No AI Slop Manifesto – A Call for Human‑Centred AI in Software Development
TL;DR – What the No AI Slop Manifesto Declares
The manifesto, authored by speakers and attendees of the AIE Code Summit 2026, defines slop as AI‑generated output that requires the consumer to invest more effort to understand than the producer invested to create. It asserts that AI should amplify, not replace, human judgement and that teams must prioritize outcomes, understanding, and system‑level fixes over raw token counts or PR size.
1. Core Definition of Slop and Why It Matters
Conclusion: Slop is the asymmetry between minimal production effort and maximal consumption effort, and it threatens code quality, maintainability, and human agency.
- Definition (manifesto): “Slop is when we put less effort into producing something than we ask the consumer to spend understanding it.” This captures the core risk of AI‑assisted development: cheap generation of large artifacts (e.g., thousand‑line PRs, ten‑page docs) that remain expensive to review.
- Why it matters: The cost of producing AI output approaches zero, while the cost of human review stays constant, creating a bottleneck that can degrade system reliability and team morale.
- Community echo: Commenter @seizethecheese notes that defining slop this way could cap AI productivity, citing Terry Tao’s view that mathematicians now spend most of their time understanding AI‑generated proofs. The tension highlights the need for a nuanced balance between productivity and comprehension.
2. Guiding Values – What the Authors Prioritize
Conclusion: The manifesto’s five values place human‑centric outcomes above raw AI output.
| Value | Intended Effect |
|---|---|
| Outcomes over output | Measure success by impact on users, not by lines of code or tokens spent. |
| Understanding over speed | Encourage deep comprehension before shipping. |
| Our own thinking over a model’s first draft | Treat AI suggestions as drafts, not final decisions. |
| Better work over easier work | Use AI to raise quality, not to shortcut effort. |
| Improving the system over patching the output | Invest in tooling, processes, and guardrails rather than fixing symptoms. |
Commentary: @gwbas1c praised the outcome‑centric line, emphasizing that “only the result matters” despite massive token expenditures.
3. Key Beliefs – The Rationale Behind the Values
Conclusion: The manifesto’s twelve beliefs explain why slop proliferates and why human oversight remains indispensable.
- AI is a powerful tool, but a swarm can outproduce human review capacity.
- Slop arises when effort spent on production is less than effort required for consumption.
- AI output is not inherently slop; misuse is.
- AI amplifies human judgement (or its lack).
- Production cost is near‑zero; review cost is constant → asymmetry.
- Even pre‑AI, “adding‑instead‑subtracting” tools generated slop.
- Organizational pressure fuels slop creation.
- Slop compounds: each sloppy artifact seeds more sloppy artifacts.
- Team morale erodes as slop accumulates, degrading communication and quality.
Commentary: @BrandoDev_it highlights the scaling problem: as LLMs generate more code, human review becomes the critical bottleneck.
4. Resolutions – Concrete Commitments for Teams
Conclusion: The manifesto lists actionable resolutions that translate values into day‑to‑day practices.
- Treat human time as a scarce resource; optimize its use.
- Fight slop because AI should improve work, not merely ease it.
- Design tools and processes for efficient intent elicitation and comprehension.
- Do your own thinking before accepting model suggestions.
- Own the intent, taste, and consequences of shipped work, regardless of AI assistance.
- Disclose AI‑generated content and verify critical parts before release.
- When receiving slop, give courteous feedback and request a review.
- Accept that some “alpha” (value) remains in any review; focus on high‑stakes artifacts.
- Use linters, agents, and verifiers, but still perform selective human review.
- Prioritize shipped impact over superficial metrics (lines, PR count, token usage).
- Recognize that not all unreviewed output is bad; high‑leverage one‑shot prompts can be acceptable when outcomes remain solid.
- When repeated mistakes surface, improve systemic guardrails rather than relying on ad‑hoc fixes.
Commentary: @courtcircuits suggests attaching the original prompt to AI‑generated PRs to make the reviewer’s job easier—a practical extension of resolution 8.
5. Community Reactions – Themes from Hacker News Discussion
Conclusion: The discussion surfaces three recurring themes: measurement focus, scalability of review, and the need for clearer terminology.
- Outcome‑centric metrics: Multiple commenters (e.g., @gwbas1c, @a3w) question whether market pressure will tolerate a shift away from velocity‑based KPIs. The manifesto’s emphasis on impact over token counts challenges prevailing performance metrics.
- Scaling human review: @BrandoDev_it and @random_cat_8745 both note the difficulty of scaling review as AI output volume grows, reinforcing resolution 9’s call for selective human oversight.
- Naming and framing: @wauter proposes a more positive name like “Human‑Serving Engineering Manifesto,” while @ozgung criticizes the opening question‑prompt as “slop” itself, highlighting the delicate balance between defining slop and avoiding meta‑slop.
- Interpretation of “alpha”: @InexSquirrel asks for clarification of “alpha” in resolution 8, indicating that terminology needs explicit definition for broader adoption.
6. Practical Takeaways for Teams
Conclusion: Teams can start implementing the manifesto today by adopting three low‑friction practices.
- Prompt Transparency: Include the exact prompt (or prompt template) alongside any AI‑generated PR or document. This satisfies the desire for reviewer context and aligns with resolution 8.
- Impact Review Checklist: Before merging, ask: What problem does this solve? Who benefits? If the answer is vague, treat the artifact as potential slop.
- Scheduled Slop Audits: Allocate a regular (e.g., bi‑weekly) slot for a small group to audit recent AI‑generated artifacts for clarity, correctness, and unnecessary complexity.
7. Final Reflection – The Manifesto Is a Work in Progress
Conclusion: The authors acknowledge that the manifesto is a living document; its value lies in sparking ongoing dialogue about responsible AI use.
“None of us has this figured out. We're each relearning what we will and won't accept, and there's no single right answer.” – Manifesto conclusion.
The community’s diverse reactions—ranging from enthusiastic endorsement to skeptical critique—show that the conversation about AI‑driven slop is just beginning. By grounding discussions in the concrete definitions, values, and resolutions outlined above, teams can begin to shape a culture where AI amplifies human judgment rather than eroding it.