The debate about the existential risks that artificial intelligence poses is intensifying. In an interview with Fortune, Lila Ibrahim, chief AI readiness officer at Google DeepMind, weighed in on whether AI could cause human extinction.
Though declining to offer a specific probability, calling estimates “just guessing,” Ibrahim was clear on her stance. The risk is “not zero”.
Ibrahim was one of the leadership figures to sign the landmark 2023 Center for AI Safety statement alongside leadership from OpenAI and Anthropic. In reaffirming this position, she stressed that while the dangers of AI exist, we must address them with equal priority as the dangers of nuclear threat and pandemics.
“It would be irresponsible not to keep talking about existential risks, especially given how fast the technology has developed.” – Lila Ibrahim, chief AI readiness officer, Google DeepMind
How tech leaders are estimating AI existential risk
Ibrahim’s cautious approach joins other estimations from various scientists and tech titans:
Leading figures in tech and science are divided over the extent to which AI could prove dangerous but agree that the risk is worthy of serious attention.
‘Godfather of AI’ Geoffrey Hinton estimates that human extinction from AI exists is between 10-20% chance in the next 30 years.
Anthropic CEO Dario Amodei estimates that extreme negative impacts from advanced AIs exist at about 25% chance.
Google DeepMind chief AI readiness officer Lila Ibrahim doesn’t put a number on it, but admits “it’s not zero” and warns against being over-optimistic.
The late Stephen Hawking famously warned that AI could completely supplant humans.
While each figure comes with unique estimations, there’s a common thread: top minds in AI are pleading for careful consideration, enhanced safety measures, and global oversight until the advent of AGI (artificial general intelligence).
Historical Precedents and Over-Optimism
When drawing parallels to other technological shifts in the past such as the advent of the internet and social media, Ibrahim pointed out how the tech industry often fails to accurately gauge risks and account for them early on. She cautioned against expecting catastrophic incidents to be those society has already foretold. In a similar way the world failed to anticipate and prepare for the pandemic.
Furthermore, Ibrahim has expressed her concerns against overly optimistic tech predictions such as a post-scarcity economy, or settling on other planets.
Instead, Ibrahim argues that AI’s full capabilities can be harnessed for the benefit of humankind while tackling global challenges such as pollution and climate change, instead of being concentrated in the hands of a few mega-corporations.