AGI is just a few years away, warns DeepMind CEO
What's the story
Demis Hassabis, the CEO of Google DeepMind, has warned that artificial general intelligence (AGI) could be just a few years away. In a recent blog post, the Nobel Prize-winning AI researcher emphasized that we are nearing a tipping point in AI development. He defined AGI as "a system that exhibits all the cognitive capabilities the brain has."
Regulatory concerns
Concerns over AI's rapid advancement
Hassabis's prediction of AGI's arrival in three to four years echoes his earlier statements from June.
However, he cautions that the focus on AI's rapid advancement is overshadowing its associated risks.
He believes governments and tech companies are more concerned about the competitive race to build powerful models than their potential dangers.
Security threats
Potential risks of advanced AI systems
Hassabis warns that frontier AI systems are already posing cybersecurity challenges.
He fears future iterations of these systems could pose even bigger problems, including biological and nuclear security risks.
"Urgent action is needed to address risks that might arise as we get closer to AGI," he wrote in his blog post.
Control measures
Need for an independent AI regulatory body
Hassabis also stresses the need for robust mechanisms to control increasingly autonomous AI systems.
This is especially important if future models can improve themselves with little human involvement.
He proposes the establishment of an independent regulatory body, similar to the Financial Industry Regulatory Authority (FINRA), to oversee AI development and deployment.
Oversight proposal
Proposed oversight body composition and functions
Hassabis's proposed AI oversight body would be composed of leading technical experts, primarily funded by the AI industry.
The regulator would establish safety standards that frontier AI models must meet before deployment.
Initially, developers would voluntarily submit their most advanced models for evaluation up to 30 days before release.
Over time, as testing procedures mature and prove reliable, participation could become mandatory.
Future prospects
Optimism about AI's potential to drive progress
Despite the uncertainties surrounding AI development, Hassabis remains optimistic about its potential.
He believes that responsibly developed AGI could revolutionize scientific discovery, healthcare, clean energy, and advanced materials. It could also help solve some of humanity's biggest problems.
However, he acknowledges that the future is uncertain and experts disagree on both AGI's timeline and capabilities.