Bill Gates has undergone a notable shift in perspective. Prior to 2026, the Microsoft co-founder was a staunch advocate for artificial intelligence development, embodying a techno-optimistic outlook. While he acknowledged potential risks related to employment and misinformation, he largely viewed them as manageable, insisting that no previous technological shift compared to the transformative power of AI. He firmly believed AI would tackle long-standing global challenges like inequitable healthcare distribution and expanding educational access.
However, by the latter half of 2026, Gates now believes AI could constitute a serious threat to human society, asserting that addressing these risks should be the world's foremost priority. He notes that the tech industry itself is aware of these dangers, yet deliberately downplays them, fearing that excessive risk warnings could jeopardize funding opportunities.
In a recent interview, Gates stated that AI has the potential to be both a powerful force for equality and a catalyst that exacerbates existing global disparities. He explained, "If AI is only applied in wealthy nations and fails to reach developing countries, inequality will only deepen. For example, in Sub-Saharan Africa, areas with the most acute doctor shortages are the least able to leverage AI for health solutions. I believe there is no longer any reason to delay changing this."
Adding to his concerns, Gates published a 6,000-word essay on his personal website on August 26, suggesting that the transition to the AI era would be one of the most turbulent periods in human history. He argues that the international community is currently unprepared for this shift, although with the right steps, AI could become a positive force that improves life for everyone. While he does not dismiss AI's development, his focus is increasingly on ensuring its benefits are widely distributed.
Where to begin tackling this issue
The urgency of the situation was underscored on September 8 when Jacob Coxon, a researcher at US AI company Anthropic, announced his resignation on social media platform X. In his resignation statement, he warned that leading AI companies are in a competitive race to develop superintelligent AI technologies that "could destroy us all within the next decade," a remark that immediately caused a stir. Coxon's concerns are increasingly shared by others in the field.
Gates noted that the Gates Foundation maintains strong collaborative relationships with leading US AI firms like Microsoft, Google, OpenAI, and Anthropic. He mentioned his deep ties with Microsoft, where he worked for many years and still offers advice. He also revealed plans to visit OpenAI and Anthropic in San Francisco soon for consultations. In January of this year, OpenAI and the Gates Foundation jointly launched the "Horizon 1000" initiative, committing a combined $50 million in funding and technical support to integrate AI with healthcare and help African nations improve their health systems. Rwanda serves as the initiative's first pilot country, with Gates expressing excitement at their rapid progress and potential to become a model for Africa.
On September 15, the Gates Foundation announced it would invest at least $1 billion over the next two years to promote broader and more equitable access to AI globally. That same day, the 2026 Goalkeepers Report, titled "Creating Change - AI, Equity, and an Urgent Choice," was released. The report emphasizes that whether AI benefits those with the most resources or reaches those with the least is not yet determined, making current decisions critically important.
Gates explained to the media that this window of opportunity may only last 12-18 months. "We should not waste any time. In healthcare, whether AI directly assists patients, helps medical staff improve their service capabilities, or enhances their decision-making and follow-up care, it can save countless lives." To ensure more people benefit from AI, Gates argues for addressing the "language" component of large language models. Early models were trained predominantly on English-language data, making AI increasingly accurate and powerful for English and other major language speakers, while stagnating for others. For instance, a Yoruba speaker in Nigeria might find current leading AI speech recognition systems severely lacking, with error rates as high as 60%, compared to under 6% for English. Such errors can even be life-threatening, as exemplified by a scenario in Malawi where a patient's local Chichewa phrase for "my water broke" could be mistranslated by AI tools as "pour the water away," leading to dangerous medical miscommunication.
Part of the Foundation's $1 billion commitment will fund digital infrastructure for equitable AI, including building language datasets not adequately covered by current tools and evaluating AI based on the daily languages of its users. Mark Suzman, CEO of the Gates Foundation, acknowledged that while $1 billion might sound modest compared to the astronomical figures associated with AI investments, it is a substantial contribution in the field of AI accessibility, with expectations that this funding will grow.
Gates' perspective on China's AI sector
Gates shows keen interest in China's AI industry and its latest developments. He observes that China places a higher emphasis on humanoid robots compared to other countries, mentioning he recently enjoyed watching the World Humanoid Robot Games over a weekend. He found the event fascinating, noting both impressive successes and spectacular failures. He believes the mere organization of such a "Robot Olympics" is mind-boggling and predicts the next edition will be even more compelling, given the rapid pace of development in the field.
Discussing humanoid robots, Gates identifies robot hand technology as a persistent bottleneck. He notes that many researchers in China are focused on this area, mentioning he has visited a company in Shanghai working on robotic hand technology and closely follows the field. He predicts that if humanoid robots achieve human-level capability in construction and manufacturing within a few years, they could work around the clock, boosting productivity. However, he cautiously adds that this is good "in a sense," as he is particularly concerned about the resulting employment challenges. "Society as a whole needs to think about how we adjust and ensure that people currently in these jobs feel they are properly taken care of and supported, and that this support is not just financial. This is a problem the world hasn't solved yet."
In his 6,000-word essay, Gates mentioned China three times, the most of any country besides the US. He proposed that "a degree of cooperation between the US and China" is necessary to address AI challenges. He acknowledges that "China's AI models are performing exceptionally well," noting that both China and the US each have four or more institutions with models maintaining state-of-the-art status. The performance gap between the top model and the next seven leading models is not significant, and current models have already surpassed the best from just four months ago. Gates characterizes this as a "hyper-competitive" environment, which means users, including African governments, have access to many high-performing, low-cost options. He further suggests, "I hope both countries could allocate about 5% of their data center computing power to support poorer nations, similar to foreign aid, for humanitarian areas like health, agriculture, and education."
Comments