AI FOMO: How Long Does the Hype Last?
And the 4R Framework for Sanity and Progress
The last Google I/O did it again. If you’re feeling overwhelmed by the nonstop firehose of “AI breakthroughs,” you’re not alone—and the numbers back you up.
🔎 SIGNAL
In a recent audit of 23 major AI product launches from the past 18 months, the average hype cycle—defined as the period between viral launch and noticeable user drop-off—was just 28 days1. Only 3 of those launches sustained meaningful engagement beyond 60 days, and none maintained their peak longer than two weeks.
Sora, OpenAI’s text-to-video model, hit 50M+ impressions on X within 72 hours of its announcement in February 2024. Released to Pro User in December 2024, today usage is marginal, community documentation is sparse, and developer activity has flatlined (take a tour around Reddit to get a feeling of what real people say).
This pattern mirrors what Gartner once called the “peak of inflated expectations” but in today’s accelerated news cycle, the descent happens faster, and deeper. Startups pivot before features mature. Enterprises stall in pilot purgatory. Users get burned out on broken promises.
What drives this?
AI FOMO (Fear of Missing Out) is engineered. Model demos are optimized for virality, not utility. Add a few “experts” on LinkedIn and X competing for dopamine points, and voilà—you've got a treadmill we can't seem to step off.
🎥 STORY | Remember Humane's AI Pin?
In 2023, Humane, a San Francisco-based startup founded by ex-Apple employees, unveiled the AI Pin. “The next big thing in personal computing.” Backed by over $230 million from high-profile investors, including Sam Altman and Marc Benioff, the AI Pin promised to revolutionize the way we interact with technology.
The hype was palpable. Demonstrations at TED Talks and Paris Fashion Week showcased the device's capabilities, and Time magazine even listed it among the "Best Inventions of 2023" before its release. The anticipation was so high that the company projected sales of 100,000 units by the end of the year.
However, reality didn't match the expectations. Upon its release in April 2024, the AI Pin faced scathing reviews. Users and tech reviewers criticized its performance, citing issues like poor battery life, unreliable voice recognition, and a lack of essential features. Influential tech reviewer Marques Brownlee labeled it "the worst product I've ever reviewed."
Sales plummeted, and return rates soared. By August, more devices were being returned than sold. Internal reports revealed that employees had raised concerns about the product's readiness, but these were dismissed by leadership. The company failed to appoint a head of marketing, and a senior software engineer was reportedly fired for questioning the launch timeline.
By February 2025, Humane ceased operations, and HP acquired its assets for $116 million. The AI Pin was discontinued, and existing devices were rendered non-functional as servers were shut down.
How shall we react, then, to the “next breakthrough!!!!”?
🧭 YOUR HUMAN OVERRIDE
We need a way off the treadmill. Not because AI isn’t transformative—it is—but because the way we talk about AI is making us dumber, not smarter. Here’s a pragmatic framework I use with clients to break the hype cycle:
The 4R Framework: Resist, Reframe, Reduce, Reinforce
1. Resist the viral instinct
If an AI model is trending, pause. Before forwarding it to your strategy team, ask:
Is this solving a real pain point we’ve already identified?
If not, archive it and move on.
2. Reframe evaluation metrics
Most teams judge AI tools by their wow factor. Flip that. Judge them by:
Stability under imperfect input
Explainability to non-technical users
Regulatory and data governance alignment
Cost-per-unit of value generated
If the answers are fuzzy, skip it.
3. Reduce pilot fatigue
Too many orgs spin up pilots with no guardrails. Instead:
Cap your AI pilots to 3 at a time
Set a “Sunset Date” (30 days max without traction → kill it)
Require a business sponsor to co-own each pilot
This creates space for deep work, not shallow tinkering.
4. Reinforce quiet successes
We glamorize disruption, but ignore improvements. Make “boring AI” sexy again.
Share internal wins on OCR, workflow triage, fraud detection
Incentivize teams who improve KPIs with AI (not just who try new models)
Build case studies around internal tooling and processes, not just customer-facing demos
When I speak with policy leaders or execs, I urge them to treat AI not as a revolution, but as a reorganization of attention and process. Yes, in the long term, if done properly, it’ll be seen as a revolution. Done badly, it leads to civil unrest.
We need to reward operational excellence over novelty-chasing. Celebrate the teams reducing call center resolution times by 40%, not the teams fiddling with unstable LLM plugins.
Most AI’s power today lies in the 70% use-case zone: augmenting existing workflows. Not replacing humans. Not redefining industries. Just making the system suck a little less.
There’s nothing headline-worthy in that. But there’s a future in it.
🔥 SPARK
Let me ask you something uncomfortable:
What if our obsession with the next big thing is just a socially sanctioned form of procrastination?
Because confronting the real work—of change management, of process redesign, of cultural alignment—is hard. It’s easier to dream of a shortcut. A silver bullet. A new model that will let us bypass the friction and arrive directly at the future.
But the future isn’t arriving.
We have to build it. Slowly. Iteratively. Often clumsily.
The problem with the hype treadmill isn’t just that it wastes time. It erodes trust. It teaches us that AI tools are overpromised and underdelivered. And eventually, it turns smart people cynical. That’s a loss we can’t afford.
So this week, instead of retweeting the next GenAI demo reel, ask:
Where in my team’s workflow could basic automation save hours?
Which AI pilots deserve to be killed?
Who is doing great “invisible” AI work that should be celebrated?
And if you want a deeper dive on this topic, here’s what I recommend:
Algorithmic impact assessment in healthcare - ADA Lovelace Institute
Resisting AI Solutionism: Where do we go from here? - A CHI 2025 Workshop
Let’s stop running in place.
Let’s walk (deliberately) toward real progress.
___
PS—If you need my help, there are a few ways we can partner: head over here and let me know.
Footnotes
Internal tracking analysis conducted across Reddit, GitHub commits, Google Trends, and LinkedIn mentions