Affiliation: Co-founder & Executive Director, Center for Humane Technology (CHT); co-host, Your Undivided Attention podcast; former Google Design Ethicist; featured in The Social Dilemma (2020) One-line position: AI is the most powerful persuasion and manipulation technology ever built — the attention economy was the warm-up act, and without a fundamental shift in how AI systems are designed and governed, we face an unprecedented assault on human agency, truth, and democratic institutions.
What he's reacting against
- The attention economy that treats human attention as a resource to extract — this is his founding critique, now extended to AI
- AI discourse that focuses only on capability and safety, ignoring the persuasion and manipulation dimensions
- Tech companies designing AI to maximize engagement rather than human wellbeing — argues this is the structural flaw, not a bug
- The assumption that AI tools are neutral — insists the design choices embed values (usually: maximize engagement)
Key claims
- "The AI Dilemma" (2023): presentation arguing that AI amplifies every danger of social media (misinformation, polarization, addiction) by orders of magnitude
- AI is asymmetric: AI systems will understand individuals better than they understand themselves — creating a permanent persuasion advantage for whoever controls the AI
- First-contact problem: humanity is encountering an alien intelligence (AI) with no cultural, institutional, or legal frameworks for managing it — the speed of capability deployment outpaces society's ability to adapt
- Cognitive security as a civilizational priority: protecting humans from AI-driven manipulation is as important as physical security
- Design choices matter: the same AI can be designed to empower or to exploit — the current default (engagement maximization) is the wrong choice
- The attention economy → AI manipulation pipeline: social media trained the infrastructure and business model; AI completes the loop by making manipulation personalized, scalable, and harder to detect
Theories aligned with
- Attention economy / persuasive technology critique
- Cognitive security as a governance domain
- Adjacent to surveillance capitalism (Zuboff) but focused on persuasion and manipulation rather than extraction
Where he overlaps / splits
- Overlaps with Raskin (co-founder, CHT) on the attention economy; they're essentially the same institutional voice
- Overlaps with Haidt on technology harming wellbeing; splits on scope — Haidt focuses on children, Harris covers everyone
- Overlaps with Zuboff on tech as a power structure; splits on emphasis — Zuboff analyzes surveillance, Harris analyzes persuasion
- Overlaps with Harari on AI threatening democratic self-correction; splits on register — Harari is historical-philosophical, Harris is activist-practical
- Splits with Ng/Karpathy on AI's default trajectory — they see AI as mostly beneficial if taught/deployed well; Harris sees the default as harmful
Notable predictions
- (2020) Social media was fundamentally harmful to democracy and mental health — outcome: broadly supported by subsequent evidence (platform role in polarization, teen mental health data)
- (2023) AI would rapidly become the most powerful manipulation tool in history — outcome: TBD; deepfakes and synthetic content are growing but mass manipulation impact is not yet measured
- (Ongoing) Without intervention, AI will degrade epistemic environment faster than social media did — outcome: TBD; the central claim
Track record
- The Social Dilemma (2020) was the most-watched technology documentary — massive public awareness impact
- Center for Humane Technology has become the most visible advocacy org for tech design reform — institutional influence is real
- Congressional testimony on social media and AI harms — direct policy engagement
- The "attention economy" critique is now mainstream — genuine Overton window shift that Harris/Raskin catalyzed
- Specific policy wins are limited — awareness ≠ regulation
Empirical vs normative
- Empirical: documentation of persuasive design patterns, attention manipulation techniques, AI capability for personalized persuasion
- Normative: technology should be designed for human wellbeing; attention should be protected; AI governance must address manipulation, not just safety
- The normative drive is the engine — the empirical work supports a pre-existing commitment to humane technology
Sources
- "The AI Dilemma" (2023): widely viewed CHT presentation
- Podcast: Your Undivided Attention (with Aza Raskin)
- Film: The Social Dilemma (2020, Netflix)
- Center for Humane Technology: humanetech.com
- Congressional testimony: multiple appearances on social media and AI regulation
Weak spots / open questions
- The "persuasion technology" framing may be overstated — people are more resistant to manipulation than the framework implies, and the evidence for mass-scale behavioral change via AI is limited
- CHT has been more effective at raising awareness than at changing industry practices — what's the concrete policy win?
- The leap from "social media was bad" to "AI will be worse" is intuitive but empirically unproven — AI could also be designed to counter manipulation
- Activist framing tends toward worst-case scenarios — may systematically overstate risks and understate the difficulty of manipulation at scale
- Less technically engaged with AI capabilities — the critique is about design incentives, not about what AI systems actually do
Rich's take
- (your synthesis here)
converts-from: personas/tristan-harris.md · schema v1 · AI & Society domain