the probability of an asteroid colliding with Earth.

Calculations grow complex at the individual corporate level.

Leading AI companies can reduce massive additional investments by slowing down the development of cutting-edge models while continuing to generate revenue with AI models they have already released.

For AI companies preparing for an initial public offering, it could also serve as an opportunity to boost corporate value and improve financial structures.

On the other hand, companies that must design and sell semiconductors and chips have no reason to welcome a slowdown.

Concerns over "regulatory capture" are also being raised, suggesting that if leading AI firms establish safety standards themselves, it could act as a high barrier to entry for latecomers lacking capital and technological power.

Furthermore, this is intertwined with political considerations that if the United States slows down while China continues development, the U.S. might lose its technological dominance.

Ultimately, behind the AI safety debate and theories of doom lie fierce power struggles over who gets to dictate the pace of development and set the rules.

(Video Editing: Park Na-young, Design: Park Tae-young)