If you are asking for human attention, demonstrate human effort
As large language models flood workplaces with effortlessly generated emails, specs and code, many engineers are pushing back against what they see as “AI slop” that wastes more human review time than it saves. Commenters argue that if you’re asking for someone’s attention or accountability — especially in code reviews or technical decisions — you should first invest real effort yourself: scoping work, self-reviewing AI output, and taking responsibility for errors. Others note that AI can be valuable when used for summarization, editing or small, well-bounded tasks, but consensus forms around a norm: don’t offload the hard thinking to a model and then make colleagues pay the cognitive cost.
Perceived problem: “AI slop” and asymmetry of effort
- Many describe being flooded with long, verbose, obviously‑LLM text (emails, specs, PRs, docs).
- Core complaint: trivial effort by the sender creates large effort for the reader/reviewer.
- People resent being forced into the role of “human in the loop” to debug or fact‑check others’ AI output.
- This is framed as antisocial and disrespectful: “If you couldn’t be bothered to write it, why should I be bothered to read it?”
Effort, respect, and attention
- Human effort is seen as a signal of care, ownership, and accountability.
- Several argue for reciprocity: match your effort to the effort shown by the other side.
- Others push back: what matters is usefulness and quality, not how hard it was to produce.
- Tension: labor‑theory‑of‑value (“effort = value”) vs “value = outcome” is repeatedly debated.
Impact on workplaces and code review
- Common pattern: coworkers pasting large, barely‑reviewed AI PRs or specs, then expecting serious human review.
- Reviewers report spending more time than the “author,” who sometimes can’t explain the code (“Claude added that”).
- This erodes trust, slows teams, and nudges reviewers toward ignoring or rubber‑stamping work.
- Counterpoint: manual PR review “doesn’t scale” in an agentic world; some suggest heavier automation and tests instead.
Proposed norms and coping strategies
- Require authors to self‑review AI output and take responsibility (“you commit it, you own it”).
- Keep PRs small and well‑explained; invest more effort in making work easy to consume.
- Some advocate:
- Label AI‑generated content and allow filtering, with strict penalties for deception.
- Default “no” on low‑effort, high‑volume submissions.
- Use AI to review AI‑generated PRs as a first pass.
- Refuse to read obvious slop or escalate to management.
Attitudes toward AI itself
- Enthusiasts: AI is great for summarization, editing, boilerplate, research assistance, and even civic activism; tool choice doesn’t matter if output is good and checked.
- Skeptics: many outputs remain brittle, fluffy, or incorrect; some find AI content and art viscerally off‑putting or “soulless.”
- Widespread concern about an AI‑to‑AI arms race (spam, hiring filters, support, governance) that shifts costs onto humans caught in the middle.