I’ve been an advocate for AI. I’ve written about its promise. I’ve built with it. I still am in fact, but it’s time for a reality check.
I’m building an AI-driven product right now. I still believe in its potential, but I also see what’s broken.
After reading Fortune’s June 11th piece on AI fatigue, I had to take a beat.
“Forty-two percent of companies have scrapped most of their AI projects this year. Forty-six percent of proof-of-concepts don’t make it past the demo.”
(Fortune, 2025)
Since then, the trend has only accelerated, and not in a good way. In early November, Business Insider reported that AI startup Cohere laid off over 50 employees, including core infrastructure engineers, despite a booming year of investment. And just this week, Bloomberg and Yahoo Finance noted how shares of OpenAI-adjacent companies have taken a hit as investor excitement continues to cool amid real-world constraints.
In my opinion, that’s not a failure of the technology. I believe that’s a failure of how we’re using it and who we’re trying to replace in the process.
Right now, AI is a glorified intern, eager, fast, but often clueless. It parrots what it’s seen, not what it knows. It’s pretty good at:
- Summarizing documents
- Generating boilerplate code
- Rewriting copy in new tones
- Auto-tagging media or emails
But it hallucinates. It misreads nuance. It confidently spits out falsehoods. It doesn’t understand context, it emulates it. It’s not capable of creatively adding to your content.
Just ask Google, whose new AI Overviews once advised users to glue cheese to pizza or eat rocks every day for digestion, with errors so bizarre they forced Google to pull back the feature for fixes (Wired, 2025).
Or ask Air Canada, which got sued after its AI assistant invented a refund policy… and lost (CBC, 2024). Or Klarna, whose AI tried to respond to a Python question in, well, literal Python code (Klarna CEO on LinkedIn, 2024).
These aren’t flukes. These are its current, predictable limits.
Here’s the reality: AI isn’t just disrupting. It’s displacing, and not in the way most people think.
Entry-level workers are the first to go. Writers, designers, junior engineers, roles traditionally meant to grow talent and build careers, are being erased in favor of speed and cost-cutting.
Meanwhile, senior professionals are being asked to “work smarter with AI,” which increasingly looks like cleaning up after its mistakes. We’re replacing creation with correction. We’re diluting skills instead of deepening them.
And that has consequences: fewer people entering, fewer opportunities to learn, and a generation of professionals who are asked to review instead of build. And in classrooms, it’s even murkier, where convenience now often takes precedence over effort, and real learning risks being replaced by shortcuts.
There’s another layer here that isn’t getting nearly enough attention: the tax law change under Section 174. Companies can no longer write off R&D in the same year as they now have to amortize over five years (Bloomberg Tax, 2023).
That change alone increased tax bills for innovation-heavy firms at the exact moment when AI became the hot thing to chase.
Layoffs spiked. Meta laid off nearly 25% of its workforce, Microsoft cut 10,000 jobs, and Google slashed over 12,000 positions, with many tied to innovation teams. For startups and mid-size firms, the hit was even harder: some reported layoffs of up to 40% due to unplanned tax burdens (Technical.ly). And instead of pointing to amortization and interest rates, many leaders blamed “AI restructuring.”
The hype hasn’t just created job cuts, it’s created a distorted narrative. Even Michael Burry, the man who predicted the 2008 housing crisis, has warned of an AI bubble inflated by hype, not results.
And just like in 2008, we’ve created financial narratives to justify poor decisions.
But let’s call it what it is: financial maneuvering, dressed up as a type of futuristic transformation.
I want AI to succeed. I want it to help people create, collaborate, and reclaim time for deep thinking. But right now, it’s being positioned as a miracle cure instead of a tool, and that narrative hurts everyone.
I talk to creatives who are terrified.
I talk to engineers who feel sidelined.
I talk to recent graduates who are wondering where the entry-level jobs went.
I have meetings where nobody can explain what an AI strategy actually means beyond vague promises and a chatbot demo.
What passes for an AI strategy today is often just a slide deck, a demo, and a hope that stock price follows suit. That’s not transformation. That’s theater.
And I’m not alone.
According to Business Insider, most CEOs can’t articulate how AI will actually reshape their company. And according to Wired, AI models are fueling misinformation at scale while enterprises roll out tools they barely understand, much less use.
But there are a few glimmers of hope in AI, and here’s where it’s working:
- Lattice, for instance, uses AI to augment HR operations, not replace workers. AI assists in things like surfacing engagement metrics or nudging managers about team performance. As CEO Sarah Franklin describes it, the goal is giving workers an Iron Man suit, not a pink slip.
- McKinsey reports AI can accelerate tasks like data synthesis, contract summarization, and content prep, enabling employees to focus more on judgment and creativity.
- Research cited by Harvard Business Review shows that AI-powered customer support tools improved productivity among junior agents and even enhanced learning through performance feedback loops.
These aren’t miracle deployments, they’re measured, intentional examples of how AI can support human potential when properly scoped and responsibly used.
Press enter or click to view image in full size
What Should We Do?
Here’s what I think responsible use looks like:
- Use AI for grunt work, not judgment. Let it write the first draft, not the final say.
- Keep humans in the loop. Especially in legal, healthcare, and product code.
- Hire juniors again. Mentorship matters! You can’t review code well if you never learned to write it.
- Tell the truth. Be honest about AI’s limits to yourself, your team, and your customers.
- Track human metrics. Burnout and disengagement should be on every dashboard.
The enemy isn’t AI. It’s how we’ve started using it.
If you’re wondering why your team feels tired, uninspired, or vaguely resentful of your new “AI-first Roadmap,” this might be why.
And if you’re reading this and feeling uneasy about where AI is headed, or where you stand in it, believe me, you’re not alone. I’ve written before (and will continue to write) about how anyone, at any level, can use AI to their advantage. Whether you’re starting out, switching careers, or trying to lead through the noise, there’s a way forward. AI should be a tool in your toolbox, not a threat to your seat at the table.
Because people know the difference between a tool and a replacement.
Because creativity doesn’t scale through prompts.
Because innovation doesn’t happen through exhaustion.
Let’s leverage AI and never stop learning!
Let’s build with clarity, not mythology, not marketing decks, and definitely not panic.
Let’s support people, not just platforms.
Let’s treat AI as the tool it is and keep humans as the superior investment.
I mean really, if we get this right, AI doesn’t replace us, it reminds us why we matter.