AI Sanity Check: Ditching the Doom and Hype

Artificial intelligence is moving fast. Faster than most organizations are equipped to absorb.

If you’ve been watching the tech world lately, you’re probably feeling anxious, or at least conflicted about what comes next.  

On one end of the spectrum, social feeds are buzzing with doom and gloom predictions of a white‑collar apocalypse. On the other, optimists imagine a near future where everything is cheap and work is optional.  

Both narratives are compelling. Both are incomplete.

A Sanity Check on the Extremes

Technology alone has never determined our future. Economic incentives, legal accountability, physical constraints, and human motivation have always shaped how major innovations play out. AI will be no different.  

What feels far more likely is a volatile, uneven transition that leaders, companies, and individuals will have to actively navigate.

That tension shows clearly in how people closest to the technology are talking about it. 

What the AI Optimists Are Actually Saying

Matt Shumer’s recent viral post, “Something Big is Happening”, reflects what he’s seeing as the CEO and cofounder of an AI company. He points out how the latest models are crossing a threshold into more autonomous, complex problem‑solving that increasingly resembles human judgment and taste.

His advice to start using these tools now is spot-on. We all should be learning how to interact with these increasingly capable systems.  

Where I’m less convinced is the jump from rapid progress to a jobless future.

A more pragmatic view is this: the fundamentals of our society, our economy, and our humanity will keep us in control.  

The ride ahead may get bumpy, but it’s unlikely to break the vehicle entirely.  

Global Economic Breakdown is Unlikely

Sci-fi stories and doomsday sayers love to paint bleak futures, where a trillionaire elite automates every job, retreats behind AI-powered isolation, and leaves everyone else behind. It’s a great movie plot, but it ignores how economic systems actually work.

Capitalism depends on revenue. Revenue depends on consumers with disposable income. If businesses automate millions of middle-class jobs all at once, the system breaks down quickly. Markets, political power, and public trust follow.  

Mass unemployment isn’t just a social problem, it’s an economic and political one.  

We’ve seen versions of this before. In the early 1900s, Henry Ford reduced work hours, doubled wages, and introduced the five-day workweek. On the surface, it was about reducing turnover. In practice, it helped create a stronger middle class that benefited the entire economy.  

When systems destabilize, intervention follows. The New Deal during the Great Depression is one example. Governments may lag innovation, but they rarely sit idle forever when stability is at risk. That doesn’t mean the transition will be painless, but history suggests survival, not collapse. 

Physical and Legal Brakes Still Apply

What often gets overlooked in AI narratives is friction.

Dario Amodei of Anthropic has shared a thought experiment about a “country of geniuses in a data center,” a massive concentration of AI-driven capability that could reshape global power. It’s unsettling and worth taking seriously.  

But this doesn’t happen overnight.

AI requires staggering amounts of electricity. You can't hide a gigawatt-scale data center. These facilities are expensive, visible, and constrained by physical infrastructure and global supply chains. The physical world still matters.  

Then there’s the legal reality. Who is responsible? Who goes to jail? AI cannot be sued, fired, or held legally accountable. Even if an AI writes perfect code or provides a flawless medical diagnosis, the law requires a human to sign off and take responsibility. Courts and regulators are only beginning to catch up, but they eventually become a governing force.  

AGI: Intelligence vs. Motivation

What about Artificial General Intelligence (AGI)?  

When people talk about a coming “singularity,” the implication is often that AI will wake up, decide it no longer needs humans, and take over. I’m skeptical.  

We often confuse intelligence with desire. They are not the same thing.

There’s a concept in AI research known as the Orthogonality Thesis, which argues that intelligence and motivation are largely independent. A system can be exceptionally good at solving problems without having any inherent desire to act on its own. Humans, by contrast, have motivation hard-coded through biology, survival instincts, status and purpose.  

AI doesn’t have any of that, unless we give it goals.  

