The future of everything is lies, I guess: Where do we go from here?
An essay series warning that large language models will flood society with “lies” has prompted intense debate over how disruptive AI really is and what, if anything, individuals should do about it. Commenters argue over parallels to the automobile—useful but societally distorting—raising concerns about deskilling, job loss, information pollution, coercive workplace adoption and the inability to truly “opt out,” while others counter that fears are overstated, technological progress is inevitable, and the focus should be on wise use, regulation, and shaping net-positive outcomes. Underneath is a broader anxiety about inequality, political economy, and whether current AI trajectories entrench a new kind of digital feudalism or eventually deliver genuine benefits like medical advances and reduced drudgery.
Use of AI at Work and Coercive Incentives
- Many feel economically forced to use LLMs despite misgivings; refusal may mean job loss or stalled careers.
- Some add “AI”/“agentic workflows” to résumés even though they personally cringe, reasoning that HR and management now expect it.
- Others view advertising AI skills as a red flag and prefer to avoid AI‑centric workplaces, even if that means lower‑status or manual jobs.
- Several describe “AI theater”: leadership mandates AI use to “accelerate” feature delivery, leading to massive, unreviewable PRs and long‑term quality worries.
Ethics, Principles, and System Constraints
- Strong divide between “stick to principles even if it hurts you” vs “individual ethics can’t beat market incentives.”
- Some argue personal boycotts are futile without structural change; others insist individual refusal still matters morally.
- There’s frustration with what some call naive systems thinking vs naive moralizing; disagreement about how much individuals can influence large‑scale trajectories.
Deskilling, Metis, and Learning
- Many resonate with the article’s concern that LLMs erode persistence, “muscle memory,” and deep understanding.
- Comparisons to writing, calculators, and Socrates’ critique of writing: new tools always deskill something, but outcomes differ by domain.
- Particular worry about students who rely on AI for coursework, then “crash” on exams; professors report unusually high failure rates.
- Some see a future premium on “pre‑AI” engineers who learned through long, manual struggle and can now use AI more judiciously.
Car Analogy and Technology Externalities
- Long, detailed debate over whether cars were a net positive and how that maps to AI.
- One side: cars (and by analogy, AI) brought vast benefits in logistics, mobility, and prosperity; you must accept some externalities.
- Other side: car‑centric planning produced sprawl, pollution, isolation, and dangerous streets; benefits could have been achieved with fewer harms via different policy.
- This is used to argue both “we should shape AI with regulation now” and “we can’t unrealistically ban or opt out of dominant tech.”
Concrete AI Use Patterns
- Common coding pattern: LLMs for boilerplate, scaffolding, refactors; humans for design, tricky logic, and cleanup.
- Some teams skip reviews for “vibe coders” and instead have stronger engineers refactor their AI‑assisted PRs directly.
- Others deliberately feed “slop” into AI‑driven review cultures, seeing it as giving management what they asked for.
Futures, Risk, and Regulation
- Views range from “LLMs are just another automation tool that will create new jobs” to “they are aligned with elite interests and could entrench feudal‑like power.”
- Suggested responses include: unions, resisting AI mandates, aggressive regulation and liability, opposing datacenter subsidies, and even coordinated strikes or bank runs (controversial and disputed as effective).
- Disagreement on how much doom is warranted; some call the tone excessive doomerism, others see it as proportionate to the stakes.
Information Ecology and Legal Context
- Concern that pervasive AI slop will make “source or GTFO” essential, yet sources themselves may be polluted.
- UK readers note the blog is geo‑blocked due to Online Safety Act concerns; this is cited as illustrating broader information‑control risks independent of AI.