How to Stay Sane in the AI Snake Oil and Attention-Mongering World
It’s February 2026, and your feed is a firehose of apocalypse and utopia at the same time. “This one prompt made me $47k while I slept.” “AGI by Q3 or we’re all toast.” “AI just wrote a better novel than any human alive.” Every LinkedIn guru, X thread-lord, and VC tweet is selling the same miracle tonic: AI will 10x your life, replace your job, or make you rich—usually all three before lunch.
Welcome to the snake-oil era of artificial intelligence.
The term isn’t new—Princeton professors Arvind Narayanan and Sayash Kapoor literally wrote the book on it (AI Snake Oil, 2024). They define it as “AI that does not and cannot work as advertised.” Think: software that claims to read your soul from a 30-second video interview, or foraging apps that confidently tell you which mushrooms won’t kill you (spoiler: they sometimes do). The hype isn’t harmless. It warps expectations, wastes money, burns people out, and makes the genuinely useful parts of AI harder to see.
Here’s how to keep your sanity while the circus rages.
1. Understand the Game: Incentives, Not Innovation
Most of the noise isn’t about technology. It’s about attention, funding, and status.
- VCs need to justify billion-dollar checks.
- Influencers need clicks and course sales.
- Companies need to look “AI-first” to investors.
- Journalists need traffic in a world where AI summaries are eating the web.
The result? Every incremental improvement gets branded “revolutionary.” Every demo video is edited to perfection. Every failure is quietly buried. This isn’t conspiracy—it’s economics. The attention economy always rewards the loudest claim, and generative AI gave it rocket fuel.
Sanity move: When you see a headline screaming “AI breakthrough,” ask: Who benefits if I believe this? Is there a reproducible demo, third-party test, or just a slick video?
2. Learn to Spot Snake Oil in the Wild
Red flags, straight from people who actually build this stuff:
- Vague magic words: “transformative,” “unprecedented,” “game-changing” with zero specifics.
- Predictive social AI (personality from video, hiring scores from resumes, emotion detection). Narayanan & Kapoor show most of this is modern phrenology—statistically shaky and ethically gross.
- “Agents that will run your business” that still hallucinate and break on Tuesday.
- Get-rich-quick wrappers around ChatGPT that anyone with five minutes can replicate.
- Claims that ignore fundamental limits: context windows, training data ceilings, energy costs, legal landmines.
Real AI wins in 2026 look boring by comparison:
- Code autocomplete that actually saves developers 30-50% time (yes, really).
- Image generators for quick mockups.
- Summarization tools that digest long PDFs.
- Narrow tools that do one thing reliably.
If it sounds like it replaces judgment, creativity, or accountability—treat it as entertainment, not gospel.
3. Build Your Personal Information Diet
Doomscrolling AI Twitter is the fastest way to lose your mind. Counter it deliberately.
- Mute aggressively. Keywords: “AGI,” “superintelligence,” “prompt engineering millionaire,” “AI will replace [job].” Curate follows: actual researchers, engineers shipping code, not professional hype men.
- Long-form only. Read the book AI Snake Oil. Skim arXiv papers. Watch conference talks. Avoid 280-character hot takes.
- One trusted source rule. Pick a newsletter or Substack from someone whose day job involves building or studying AI (not selling it). Check it weekly, not hourly.
- Experiment, don’t consume. Spend 30 minutes every week actually using tools on your own problems. You’ll quickly discover what’s hype (most “AI employees”) versus quietly useful (Claude for outlining, Cursor for refactoring).
4. Protect Your Brain Like It’s the Last Quiet Place on Earth
The mental toll is real. FOMO, comparison, constant low-grade anxiety that you’re “falling behind.” People report burnout from the pressure to “learn AI” while still doing their actual job.
Practical defenses:
- Time-box social media. 20 minutes morning, 20 evening. Use screen-time limits ruthlessly.
- Schedule learning like a meeting. One focused hour per week on a real project (build a small tool, automate something tedious). No passive consumption.
- Touch grass. Literally. Walk without headphones. Talk to humans face-to-face. Create something with your hands. AI can’t replicate the grounding effect of reality.
- Set your own benchmark. Decide what “good” means for you—not whatever viral thread says. Maybe it’s saving two hours a week on email. That’s enough.
5. Keep Perspective: This Is Just Another Technology
Electricity didn’t make everyone geniuses. The internet didn’t end boredom or inequality. AI won’t either. It’s a powerful pattern-matching machine trained on human data. Incredible at some things, comically bad at others.
The people who thrive long-term won’t be the ones who believe the hype hardest. They’ll be the ones who:
- Use AI as a multiplier for real skills.
- Stay skeptical without becoming cynics.
- Build things that matter to actual humans.
The loudest voices today will be footnotes tomorrow. The quiet builders will still be here.
You don’t need to master every new model. You don’t need to have an opinion on every doomer/utopian tweet. You don’t need to turn your life into an AI side-hustle.
You just need to stay curious, test things yourself, and refuse to drink the Kool-Aid—sorry, the snake oil.
The technology is real. The frenzy is mostly theater. Your sanity is non-negotiable.
Keep it. The rest will sort itself out.
(And if an AI ever writes a better article than this one… well, at least I’ll have had a good run.)