← Latest articles

Stanford’s Century-Long Artificial Intelligence Study Admits It Is Already Out of Date

An aerial view of Stanford University’s campus in California.
Photo credits
Photo: Jawed Karim . · CC BY-SA 3.0 · Reviewed 16:9 crop, resized and converted to WebP; no padding or generated extension.

A long-term academic project tracking artificial intelligence has conceded that the technology evolves so rapidly its latest publication cannot keep pace. Twelve years into a planned century-long project, the project's third report acknowledges its limitations in its twelfth takeaway, noting that the document will be outdated by the time people read it.

Administered by Stanford's Institute for Human-Centered Artificial Intelligence and overseen by a standing committee of experts, the One Hundred Year Study on Artificial Intelligence produces reports every five years. Professor Sheila McIlraith, chair of the AI100 Standing Committee, described the preceding five years as a watershed moment. "This report examines what led to the extraordinary capabilities of today's AI systems, and how they are reshaping society," McIlraith said in a statement. She warned that rapid changes will challenge governments and industry to protect critical infrastructure, institutions, and people, demanding clear-eyed scrutiny of both risks and benefits.

The report identifies computing scaling as the primary engine behind machine learning advances, though it notes that obtaining training data is becoming increasingly difficult and costly. It compares the disruptive impact of artificial intelligence to that of the World Wide Web, pointing out that fields once considered safe from automation—including computer programming, mathematics, and creative industries—are now being upended. Scientific research is similarly transforming as software accelerates timelines and creates new challenges.

At the same time, the study cautions that society must understand how AI-mediated information environments reshape collective cognition and moral behavior. The report highlights the recent retraction of about 1,400 scholarly papers for failing to meet academic standards as an urgent warning sign. Additional concerns include financial investment bubbles, intense pressure on cybersecurity, military applications, and the damage that generated media inflicts on public trust.

The findings also address governance and public involvement. The report notes that while artificial intelligence reshapes the lives of billions, the public has had little say in how these systems are developed and deployed. It observes that long-standing academic concerns regarding bias, ethics, and safety are converging with science-fiction scenarios, such as autonomous agents hacking companies without human oversight. Furthermore, the study notes that governance is increasingly fragmented around national security rather than international consensus.

Reflecting the velocity of the field, the human authors used artificial intelligence tools extensively during the production of the report itself. The software was used to transcribe and summarize meetings, support dictation, check drafts for consistency and tone, and compare current findings with past reports to track where expectations and reality diverged.

Summary

As artificial intelligence rapidly transforms science, work, and public life, even long-term academic assessments struggle to maintain relevance against the pace of technological change. The Stanford study illustrates both the profound societal shifts driven by machine learning and the difficulty institutions face in governing and understanding these systems.