I don't want anything your AI generates
Generative AI is polarizing people who work in tech, with some arguing they want nothing produced by large models because it’s derivative, trained on unconsenting labor, environmentally costly, and is already flooding the internet and workplaces with low‑quality “noise.” Others counter that humans are equally derivative, that AI is a practical tool for tedious communication, coding and recommendations, and that it can democratize creativity and productivity for people and small businesses who lack resources. Underneath are deeper disputes about consent for training data, the real versus hyped benefits to end users, whether AI will deskill or augment work, and how much society should try to slow or redirect this wave of automation.
Perceived flood of AI-generated content
- Many commenters say even if they personally avoid AI tools, they can’t avoid AI output: art sites, hobby groups, search results, social media, SEO spam, and codebases are increasingly saturated.
- This is framed as a “signal-to-noise” collapse and an “infinite noise generator,” making online spaces less useful unless heavily moderated.
Quality, usefulness, and “mediocrity”
- Strong view that LLM output is “regression to the mean”: generic, verbose, lowest‑common‑denominator, especially for writing and code.
- Others argue that “mediocre average” is often exactly what’s needed: stock policies, tourist guides, recipes, boilerplate emails.
- Some say AI is already better than many humans for mundane tasks (summaries, classification, routing tickets, code scaffolding).
Creativity, originality, and derivative nature
- One camp: every great work is an outlier; models trained on the corpus of human output inherently push toward bland, safe ideas and cannot sustain genuinely contrarian or unique positions.
- Counter‑camp: all human work is also derivative; models “learn” from examples much like humans do, and can be powerful tools for exploring idea space or augmenting artists.
Labor, jobs, and economic effects
- Fears that front‑end devs, call centers, data entry, illustrators, and some middle management will be displaced; critics highlight unequal gains and lack of safety nets.
- Supporters emphasize new capabilities for small teams, indie devs, and non‑experts (e.g., “two‑person brands” competing with big ones, indie games, solo creators).
Search, recommendations, and discovery
- Disagreement on whether AI improves or worsens search: some see LLMs as a poor replacement and a driver of SEO spam; others find AI‑augmented search and tools like music recommenders significantly better than current search or radio.
Communication, social interaction, and workplace use
- Many dislike AI‑mediated social interaction (chatbots, AI therapy, AI in every text field) and foresee loops where AI generates verbose corporate prose that other AIs then summarize.
- Others welcome offloading “unimportant” but socially required communication (polite emails, reports), seeing this as damage control.
Data, consent, and copyright
- Contentious debate over training on unlicensed public data: some say public posting implies acceptance of reuse and learning; others insist this is exploitation and want models limited to properly licensed corpora.
- Related concern: low‑paid workers labeling disturbing data for safety and filters.
Environmental impact
- Several posts argue large models have significant carbon and resource costs, analogizing to Bitcoin and warning about projected data‑center energy use.
- Others counter that inference is relatively cheap, data centers increasingly use renewables, and automation may lower overall emissions compared to human labor.
Cultural / philosophical reactions
- Some see criticisms as “Luddite” or emotionally driven tech rejection; others argue strong pushback is necessary to influence how AI is deployed.
- A recurring theme: distinction between AI as invisible “glue” in systems vs. AI as a direct producer of human‑facing content.