 |
| Evan Hubinger warns there’s a >10% chance AI could wipe out humanity within ten years. Credit: Pinterest |
Others, including Bill Gates, warn the bigger danger may be losing our ability to think for ourselves.
The debate over artificial intelligence is no longer about whether it’s useful. It’s about whether it’s survivable.
On August 26, 2026, Bill Gates published a 6,000‑word memo titled The turbulent AI era is here.
He wrote: “The transition to the AI era will be one of the most turbulent times in human history.
Right now, we are not preparing adequately for that transition.”
Gates believes AI can leave humanity better off, but only if the right steps are taken now.
Not everyone shares his optimism. On September 9, Evan Hubinger, a safety researcher at Anthropic, replied to a public thread by former OpenAI and Anthropic researcher Jacob Coxon.
Coxon had suggested that “the people building AI earnestly believe that it could kill us all by the end of the decade.” Hubinger agreed, adding: “I personally think it is >10% within the next decade.
I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.”
Alignment is the technical term for ensuring an AI system’s actions match human intentions.
Misalignment means the system could “go rogue,” taking actions its creators never intended.
The risk grows if Artificial General Intelligence, or AGI, is achieved.
AGI refers to systems that can match or surpass human capabilities across all cognitive tasks.
A misaligned AGI could reprogram itself, accelerate beyond human control, and act against human interests.
Market forces complicate the picture.
The developer of the most powerful AI stands to become the most powerful company in the world.
That incentive pushes progress forward even when safety lags. Stephen Cobb, a former ESET researcher, argued that AI won’t destroy humanity unless we allow it.
Journalist Dan Tynan countered with a comparison to cigarettes: companies may knowingly push harmful products if profit outweighs safety.
OpenAI’s leadership has already slowed parts of model development.
Sam Altman confirmed that internal training was paused due to safety concerns, and that he has discussed pacing AI progress with U.S. officials.
That acknowledgment shows the risks are not theoretical — they are shaping decisions inside the labs building these systems.
But survival isn’t the only concern. Gates pointed to a survey showing that heavier AI use correlated with reduced critical thinking, especially among younger people.
Even aligned AI could erode the reasoning ability that defines human uniqueness.
The danger isn’t just machines acting against us. It’s us forgetting how to act without them.
The race to control AI is therefore about more than existential risk.
It’s about protecting the very qualities that make us human — judgment, independence, and the ability to think critically.
Whether the threat comes from misaligned superintelligence or from our own overreliance, the stakes are the same: what survives of us in the age of AI.
0Comments