New report warns AI could exceed humans by 2031

By Lauren Wilkin

The Alan Turing Institute has released a report warning people that "superintelligent" AI systems could overtake human capabilities in just five years.

The report, titled "Frontier AI Risks: A practical way forward," makes the case that we are not powerless to act against the worst case scenarios of AI development because many of the underlying capabilities, behaviors and harms of AI can be observed, studied and acted on.

The report by Turing - the UK's independent national institute for data science and artificial intelligence - identifies five main risks of the technology, including:

• Cybersecurity
• Democratic instability
• Misalignment
• Loss of meaningful human control and oversight
• Chemical and biological threats

The report stated that: "There is a realistic possibility that 'superintelligent' AI systems emerge in the next five years, exceeding humans across all cognitive tasks."

It goes on to say: "If AI systems devise strategies that institutions cannot adequately evaluate and act faster than organizations can respond, substantive human control could disappear."

Because of this, the Institute is calling for immediate action to be taken.

More from this section

AI tools can help Americans meet their civil legal needs

+4

New research reveals sick sea monsters swam to shore to vomit

Third-party data breaches rose 60% in a year. Most vendor reviews still happen once.

George Williamson, CEO of the Alan Turing Institute, said: "Questions about humanity's long-term future deserve attention, but discussions about advanced AI cannot be left solely to technology companies or framed only around the most extreme scenarios.

"We know enough about many serious AI risks to act, and with AI capabilities advancing rapidly, there is a clear need to move at pace.

"That means bringing together expertise from government, industry, academia and civil society to build the evidence, tools and partnerships needed to manage those risks.

"That's where the Alan Turing Institute wants to contribute, helping develop practical approaches that strengthen the UK's resilience, as our AI systems become more capable."

The group says that effective shutdown mechanisms would need to be supported by governance processes, safe fallback arrangements and clear operational procedures, while a meaningful slowdown in frontier AI development would require levels of regulatory, industrial and international cooperation that appear unlikely under current conditions.

Jason McEwen, interim chief scientist at the Alan Turing Institute, said: "The risks posed by advanced AI are real, but they are not beyond our control.

"They involve observable behaviors and evidenced harms, which means they can be studied, tested and acted on.

"The scientific work now is to ensure reasoning of powerful AI models remains observable as they become more capable, or to build the tools to read it where it is not, and to verify that complete AI systems, not just the models inside them, can be relied upon in real-world use."

The report says that there is now broad agreement that catastrophic AI risks are a realistic possibility in the short to medium term.