Delaying Artificial Intelligence The Real Existential Risk Says Joscha Bach, Scientist Says He's More Afraid That 'We Don't Build AI'

German cognitive scientist Joscha Bach says humanity faces a bigger existential threat from dragging its feet on artificial intelligence than from building it too fast.

What Happened: Speaking last week at the California Institute for Machine Consciousness, the longtime AI theorist argued that climate change, super-volcanoes and our own "inability to stay aligned with survival" pose a higher kill-shot probability than a runaway algorithm — unless people use the current "technological bubble" to build smarter machines fast.

"Over a long enough time span, it’s certain something will lead to our extinction … I'm much more afraid that we don't build AI than that we build it," he told the audience.

Bach, who researches machine consciousness and computational models of mind, has previously written that society must "make AI beautiful" — safe, aligned systems capable of steering civilisation through the next evolutionary bottleneck.

Without AI, Bach argues, civilization's respite from predators and famine is a temporary "bubble" that will burst. Scaling intelligent machines, he says, is the only credible way to invent planetary-scale defenses.

Why It Matters: The plea drops into a rift among AI luminaries. OpenAI chief Sam Altman has shifted from calling for sweeping guardrails in 2023 to urging "sensible regulation” that “does not slow us down," framing speed as vital to out-innovate China. Elon Musk and Apple AAPL co-founder Steve Wozniak joined a 2024 open letter demanding a six-month moratorium on systems beyond GPT-4, though Musk later called the effort "futile."

Geoffrey Hinton, the "Godfather of AI," quit Google GOOGL GOOG in 2023 after concluding there is a 50-50 chance advanced systems could wipe out humanity within two decades. At Davos, fellow Turing Award winner Yoshua Bengio told Business Insider that autonomous "agentic" models could be the most dangerous route and pushed for international safety rules before such systems scale.

Meta's META Yann LeCun, in an interview with WIRED, dismisses such doomsday talk as "stupid," insisting today's models still lack basic reasoning and common sense.

Photo Courtesy: Tapati Rinchumrus on Shutterstock.com