Yuval Noah Harari predicted financial disaster due to artificial intelligence

Artificial intelligence could cause a financial crisis with “catastrophic” consequences, according to historian and author Yuval Noah Harari, who says the complexity of the technology makes it difficult to predict its dangers.

Harari told The Guardian that concerns about the safety of testing artificial intelligence models stem from anticipating all the problems a powerful system could cause. Unlike nuclear weapons, he said, there was no “big, dangerous scenario” that everyone understood.

“With artificial intelligence, what you’re talking about is a very large number of dangerous scenarios, each of which has a relatively small probability that taken together … poses an existential threat to the survival of human civilization,” Harari notes.

The Sapiens author, who has become a prominent voice on concerns about AI development, said the multilateral declaration adopted at the recent global summit on AI safety at Bletchley Park was a “very important step forward” because leading governments came together to express concerns about technology and do something about it.

“Perhaps the most encouraging thing about this is that they were able to get not only the European Union, the United Kingdom, the United States, but also the Chinese government to sign the declaration,” he said. “I think it was a very positive sign.” Without global cooperation, it will be extremely difficult, if not impossible, to curb the most dangerous potential of artificial intelligence.”

The summit ended with an agreement between 10 governments, including the UK and US, but not China, the EU and major AI companies, including ChatGPT developer OpenAI and Google, to collaborate on testing cutting-edge AI models before and after they are released.

Harari said one of the challenges in testing models for safety was anticipating all the problems a powerful system could cause. “Artificial intelligence is different from all previous technologies in human history because it is the first technology that can make decisions on its own, that can create new ideas on its own, and that can learn and develop on its own. Almost by definition, it is extremely difficult for humans, even those who created the technology, to anticipate all the potential dangers and problems.”

Governments have cited the threat of artificial intelligence systems helping create biological weapons when trying to explain their danger to the public, Harari said, but there were other scenarios that could be considered. The author pointed to finance as a sector ideal for artificial intelligence systems—“it’s an ideal area for AI because it’s just data”—and a potential source of a major AI-driven crisis.

“What would happen if AI not only gained more control over the world’s financial system, but also started creating new financial devices that only AI could understand and that no human could understand?” Harari said, adding that the 2007-2008 financial crisis was caused by debt instruments such as collateralized debt obligations (CDOs), which were poorly understood and thus inadequately regulated.

“Artificial intelligence has the potential to create financial devices that are orders of magnitude more complex than CDOs. And just imagine a situation where we have a financial system that no one can understand and therefore also cannot regulate,” he said. “And then the financial crisis comes and no one understands what’s happening.”

Last month, the UK government raised concerns about an advanced artificial intelligence model potentially posing an existential threat by controlling and manipulating financial systems, The Guardian notes. But Harari said a financial crisis created by artificial intelligence would not destroy human civilization – “at least not directly.” He added: “This could happen indirectly if it provokes certain types of wars or conflicts. This is a catastrophic risk – economic, social, political – but in itself I would not call it existential.”

An Israeli author who has supported calls for a six-month pause in the development of advanced AI and supports holding artificial intelligence companies accountable for damages caused by their products said the focus should not be on specific regulations and laws, but on regulatory institutions that know technology that can quickly respond to new breakthroughs.

“We need to quickly build strong regulatory institutions that can identify and respond to dangers as they arise, recognizing that we cannot predict all dangers and problems in advance and legislate against them in advance.” Harari added: “This should be the main effort, not trying now to write some very long and complex regulation that may be out of date by the time it is passed by parliament or congress.”

According to Harari, as part of this program, AI security institutes should hire experts who understand the potential impact of artificial intelligence on the world of finance.

Rishi Sunak announced the creation of the UK’s AI Safety Institute last month, and days later the White House announced plans to create a similar body, with both expected to play a key role in testing advanced AI models. Speaking at the summit, Sunak said the UK needed to first understand the capabilities of advanced models before introducing legislation to combat them.