Bill Gates has warned that humanity faces two decades of AI disruption, a period he believes could fracture society between those comfortable near the technological frontier and those who reject it.
The former Microsoft CEO and Gates Foundation chairman has been speaking publicly about artificial intelligence for several months. This follows an essay he published in late August titled “The turbulent AI era is here. The choices we make now are critical.” In that piece, Gates wrote that the transition to this new era would be one of the most turbulent times in human history.
Recent interviews reinforce his stance. He has described AI as “the most dangerous thing that humans have ever gone near,” adding that it “makes nuclear weapons look like nothing.” His view is that neither the current policy response from the White House nor coordinated efforts from the AI industry are sufficient to combat the threat.
Interview highlights
The conversation focused on whether Gates is an AI doomer, how to control the technology, and the difficulty of defining it before we can align it with human interests.
Gates stated that the response to AI has broadly been inadequate to the speed at which the challenge is arising. He expressed concern that bad outcomes will not be minimised and that beneficial applications could progress slowly if management fails.
He noted that progress over the last year has been remarkable. This includes the ability to write code and find cyber flaws, reason in biology, form relationships with children and other people, and disrupt the job market. He believes society needs to be ready for every one of these developments.
The nature of the risk
Gates clarified that the risk is not one model versus another. He said the near-term danger comes from models where safeguards can be removed. In contrast, models where activity is monitored to prevent cyberattackers and bioterrorists from exploiting the technology pose less immediate risk.
On a longer timeframe, the control issue relates to any state-of-the-art model, but it is always easier when safeguards are maintained.
When asked if he trusts specific leaders like Satya Nadella at Microsoft more than others, Gates replied that for someone conducting a cyberattack, the choice of model does not matter.
He explained that models with safeguards in place, where effort is made to monitor usage, do not present gigantic near-term risks. The danger lies where someone can remove all safeguards, making cyberattacks available today.
Crossing the threshold
Gates cited the basic power of models towards social relationships, the job market, cyberattacks, and bioterrorism attacks as the reason for his urgency. He stated that we have crossed the threshold of capability where this is by far the most dangerous technology ever invented.
He added that it will have pervasive effects in the economy and how war is fought.
He argued that while the industry speaks up about these issues, awareness and willingness to believe them are very low outside the sector. He noted that people often look at the past and dismiss fears about technology disruption, including impacts on the job market, as untrue.
He rejected the suggestion that industry insiders are simply seeking better public relations. He insisted that the warnings are the truth and are universal. He pointed to the founding story of OpenAI, which was designed with the caveat that if it became too powerful, it might open Pandora’s box, as evidence.
He also referenced DeepMind’s negotiation with Google management.
The people close to the technology are very concerned about misuse, job impacts, and social relationship impacts. Gates found it impressive to have an industry speaking out as vividly as it does.
What it means
For people making things, the implication is a long-term adjustment period. Gates suggests we should prepare for roughly twenty years of change. This era could be so painful that it splits society into two groups: those comfortable living near the technological frontier, and those who reject it.
He also highlighted labour considerations and the nihilism of younger generations toward technological progress. While he did not offer a definitive solution to alignment for open-source models in this interview, he emphasised that technical solutions must be enforced, not just applied to closed-source systems.