That’s where alignment comes in. A poorly specified instruction like “improve the world” can be interpreted in extreme ways when constraints are weak. In a purely theoretical sense, even something as absurd as “eliminate all humans” could be rationalized as a way to reduce suffering or environmental harm. That’s the stuff of sci-fi thrillers and blockbuster movies.  

It makes for great cinema. In practice, the more realistic risks don’t come from machines developing their own ambitions. They come from human misuse, vague objectives, weak guardrails, and misplaced trust in systems we don’t fully understand or supervise. 

The Utopian View

On the other side of the panic is the Star Trek vision of a post-scarcity economy. Elon Musk has talked about a future where AI and robotics make goods and services so cheap that work becomes optional. But this view ignores human nature just as much as the societal collapse narrative does. Humans are status-driven creatures. When basic needs are met, desires don’t disappear, they move up Maslow's hierarchy. We invent new scarcities. Who gets the waterfront property in Hawaii? Who owns the original human-made painting instead of an AI-generated one?

There’s also identity. Untangling human dignity from work wouldn’t be a smooth slide into leisure. It would be messy and disruptive for many people. Star Trek skips the transition phase. Real life doesn’t.

Our Human Advantage

Here’s the hard truth. AI will generate genius-level breakthroughs. We’re already seeing it, including Nobel-recognized advances such as AI-assisted protein-folding research in Chemistry. The natural question is what happens when humans are no longer the primary source of the best ideas.  

AI still lacks desire. It doesn’t care what should be built or why it matters. Humans provide the directives and the purpose.  An AI can help figure out how to cure a disease, but only humans care whether it gets cured. As AI becomes widely accessible, it will generate more of everything. When everyone can create content, options explode and signal becomes harder to find. We’ve seen this before. The world doesn’t need another thousand to-do apps, it needs a few that people actually choose and trust. AI can generate possibilities, but humans still decide what to focus on when multiple paths exist.  

That’s where human leadership remains essential. Trust, judgement, and accountability still matter. Having a human who can look you in the eye, stand behind a call, and take responsibility is something AI will struggle to replicate. 

A Bumpy Ride

I don’t believe we’re headed for a global collapse, and I don’t think a magical utopia is around the corner either. What’s ahead looks like a painful middle path.

Large portions of data-work will be automated this decade, disrupting tens of millions of jobs. That’s not a cliff, but it is a steep slope.

At the same time, new roles will emerge. AI orchestration, validation, governance, and oversight will all require humans. In many cases, AI will look less like a replacement and more like a powerful assistant that still needs direction.

Technology rarely wipes out whole industries overnight. It raises the bar. As Dave Gilboa of Warby Parker once said, “Retail is not dead. Mediocre retail is dead.” The same logic applies here.

This transition will hurt some people. That shouldn’t be minimized. Governments, companies, and educational institutions will need to invest in reskilling and set realistic expectations as systems evolve.

Buckle Up

So, how do we prepare?

1. Embrace Change – Resistance, not age, is what makes people obsolete.
2. Learn AI – Use it. Test it. Understand its strengths and limits.  
3. Manage AI – Treat it like a capable coworker or intern. Delegate, review, guide.
4. Hone Your Humanity – Apply empathy, context, and judgment to AI output.
5. Stay Flexible – Tools will change fast. Systems thinking matters more than mastering one platform.
6. Generalize – Broader thinkers are better positioned to ask the right questions.
7. Watch the Job Market – Signals appear before headlines do.  

The Reality Leaders Should Plan For

AI isn’t here to replace the human soul or to destroy our way of life. It’s here to raise expectations and expand what’s possible. What remains scarce is clarity of intent, accountability, and trust.

The next few years will be uneven. As voices like Matt Shumer and Dario Amodei have noted in different ways, the technology is advancing quickly, but it remains bounded by physical reality, legal structures, and human choice.

This isn’t about collapse or utopia. It’s about navigating the middle with discipline and humanity. If we treat AI as something to guide rather than surrender to, we can adapt and thrive, just like we always have.

I explore this human‑centric idea further in a previous piece on building for humans, even when using AI.