I Am Tired of AI
Many technologists say they’re exhausted by the relentless push to bolt generative AI onto everything, arguing it produces mediocre “slop,” erodes trust in writing and art, and accelerates spam, surveillance and job precarity more than it solves real problems. Others counter that LLMs and related tools already deliver major productivity gains in coding, research and accessibility, and liken current backlash to past resistance to calculators or the internet. Beneath the surface runs a deeper conflict over copyright, data scraping and who captures the value of human-created content, with some calling for lawsuits and regulation while others argue the real problem is the capitalist incentive to prioritize scale and profit over quality and human agency.
AI and Jobs / Adoption Pressure
- Some argue ignoring AI risks unemployment; others with established careers say they can safely avoid it and see “AI or jobless” as fearmongering.
- Many expect white‑collar roles (coding, writing, support) to be heavily automated; others note that in operations/IT they haven’t yet seen jobs lost for not using AI.
- A recurring view: people will use AI even badly, and everyone else will bear the consequences.
Quality and Detectability of AI Output
- Many say AI text has a bland, median “TOEFL essay” tone, overly polite and generic, often obvious on sight.
- Others point out studies showing humans are bad at reliably detecting AI text; “you only notice the bad ones.”
- Some see current AI art/text/music as mediocre, but note that cheap, “good-enough” output can still transform markets.
Copyright, “Theft”, and Training Data
- Large sub‑thread on whether mass scraping for training is “the biggest theft in history” or just copying information.
- Disputes over law: some argue copyright only cares about outputs, not training; others think model weights themselves are derivative works if memorized text can be recovered.
- Many resent that corporations get away with training on others’ work while aggressively enforcing their own IP.
- Split between “abolish or weaken copyright for everyone” vs “tool it (GPL‑style) to force open models or shared benefits.”
Centralization, Capitalism, and Power
- Widespread concern that AI will deepen corporate concentration (OpenAI, Google, Meta) and build moats via exclusive data deals.
- Others counter that open‑weight models (e.g., Llama family) push in a more decentralized direction and may compress profits.
Trust, Information Overload, and “Slop”
- Many feel they can no longer trust new writing, since it may be partially or wholly AI‑generated; this erodes the sense of human connection and “proof of work.”
- Worries that AI is a force multiplier for spam, SEO sludge, propaganda and synthetic reviews, making it much harder and costlier to find reliable information.
- Some argue everything was already full of low‑quality content; AI just changes scale, not nature.
Usefulness vs Limitations of Current Tools
- Enthusiasts report real productivity gains: coding assistants (Cursor, Claude, GPT‑4/o1) for boilerplate, refactors, tests; summarization; translation; quick scripts.
- Common pattern: treat models as a “junior dev” or “bad intern” whose work must be reviewed line‑by‑line.
- Others find tools like Copilot unreliable or net‑negative and feel gaslit by hype.
- Strong consensus that AI is stochastic and must not be used as a trusted oracle or sole decision‑maker.
Ethics, Creativity, and Human Work
- Creators fear devaluation of human writing, art, and tests; some pledge never to use AI and market “100% human” work as a differentiator.
- Others see AI as a powerful editor, idea generator, or on‑ramp, with humans still responsible for taste, intent, and final judgment.
- Thread repeatedly contrasts excitement about technical progress with fatigue over relentless hype, “AI‑washing” of products, and its murky social costs.