The growing AI backlash

Gary Marcus · newsletter · 2026-05-04 source ↗ Reader 🖥️

AI Backlash AI Bubble Debate AI Regulation & Governance (External)AI Slop LLM Limitations

Generative AI has been a net negative for society outside a few domains like coding, the backlash was predictable and is now accelerating toward political salience, and the industry bears primary responsibility for privatizing gains while externalizing harms.

practical · polemic/opinion — A short, persuasive opinion piece marshalling anecdotal evidence (tweets, news links, personal predictions) into a thesis about public sentiment turning against generative AI. Primarily argumentative with light evidentiary support.

Structure

Part	Role
Opening: evidence montage of growing backlash	Establishes that anti-AI sentiment is real and widespread via social media examples, QuitGPT movement, Fortune coverage
Political prediction: backlash will shape 2028 elections	Escalates the stakes from cultural sentiment to political force, references a year-old prediction now 'on track'
Core theory: GenAI has been a net negative	Central argument enumerating harms (education undermined, surveillance, disinformation, deepfakes, bias, slop, data centers) against narrow benefits (coding, brainstorming)
Attribution of blame: industry irresponsibility	Identifies both technical causes (LLM unreliability) and political/economic causes (lack of regulation, privatized gains/socialized costs)
Coda: future AI might be great, but current GenAI has hurt more than helped	Hedges the polemic to preserve credibility as someone open to AI in principle

Key terms

net negative AI Backlash
The aggregate societal impact of generative AI, weighed across all domains, is harmful — the harms outweigh the benefits
slop AI Slop
Low-quality AI-generated content flooding the internet and degrading information ecosystems
QuitGPT AI Backlash
A grassroots movement of people abandoning ChatGPT and similar tools as a form of protest
over-leveraged AI Bubble Debate
Data center buildout that exceeds actual demand or justified return, creating economic and environmental risk

Propositions

explicitcontestedOutside of coding and a handful of other domains (e.g., brainstorming), generative AI has been a net negative for society.conf: medium
contradicts ←I'm done using AI · p2 — The backlash piece treats coding as one of the few domains where GenAI is net-positive/exempt from its 'net negative' verdict; Brett's own coding-specific experience was net negative for him (depression, apathy, worse software), directly disputing that exemption.
explicitcorroboratedGenAI is undermining education, enabling mass surveillance, increasing disinformation, deepfake porn, bias, cybercrime, and economic disparity, while drowning the world in slop and environmentally damaging data centers.
extends ←AI's environmental costs threaten water, land and climate · p8 — Supplies concrete distributional/environmental-justice detail (local burden vs. global benefit, e-waste falling on lower-income countries) for the backlash piece's general claim that GenAI is 'environmentally damaging' via data centers.
supports ←AI Surveillance and Social Progress · p4 — The-growing-ai-backlash#p2 lists 'enabling mass surveillance' among GenAI's externalized harms; this essay provides the elaborated, mechanism-level case for that specific harm.
explicitThe AI industry has privatized almost all gains while leaving all downsides to society, taking almost zero responsibility.conf: medium
explicitMost of these downsides were predictable and were in fact predicted — by Marcus in 2023 Senate testimony, by others in AI ethics, and even by OpenAI itself in 2019.conf: medium
explicitThe backlash against AI will be a major factor in the 2028 US Presidential elections and may factor into the midterms.
explicitSome future form of AI might be great, but current generative AI has hurt more than it has helped.conf: medium
implicitLLM unreliability is a root technical cause of GenAI's negative societal impact.
implicitResponsible regulation could have mitigated many of these harms.

Arguments

The backlash against AI is growing and justified

Implicit premises:

People eventually push back against technologies whose harms they experience directly
The listed harms are real and attributable primarily to generative AI rather than other causes

Nobody should be surprised by the backlash

Implicit premises:

Predicted harms that materialize validate the prediction and make the public response foreseeable

Narratives

Questions addressed

✅ Why is public sentiment turning against AI?
Because GenAI has been a net negative outside coding — harms are real, visible, and personally felt.
✅ Is the backlash justified?
Yes — harms outweigh benefits and were predictable.
✅ Could the downsides have been foreseen?
Yes — predicted by Marcus (2023), AI ethicists, and OpenAI itself (2019).
⭕ Will AI backlash become politically significant?
Predicted to be a major factor in 2028 elections, claimed to be 'on track'.

Gaps

evidentiaryNo quantitative evidence is offered for any of the enumerated harms. The 'net negative' claim rests entirely on an unweighted list of negatives versus two positives.
analyticalThe exemption of coding (and brainstorming) from the 'net negative' verdict is asserted without criteria for what makes a domain qualify as net-positive. No framework for how to weigh domain-specific benefits against cross-cutting harms.
closed byI'm done using AI — Brett supplies one concrete counter-case to the ungrounded coding exemption: for him, AI-assisted coding was net negative (mental health, deskilling, worse software), disputing that coding uniformly qualifies as a domain where GenAI's benefits outweigh its harms.
dialecticalNo engagement with counterarguments: productivity gains in knowledge work, medical applications, accessibility tools, scientific research acceleration. The piece treats these as non-existent rather than as insufficient to offset harms.
predictiveThe political prediction (2028 elections) lacks any mechanism story — how does diffuse frustration convert into voting behavior or policy change?
constructiveThe closing hope that 'some future form of AI might be great' is unconnected to any analysis of what would make future AI different from current GenAI.