Stanford CS547 HCI Seminar | Winter 2026 | What's Up with AI?

Current debates echo decades-old predictions

Winograd notes that both utopian visions like Dario Amodei's 'Machines of Loving Grace' and doomer predictions such as Marvin Minsky's 1970 forecast of superintelligence within 8 years mirror historical patterns where society ignores long-term side effects until they become crises, much like early automobile enthusiasts dismissed concerns about pollution.

AI acts as a social accelerant

Rather than creating novel problems, contemporary AI amplifies pre-existing tensions around labor, resources, and truth, making the social context of technological deployment more critical than the technical capabilities themselves.

White-collar automation is immediate

Citing recent Stanford graduate employment data and Anthropic's Dario Amodei, Winograd warns that AI is rapidly eliminating entry-level software engineering positions, not just routine data entry, fundamentally altering career paths for current students.

Inequality over efficiency

Quoting Geoffrey Hinton, Winograd emphasizes that AI deployment primarily benefits wealthy owners through workforce replacement and profit concentration, while creating massive unemployment without addressing the social and psychological functions of work.

Political backlash is emerging

Data center construction has become a rallying cry for 2026 political candidates, signaling voter anger about AI's resource consumption and economic displacement that transcends technical debates about energy efficiency.

Fakes become frictionless

While manipulated media existed before AI—such as a 2004 doctored photo of John Kerry—modern tools make synthetic content so accessible that 'AI slop,' or low-quality algorithmic misinformation, threatens to collapse shared reality and trust in visual evidence.

The business model of unreality

Winograd cites industry insiders admitting that releasing deepfake tools constitutes the 'right business decision' despite societal harm, using competitive pressure ('if we don't, China will') to justify the erosion of epistemic trust.

Algorithmic authority replaces judgment

He shares a personal example where Stanford's automated systems falsely listed him in a maternal health institute due to grant association algorithms, illustrating how AI systems generate authoritative-sounding but nonsensical information without human oversight.