The Dark Side of AI. What You Must Know.
Read time: 6 minutes
Welcome to AI-Empowered Leaders. In this weekly email, I share actionable advice on AI adoption, use cases & strategic thinking from my experience as AI Trainer, Leadership Coach, and Consultant.
AGI is a marketing term. And 3 other AI truths worth weighing.
If you've followed me for a while, you know where I stand.
Critical Thinking + AI.
It's what I teach companies. What I lecture at universities. It's how I differentiate my executive AI advisory.
Critical thinking means looking at things from all angles. Not just the comforting one.
The dark side, too.
No fast-moving industry ever arrived without a cost. AI is no exception.
As we rush to adopt it, there are things we have to see clearly. Things we have to weigh as we make decisions.
Ignoring them is closing your eyes. That's not what we do.
We look at all angles. Then we decide.
So today: The dark side of AI.
Not to fearmonger. Not to slow you down.
But to inform you. So you make shaper calls on strategy, tools, and governance.
Pass on the knowledge so we can build a better AI age together.
The Briefing
Karen Hao spent years reporting on OpenAI for MIT Tech Review, the WSJ, and The Atlantic.
Her book, Empire of AI, draws on 250+ interviews and internal documents. OpenAI declined to cooperate. Sam Altman criticized it publicly. Make of that what you will.
Her core argument.
The AI industry isn't chasing science for humanity. It's building empires. (No surprise here, given the amount of capital behind these companies)
However, they use the language of utopia to distract from what's happening right now.
Extraction. Exploitatoin. Concentration of power.
I don't necessarily buy all of it. The empire frame feels heavy, and AI is already creating a ton of real value in every industry.
But the specifics she documents deserve your attention.
Not because the robots are coming.
Because the business model has costs that rarely show up on the slide deck you're being sold.
The Real Story
AGI (Artificial General Intelligence) is a marketing term, not a milestone.
Sam Altman, founder and CEO at OpenAI, framed it differently depending on who he is speaking to.
To consumers: AGI is this amazing digital assistant that will solve all your problems, so you should definitely keep paying for ChatGPT.
To Microsoft: a system that generates $100 billion in profit.
To Congress: a civilization-scale technology too important to regulate.
That's not a scientific term. It's positioning.
Why it matters for you: when a vendor sells you "the road to AGI," they're selling a horizon that never quite arrives. Buy capabilities you can use this quarter. Not promises about 2030.
The displacement story is real. And more complicated than the headlines.
Klarna became the poster child. It replaced around 700 customer service agents with an OpenAI-powered bot and bragged the AI did the work of all of them.
Then came the part the doom headlines skip.
Klarna's CEO admitted the AI-only approach produced "lower quality" service and started hiring humans again.
Gartner now predicts half of companies that cut customer service staff for AI will need to rehire by 2027.
The lesson isn't "AI won't touch jobs." It will.
The lesson is that full replacement often fails on quality, and the savings evaporate when you have to unwind it.
The real risk to your business isn't being too slow. It's automating the wrong thing, badly, and paying for it twice.
Someone pays the price you don't see.
There are at least two costs that are being hidden from us.
- Human: Behind "safe" AI outputs sit data annotators in Kenya, the Philippines, and elsewhere, reviewing the worst content on the internet for a few dollars an hour to train the filters. The trauma gets outsourced so your experience feels clean.
- Environmental. Data centers consume enormous electricity and water. Communities from Chile to Arizona have protested water use in drought-stressed regions. This is measurable, and it's happening now. It just doesn't make the keynote.
You don't have to solve any of this. But calling the technology "clean" is a thinking error. Know the full cost of what you adopt.
Watch how they treat dissent.
Hao documents a consistent pattern. People inside these companies raise concerns, then meet NDAs, pressure, or a quiet exit. One former OpenAI researcher said his team was pushed to behave like a "propaganda arm."
I'm not here to insinuate anything beyond what's documented. But the pattern itself is a signal.
What it tells you as a buyer: a company that suppresses its own internal critics is unlikely to be straight with you about its product's limits. Trust the vendor who tells you what their tool can't do.
The Playbook
Don't doomscroll. Decide better. Four questions to run before your next AI move.
- Buy capability, not destiny. Ask "what can this do for us this quarter?" Ignore the AGI roadmap.
- Automate tasks, not people. Pilot on a narrow task and measure quality, not just speed. If quality drops, you've spotted the Klarna trap before it finds you.
- Demand the limits. Ask any vendor "what is your tool bad at?" The honest answer tells you more than the demo ever will.
- Price the full cost. Governance, oversight, and the rework when AI gets it wrong are part of the bill. The cheapest tool is rarely the cheapest decision.
- Manage the Risk: Anchor to an established standard: ISO/IEC 42001 (the international AI management standard), the NIST AI Risk Management Framework (US, voluntary), or the EU AI Act if you operate in Europe. You don't need full certification on day one. You need a shared checklist so AI risk stops living in one person's head.
The Monday Test
This week, pick one AI tool you already rely on. One you'd feel the loss of if it vanished.
Then sit with one question: What am I not seeing because this works so well?
Whose labor trained it. What it quietly gets wrong. What you've stopped checking because the output looks clean.
You don't have to act on the answer. Just refuse to look away from it.
That's the whole discipline. Comfort makes us stop asking. Leaders keep asking anyway.
Keep that in mind for future discussions and decisions.
Optimism is easy. Informed optimism is leadership.
As promised, click here to access the AI Strategy Consultant:
Whenever you’re ready, here’s how I can help you win with AI:
1) AI Business Advisory
Spot, plan & launch AI use cases that save hours and unlock new value.
2) AI Enablement
Take your team on a journey from AI beginners to critical-thinking power-users—working securely across tools, saving costs, and driving results.
I’ve already trained and coached 3,000+ leaders who are saving hours and performing at a higher level. Your team could be next.
Have questions? Hit reply to this email, and I'll help out!
Talk soon,
Alex