AI Is Asbestos in Our Walls - Rip Out the Hype, Keep What Works

AI won't replace your voice; warped incentives might. Be a centaur: use tools for grunt work, keep the thinking yours, and refuse cleanup gigs that dodge blame.

Categorized in: AI News Writers
Published on: Jan 19, 2026
AI Is Asbestos in Our Walls - Rip Out the Hype, Keep What Works

Writers: AI won't replace you - but bad incentives might

AI companies need a story big enough to keep Wall Street hooked. The story is simple: "AI can do your job. Your boss will keep half your salary and pay us the other half."

That narrative isn't about craft. It's about PE ratios, hype cycles, and buying time for monopolies that ran out of organic growth. AI becomes the excuse to squeeze labor while selling investors "infinite scale."

Centaur vs reverse centaur

A good tool makes you faster and braver. That's centaur energy - human judgment with machine support.

A bad deployment turns you into a "reverse centaur": a human appendage for a machine that can't actually do the job, but needs you to absorb blame when it fails. That's the AI "human in the loop" most C-suites are angling for. An accountability sink.

AI can assist your writing. It can't own your voice

AI is a guesser. It predicts the next token. It can draft, summarize, or remix. It cannot carry intent across a thousand choices the way your editorial sense does.

If you force AI to "do the hard part" - the thinking, the throughline, the taste - you leave humans stuck with the worst part: sifting subtle errors that look fine on the surface. That's not creative work. That's cleanup duty.

The business model behind the hype

Once a tech company dominates its market, growth stalls. Growth stocks can't admit that. So they manufacture the next big thing - video, crypto, NFTs, the metaverse, AI - to keep multiples high.

This isn't a conspiracy, it's math. When growth looks soft, the market punishes hard. So "AI will replace humans" isn't truth - it's a pitch designed to defend valuation.

Why AI art feels empty

Great work compresses a complicated internal feeling into a form others can feel. Prompts don't carry enough intent to do that. You can get striking images or passable prose. You don't get voice without a human deliberately infusing it.

That eerie hollowness you notice? That's the absence of a real point of view.

Copyright: your leverage is stronger than you think

Here's the practical edge for writers: in the US, works generated by AI alone aren't copyrightable. No human authorship, no copyright. That means if a client wants AI-only deliverables, they're buying something they can't own. That's a deal-breaker for serious businesses.

Use this to protect your work and fees. If they want IP they can own, they need you. Source: the US Copyright Office has made this explicit (policy and guidance).

Copy/paste contract clauses to protect your work

  • Human authorship warranty: "Deliverables will be created by the Writer and include significant human authorship sufficient to establish copyright."
  • AI use disclosure: "If AI-assisted tools are used for drafting, ideation, or editing, final deliverables will be materially authored, revised, and approved by the Writer."
  • No AI-only deliverables: "Client acknowledges AI-generated content without human authorship may be uncopyrightable; Client requires human-authored deliverables."
  • Training rights: "Client will not use drafts, deliverables, or datasets to train models without separate, written permission and compensation."
  • Attribution (if relevant): "Public use will credit the Writer unless otherwise agreed."

How to avoid becoming a reverse centaur

  • Own the thinking: Use AI for transcripts, summaries, outlines, and alt versions - but make the structure, angles, and argument yours.
  • Keep sensitive work local: If you must use AI, prefer local/open-source tools for drafts, notes, and analysis to avoid data leakage.
  • Refuse "reviewer-of-machine" roles: If a client wants you to rubber-stamp AI drafts at scale, price it as risk management - or pass.
  • Bill for IP, not words: Charge for strategy, research, and brand voice. AI can bulk up word count; it can't replace editorial sense.
  • Set accuracy terms: "Client-provided AI drafts are not guaranteed accurate. Fact-checking is a separate, billable service."

Coalitions that actually work

Fighting AI by expanding copyright into model training is a trap. Big media will capture the benefits long before individual writers see a cent.

What does move the needle: collective power and sector-wide standards. That's how writers have won before - through organized action and agreements that set a floor across employers (sectoral bargaining overview).

What survives when the bubble pops

Hype cycles end. Budgets dry up. Data centers shut down. What remains is what's actually useful: open models that run locally, tools that transcribe, summarize, and edit grunt work, and a lot of affordable hardware for people who make real things.

Think of these as plugins, not replacements. Use them to cut busywork, then spend saved time on craft and client results.

Red flags in AI pitches to writers

  • "We'll replace most of the team with AI; you'll just review." Translation: accountability sink at discount rates.
  • "We don't need copyright." Translation: they don't understand IP risk - or plan to offload it onto you.
  • "Prompt engineer = writer." Translation: they don't value editorial judgment, research, or brand voice.

Practical next steps for writers

  • Productize your brain: Offer strategy sprints, voice guides, messaging architectures, and editorial calendars.
  • Set tool policies: Create a one-pager on when/how you use AI, privacy limits, and QA steps. Clients trust clarity.
  • Build a "human-proof" portfolio: Case studies that show business outcomes, not just samples. AI can't claim those.
  • Join or form a guild/collective: Share rates, clauses, and standards. Power compounds.
  • Keep learning - with intent: Focus on workflows that cut grunt work while keeping your name on the byline.

If you want structured training to level up your centaur workflow, browse curated options by job role here: Complete AI Training - Courses by Job.

The point

AI won't do your job the way investors say. But it can be used to deskill your role if you accept reviewer duty at a discount.

Hold the line on human authorship, price the thinking, and use tools where they actually save time. Be the centaur - not the bolt-on.


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)
Advertisement
Stream Watch Guide