Concerns about the potential risks of artificial intelligence have reached a critical point, according to Bill Gates. He said governments and technology companies have crossed several previously identified safety thresholds and must now move from voluntary reviews to meaningful action. Gates also called for greater cooperation between the United States and China, emphasizing that neither country needs to ban AI outright to develop shared safety standards.
“You have to say what you want,” Gates said. He acknowledged that the technology industry is working to improve its public image, particularly as companies seek to raise trillions of dollars in investment. However, he argued that public relations should not replace honest discussions about AI risks, regulation, and accountability.
Gates believes the United States and China can find mutually beneficial ways to cooperate beyond artificial intelligence. Still, he considers AI safety the most urgent area for collaboration. He said countries should clearly explain their domestic plans for regulating advanced AI systems, including models capable of designing molecules. Such systems, he argued, should be monitored to reduce potential risks involving biotechnology and public health.
Although Gates acknowledged that monitoring AI models would not be perfect, he said it is unreasonable that governments are discussing numerous secondary issues while failing to address the most serious threats. He also reflected on the unusual position he occupies as one of the few public figures speaking so directly about the dangers of advanced technology.
MIT Technology Review: You have been viewed for much of your career as an effective communicator who attracts significant public attention. That may be one reason you are speaking about these issues now. You have also expressed regret about your past relationship with Jeffrey Epstein, while other controversies—including conspiracy theories surrounding COVID-19 vaccines—were largely outside your control. Do you still believe you can be an effective messenger, and how do you view your public legacy?
Bill Gates: I am not particularly focused on legacy. However, I recognize that I have made mistakes and could have handled certain situations better. I was criticized during the antitrust trial, and I believe I could have responded more effectively. My first marriage was also unsuccessful, and I made significant mistakes in that relationship.
Spending time with Epstein was an extremely foolish decision. It placed the reputation of the Gates Foundation at risk, and that reputation is essential to the organization’s ability to carry out its work. I should have addressed the issue more directly, appeared before Congress, answered questions, and acknowledged that my judgment was wrong.
I am an imperfect messenger. I became wealthy through technology and innovation, which may make my criticism of emerging technologies seem unusual. However, innovation without effective policies and safeguards could ultimately have a negative impact on humanity. That risk is not receiving enough attention in the broader public debate.
Since 2008, I have used my access to political leaders and government officials to focus primarily on global health and international development. My priorities have included eradicating malaria, expanding access to vaccines, increasing foreign aid, supporting medical research, and reducing childhood mortality. When I meet leaders such as Donald Trump, Xi Jinping, or Emmanuel Macron, I generally want to discuss saving lives and helping the world’s poorest communities.
AI safety is now an important part of that work. It may take time away from conversations about global health, foreign aid, and disease prevention, but I intend to use my platform to discuss how societies can reduce the risks associated with advanced AI. That includes speaking with journalists, meeting political leaders, and encouraging greater public awareness.
Gates said several important AI capability thresholds have already been crossed. Late last year, he was particularly concerned by advances in coding, context windows, autonomous agents, and underlying AI models. What initially appeared to be a major coding milestone soon became evidence of a broader capability: AI systems that could potentially support sophisticated cyberattacks.
Despite those developments, Gates said the response from governments and the technology sector has been limited. He argued that the public needs to understand how quickly AI capabilities are advancing and why risks involving cybersecurity, biotechnology, and psychological manipulation deserve urgent attention.
“Yes, I am an imperfect messenger,” Gates said. “Find your perfect messenger, and I will tell that person everything I think. I am being somewhat cynical because I do not know whether a perfect messenger exists. But people need to understand the technology, the trends shaping it, and the potential consequences for cybersecurity, biology, and society.”
Source: www.technologyreview.com