AI100 review warns AI’s mainstream rise has left the public with little say

AI100 released its third five-year report on November 10, 2026, describing a technology whose reach has grown faster than public influence over its direction. The assessment credits larger datasets and more computing power for generative AI’s advances, while warning that resource constraints, concentrated development and fragmented governance are shaping what comes next.

Titled AI Goes Mainstream: The Scaling Era and the Rise of Generative AI, the report comes from the One Hundred Year Study on Artificial Intelligence, administered by Stanford’s Institute for Human-Centered Artificial Intelligence. Its international panel spans computer science, philosophy, sociology and related fields. It answers 12 standing questions about AI’s advances, opportunities, dangers and effects on society.

From bigger training runs to more work on answers

The panel identifies scaling as the main engine of generative AI’s progress over the past five years. The basic approach is to build systems with ever more data and computing power. That combination, rather than a single application or breakthrough, sits at the center of its account of the period.

But this approach faces constraints in both data and computing resources. The panel says those limits are shifting the frontier toward reasoning and inference-time compute—the computing work involved in producing answers. Its assessment describes a change in where further progress is being pursued, not an end to AI development.

The consequences already extend into creative industries, software development and mathematics, fields the report describes as once thought beyond AI’s reach. In science, it says AI is compressing research timelines from years to months or weeks. That acceleration also brings new scientific challenges requiring attention, rather than simply removing obstacles to discovery.

Infrastructure costs and concentrated power

The physical demands of that progress feature prominently. The panel describes AI infrastructure investment as unprecedented in scale and speed, supporting parts of the U.S. economy and dominating capital markets. It also links the boom to fears of a financial bubble and pressure on energy and freshwater supplies.

Development of the most advanced AI is concentrated among a small number of U.S. and Chinese actors, the report says. That concentration raises questions about who will own and gain access to leading systems. Meanwhile, governance remains fragmented, increasingly framed around national security, with little international agreement on its direction.

The panel also identifies concrete areas where AI’s expanding capabilities are changing the stakes:

• Cybersecurity has become a major area of potential AI abuse and a possible route to real-world harm.
• Military AI has moved from hypothetical discussion to a transformative force in active conflicts, according to the panel.
• Generated images, audio and video challenge the assumption that recordings are inherently trustworthy evidence.

AI is reshaping the lives of billions of people, yet the public has had little say in how it is developed and deployed.

2026 AI100 Study Panel report

The limits of a five-year snapshot

AI100’s assessments began in 2016, followed by the 2021 edition. Panel member Iyad Rahwan says that earlier report was optimistic about rapid progress but did not foresee the “ChatGPT moment” that made AI a mainstream topic. The new report acknowledges the difficulty of keeping pace: “This report will be out of date by the time you read it.”

Its production also reflects the shift it describes. According to the disclosure quoted by The Register, humans wrote the report while using AI extensively for meeting transcription and summaries, dictation and draft checks. AI also helped compare this edition with earlier reports to identify where expectations and reality had converged or diverged, and checked the completed text for inadvertent plagiarism and signs of AI-generated content.