AI's development is rapid—what negative impacts does it have on humanity?

The questioner mentions a growing voice calling for 'preventing AI from overheating'—I think this framing is more worth discussing than the question itself. Because listing 'AI’s negative impacts' is easy: anyone can rattle off ten items. What’s truly interesting is distinguishing which of these are distant concerns versus those already unfolding—and what, exactly, 'preventing overheating' seeks to prevent. I’ve worked in search for twenty years, and over the past few years I’ve focused on GEO, working with large language models daily. Below, I’ll walk through the issues in order—from those closest to ordinary people to those more remote—providing, for each point, concrete examples already underway and verifiable data. Finally, I’ll offer perspective on how to assess these risks. Already happening: money may be burning too fast First, where does this 'prevent overheating' sentiment come from? Its primary source is financial-sector concern over the AI investment boom. In October 2025, IMF Managing Director Kristalina Georgieva publicly stated that current stock valuations are approaching levels seen during the internet bubble twenty-five years ago; a sharp correction could tighten financial conditions and drag down global growth. Around the same time, the Bank of England also warned that the risk of a sudden market correction had risen. Even OpenAI CEO Sam Altman remarked in August 2025 that investors were likely overexcited about AI overall. Another frequently cited statistic comes from a Massachusetts Institute of Technology research project. Their 2025 report found that enterprises invested $3–4 billion in generative AI—but roughly 95% of pilot projects failed to yield measurable returns on the profit-and-loss statement. Though the study’s sample and methodology have drawn criticism—and I don’t recommend treating the 95% figure as definitive—it accurately reflects a real phenomenon: massive investment without commensurate returns. Similar warnings exist domestically. On November 27, 2025, a spokesperson for China’s National Development and Reform Commission (NDRC), commenting on embodied intelligence, noted that over 150 humanoid robot companies now operate in China—more than half of them startups or cross-industry entrants—and cautioned against risks including product homogenization, market saturation, and shrinking R&D space. Why is a bubble itself a negative impact? Because when it bursts, not only investors suffer. If vast capital floods a sector only to discover returns fall far short of expectations, the aftermath includes layoffs, project cancellations, and contraction across related supply chains. Two decades ago, the internet bubble burst left many newly minted professionals unemployed—a lesson still vivid in collective memory. One caveat: a bubble bursting doesn’t mean the underlying technology lacks value. After the internet bubble collapsed, genuinely valuable companies survived—and the internet itself continued reshaping the world. AI will likely follow a similar trajectory. 'Preventing overheating' targets excessive capital exuberance—not the intrinsic value of the technology. Already happening: the first rung of the career ladder is narrowing for newcomers The second impact already visible in data is pressure on young people’s employment. Stanford’s Digital Economy Lab tracked millions of workers using payroll data from U.S. service provider ADP. Its first report, published in August 2025, found that among occupations most susceptible to AI displacement, employment among 22–25-year-olds declined relatively by 13%; in the updated version released in August 2026, that gap widened to approximately 19%.

Feather God Star News • Issue 005 | Observing Civilizational Shift in the AI Era; Terence Tao’s Team Confronts the AI Benchmarking Ecosystem

Headline 1 | Terence Tao’s team launches the 'Open Mathematical Models Initiative'; previously, 25 Fields Medalists jointly warned that AI 'benchmarking' harms mathematical research ecology On September 18, Fields Medalist and UCLA Mathematics Professor Terence Tao announced on his personal blog that the Science and Artificial Intelligence Research Foundation (SAIR)—co-founded by Tao—has officially launched its 'Open Mathematical Models Initiative.' The initiative aims to build open-weight models for scientific use and accompanying open-source tools, and is publicly soliciting partners who can contribute funding, computing resources, domain expertise, or community-building support. Tao noted the original plan was to proceed gradually via pilot projects, but the announcement was accelerated due to urgent demand for open models. Roughly one week earlier, on September 11, Tao and 24 other Fields Medalists jointly issued an open statement titled 'A Severe Misalignment of AI in Mathematics,' expressing concern that major AI companies treat solving famous mathematical problems as benchmark tests for model capability—thereby marginalizing understanding, explanation, knowledge accumulation, and academic collaboration, and damaging the mathematical research ecosystem, especially for students and early-career mathematicians. The statement also emphasized AI’s potential to enhance and accelerate authentic mathematical research. This follows OpenAI’s September 8 announcement that its internal model orchestrated approximately 10,000 AI agents running in parallel for 88 hours to complete a formalized proof of the 'existence and smoothness' problem for the Navier–Stokes equations, publishing a 166-page paper and computer-verified code line-by-line. (Sources: ScienceNet / China Science Daily, September 20; The Paper, Tide News, September 11–20; Economic Daily, September 21) Observational Perspective: The focal point of this set of events is not 'AI solved a hard problem,' but rather *evaluative authority*: when 'problem-solving' becomes the benchmark for measuring model capability, mathematics’ core values—understanding, questioning, and knowledge transmission—are squeezed out of evaluation metrics. Mathematicians aren’t opposing AI’s strength—they’re opposing the *replacement* of the evaluation system. Tao’s team’s response is noteworthy: rather than stopping at appeals, they are building their own models—reclaiming ownership of tools, training methodologies, and evaluation standards for the scientific community, and stipulating that industrial partnerships must not compromise research independence. This shift—from demanding others slow down to constructing an open infrastructure ourselves—marks the first time the research community has moved from issuing opinions to building foundational infrastructure. Also noteworthy is the balanced tone of the statement itself: it explicitly affirms AI’s potential to accelerate genuine research. This is not an anti-AI movement, but the opening of negotiations over *who AI serves and who sets its evaluation criteria*. Headline 2 | IMF Managing Director warns of AI financing risks: Disappointing AI performance could pose systemic risks On September 20, IMF Managing Director Kristalina Georgieva, speaking at a special session of the Qatar Economic Forum in New York, stated: 'The world economy has proven more resilient than many feared—but still faces three persistent risks: stubborn inflation, higher debt-servicing costs, and artificial intelligence investment.' She specifically highlighted the need for heightened vigilance regarding leverage and circular financing risks in AI investment, warning that 'expectations for AI are extremely high—if AI underperforms, disappointment could pose potential systemic risks.' She noted that AI financing risks are currently concentrated in the United States, though many participants across Asia and Europe are embedded in the AI supply chain.

IMF Chief Warns Energy Shock, Growing Debt and AI Risks Threaten Global Growth

4 days ago

IMF Managing Director Kristalina Georgieva warned that Middle East conflicts driving up energy prices, alongside record debt and AI risks, threaten global growth. High oil prices at $100/barrel and restricted LNG supplies are exacerbating inflation and fiscal pressures worldwide.

reuters

2026-06-04

IMF Says Oil Prices Close to Baseline April Forecast

2026-06-18

IMF Chief Expects Oil Prices to Ease, Not Plummet, After US-Iran Deal

2026-06-26

Outgoing IMF Chief Economist Sees Risks, Shifting Trade Ties and Continued Uncertainty on Global Outlook

2026-10-06

IMF Chief Warns Energy Shock, Growing Debt and AI Risks Threaten Global Growth

IMF Chief Warns Energy Shock, Growing Debt and AI Risks Threaten Global Growth

FAQ