Is AI All Hype? When Will AI Go Away? (Spoiler: Never)

There are two conversations about AI happening at once. One is the pumping of recycled venture funding in the stock market and the chatter of the media predicting AGI and the end of employment as we know it. The other is quieter and practical, about automating boring work, finding patterns in messy data, or helping people prototype faster. The former is that typical gold rush that happens whenever a new technology grabs the spotlight. The latter is the real utility of AI.

The Gartner Hype Cycle fits AI pretty well right now, lots of inflated expectations and sky-high valuations as the market chases the next Great Unicorn. There’s a real phenomenon happening, where massive sums of money get recycled between a small number of very wealthy oligarchs and their companies, pushing valuations to absurd levels. That matters, not just for markets, but for the environment and for people who worry about what this future means. Large models burn enormous amounts of energy to train, and we can’t ignore these resource and environmental costs that come with the frenzy.

But hype is not the same thing as usefulness. The internet had its dot-com bubble, and it deflated. I lost my first business to that crash. But the useful parts remained and became so ubiquitous that it's hard to remember driving without Google Maps. AI is the same. Over-enthusiasm will collapse in places where the value doesn’t hold, while genuinely useful tools will remain. The question is not whether AI will go away, it is how AI will settle into everyday life.

Why people feel freaked out

The job anxiety is real. Headlines that claim AI will replace every role amplify fear, especially for people early in their careers. It feels dark to imagine your future ruled by a handful of wealthy people and their robots.

The dream of Artificial General Intelligence, something that thinks like a human and takes over everything, is irresistible to the press. It makes for dramatic stories, but it is not the same as the actual useful tools most firms are deploying today.

There’s a cultural loss dimension, too, that I think matters. When we swap slow, handcrafted techniques for faster, automated ones we lose the edges, the raw creativity that produces a Kerouac or Joyce. Think of hand-hewn carpentry versus power tools or a writer’s messy first drafts compared to the sterility of a document edited to death in Word. We definitely lose a lot (hopefully, not all) of those aesthetics but not the underlying human creativity.

The bright side

AI will remain because it solves real, persistent problems. The patterns are pretty consistent. AI helps most where tasks are repetitive, error-prone, or time-constrained (e.g., you don't write tests because you have to launch the product tomorrow), and where automation frees humans to do the thinking that matters more and to simply do more, better work.

Practical examples

• Software: code completion and refactoring suggestions, reducing repetitive typing and helping developers experiment faster, while engineers still design systems, handle complex architectures, and guide the coding agent away from drunken hallucination.
• Legal: document review and contract clause extraction, not replacing lawyers, but slashing the time spent on rote review so lawyers can focus on strategy and negotiation.
• Healthcare: triaging medical images or flagging likely cases for human radiologists, reducing time to diagnosis and letting specialists concentrate on hard cases.
• Operations: automated reconciliation of invoices, expense audit bots, or data cleaning pipelines, cutting hours of tedious work and reducing human error.
• Creative work: idea brainstorming, rough drafts of marketing copy, or thumbnail sketches or "pre-shoots" for design, which creators can then refine and make original.

Why "AI slop" is everywhere right now

Because it’s cheap to produce and still sells. Mass-produced, low-effort outputs work for many business models in the short term of these bubbles, like churn-based content farms or automated AI customer services responses that seem like they should genuinely benefit the consumer (not in my experience). Over time, users are learning to filter this out, and markets will eventually push back against this garbage content. Expect an evolution where quality signals matter more than volume.

How to think about integrating AI

• Map the work first, tool later. Document your processes, track time and mistakes, identify the steps that are rote or rule-based. These are your high-potential wins.

• Start small with pilots. Pick one repetitive task, run a short experiment, measure speed and error rate, then compare to baseline. If it fails, iterate, don’t double down blindly.

• Keep humans in the loop. For anything with nuance or consequence, design a human review step. Let AI suggest, humans decide.

• Measure the right things. Track time saved, error reduction, tip-offs for escalation, and downstream effects like customer satisfaction or regulatory compliance.

• Watch costs. Use the most efficient models. Don't blindly use AI when a simple API will work.

• Invest in skills. Help your team learn how to prompt, evaluate, and integrate AI outputs, while strengthening the uniquely human skills of judgment, synthesis, and storytelling.

• Share successes and failures transparently, so teams learn what to automate and why not to sometimes.

• (Bonus Thought!) Push for policies at the governmental level that measure and make plans to mitigate environmental impact and support fair, future-looking labor transitions. Maybe tax the rich while you're at it. (AI Says: The top US federal income tax bracket rate during most of the 1950s was 91%. This top marginal rate applied to all income over $200,000 for a household (which is equivalent to roughly $2 million in today's dollars). The top federal income tax bracket for 2025 is 37%.)

AI, in all its many forms, is not going away. The hype will settle, the worst excesses will be filtered out, and a new normal will emerge where AI is just another tool. The trick is to avoid getting dazzled by the stories of the apocalypse or promises of technological nirvana, and instead focus on concrete problems where AI actually helps: repetitive, error-prone, and time-constrained work. That is where AI frees people to do the higher-level, (meaningfully) creative, and larger-context-aware work it cannot, and probably never will, do. The real challenge will be to steer that process so it amplifies human work and creativity, while protecting the environment and not just further inflating the wealth of those who already have too much and filling the world with quantity over quality AI slop.