AI Pioneers Split Over Risk As Compute Buildout Accelerates

Three of AI’s best-known researchers used a rare joint appearance at the Ai4 conference in Las Vegas to draw sharply different lines around how advanced AI should be released, regulated and used at work, Data Center Knowledge reported.

The session, billed as “The Architects of Intelligence: A Historic Convergence,” brought together Geoffrey Hinton, Fei-Fei Li and Andrew Ng.

All three agreed that AI will reshape nearly every industry.

They disagreed over how soon it could exceed human intelligence, whether it will replace routine intellectual labor, and how aggressively governments should intervene.

Hinton, a pioneering deep learning researcher and AI safety advocate, said artificial intelligence could surpass human intelligence within five to 20 years.

He argued that AI would eliminate many forms of routine intellectual work, comparing today’s white-collar occupations with manual labor displaced by mechanization.

“If AI can do routine intellectual labor, any job that consists mainly of routine intellectual labor is going to be done by AI,” Hinton said.

He also warned that increasingly capable systems could enable cyberattacks and concentrate power among leading AI companies.

In his view, stronger government oversight is needed before those systems are deployed.

Ng, the founder of DeepLearning.AI and Landing AI, challenged the idea that AI is already causing widespread job losses.

Current evidence, he argued, points instead to workers using AI tools to become more productive.

The systems automate individual tasks while allowing employees to take on broader responsibilities, rather than replacing entire professions at once.

Li, co-director of the Stanford Institute for Human-Centered AI, called for a more measured debate.

She said extreme narratives were obscuring practical questions about deployment, education and policy.

AI should primarily be treated as a tool that augments human capabilities, she argued, while existing rules are updated for sectors such as healthcare, transportation, finance and education.

Li also called for greater public investment in AI research and education, describing AI as foundational infrastructure whose long-term development should not be driven solely by private companies.

The sharpest disagreement concerned open-weight models.

Ng defended them as essential to innovation, competition and broader access, warning against allowing a handful of companies to control advanced AI technology.

Hinton said releasing model weights could make it easier for malicious actors to adapt advanced systems for cyberattacks and other harmful uses.

Li rejected a simple choice between open and closed models.

Different applications, she said, require different levels of openness depending on their risks.

The debate offered no common timetable or regulatory formula.

Its disagreements now sit alongside the capital being committed by data center developers, utilities and infrastructure investors, who are betting that demand for AI compute will continue rising for years.

The rules governing access, safety and deployment could influence how quickly that capacity becomes working infrastructure.