The Wrong Way to Do AI (And Why Everyone’s Doing It)
Duolingo’s $3B AI Mistake: The Real Lesson is Leadership, Not Technology
The Duolingo Meltdown
How a viral brand destroyed its own goodwill in the name of efficiency
In May 2025, Duolingo went dark.
Its TikTok and Instagram accounts, once buzzing with humor and viral stunts, were suddenly wiped clean. In their place: cryptic captions, dead rose emojis, and silence.
The cause? A memo from CEO Luis von Ahn announcing that Duolingo would become “AI-first.” Contractors would be phased out. Hiring would only continue if teams could prove a human was essential. AI fluency became part of employee performance reviews.
What followed wasn’t pride or excitement. It was panic. A masked employee appeared in a now-infamous video, wearing the company’s owl mascot and stating flatly, “Duolingo was never funny. We were.” It was a protest. And it hit a nerve.
The company pulled back. Social feeds were deleted. Comments turned off. The silence was louder than any apology.
This wasn’t a tech failure. It was a leadership one.
Duolingo’s success was built on cultural understanding, emotional resonance, and the human touch. But the shift to AI-first treated those qualities as irrelevant. The very elements that made Duolingo different were discarded in service of speed and efficiency.
Replacing people with AI might look good on a spreadsheet. But it erodes trust, weakens the brand, and breaks the creative engine that drives real value.
The real lesson isn't about AI. It's about intent.
So ask yourself: What are you optimizing for? Short-term savings? Or long-term growth of customer loyalty, employee capability, and your ability to lead with clarity in a world where AI is no longer optional?
Because the technology won't lead. You will.
Cutting Staff, Cutting Corners
Why using AI to reduce headcount is exactly the wrong instinct
Let’s be honest. Many leaders saw the AI boom and immediately thought one thing: efficiency.
Reduce headcount. Streamline operations. Automate the mess. Boards loved it. Investors applauded. Consultants were ready with playbooks. AI became the latest excuse to do what companies have always wanted to do, cut costs faster than their competitors.
Duolingo wasn’t alone. Salesforce laid off over a thousand staff in February, only to turn around and hire two thousand AI-focused roles. The net message to employees? You’re replaceable. The company will survive, just not with you.
You don’t build loyalty by making people feel disposable.
What these companies missed is the deeper cost of treating AI as a shortcut. Efficiency isn’t just about doing the same work with fewer people. It’s about designing systems that unlock new levels of performance, creative, cultural, operational, that weren’t possible before.
That only happens when humans and AI are in partnership. Not opposition.
When you replace expertise instead of enhancing it, you lose more than just headcount. You lose context. Judgment. Institutional memory. The unwritten rules that keep your company from drifting into chaos. And most dangerously, you lose trust.
Because no matter how good the AI is, your workforce is watching. And they know whether they’re being asked to grow or to get out of the way.
If you use AI to cut corners, your people will pull back. They will stop sharing insights. They will stop experimenting. They will do just enough to survive.
But if you use AI to elevate them, to take away the drudge work, to help them operate at a higher level, they will give you everything they’ve got.
The lesson is simple. You don’t get extraordinary results by shrinking your people. You get them by expanding what they’re capable of.
Beyond Automation: Building Human-AI Teams
The companies that win won’t replace humans. They’ll design systems that elevate them.
AI isn’t here to replace your team. It’s here to give them leverage.
Done right, AI agents act like a support crew for every worker. They handle the repetitive tasks, synthesize the data, draft the first versions. That doesn’t make people less important. It makes them more effective.
Imagine a team where each person leads a small orchestra of AI agents. One generates options. Another checks for compliance. A third tracks performance and recommends the next best action. The human still makes the call. But they do it with better insight, faster feedback, and less friction.
The companies that learn how to orchestrate these teams will outperform those that treat AI like a blunt instrument for headcount reduction. The key is knowing what to automate and what to elevate.
Some tasks can and should be fully automated. But others demand human oversight, judgment, or creativity. The best systems are hybrid by design. They match the right type of work to the right kind of intelligence, human or machine.
Luis von Ahn understands this, at least in part. On the No Priors podcast, he laid out a bold vision: AI tutors that personalize learning for every student, freeing schools to focus more on care and community. In countries like South Korea, that future is already arriving. It’s a strong, necessary idea.
But vision without skilful execution is dangerous.
Duolingo tried to take that sharp turn at full speed. They rolled the car. Not because the destination was wrong, but because the driver ignored the limits of the road.
AI gives us power. Leadership decides how we use it.
Designing for Trust and Transformation
The updated 7-step plan for bringing your people with you
The fear that AI will replace people isn’t irrational. It’s protective.
In many companies, workers are already pulling back. They hesitate to share knowledge. They avoid contributing to pilot programs. Not because they don’t care, but because they don’t trust what comes next.
The only way through is to build that trust deliberately. That means starting the AI conversation before fear fills the silence.
This is where leadership earns its reputation. Not in the moment AI is deployed, but in the months before, when choices are still reversible and people are still listening.
The most effective leaders don’t just adopt AI. They guide their teams through it. They don’t soften the message, but they do shape the experience. The plan below, developed with business psychologist Aparna Uberoy, is built for that.
Here’s how you lead your organization through an AI transformation without leaving your people behind:
Prepare for External Influence - Acknowledge that employees hear about AI everywhere. Be the source of truth.
Communicate Early and Honestly - Don’t wait. Set the tone before speculation takes hold.
Invite Employees Into the Process - Surveys and workshops aren’t cosmetic. They shape buy-in.
Redesign Roles Together - Make it clear what AI will take off people’s plates—and what new value they can create.
Include Employees in Tool Development - Use their insight to make the tools better. Let them see their fingerprints on the solution.
Pilot with Purpose - Start small. Share results. Celebrate what works and fix what doesn’t.
Empower with Training -Give people the skills to work with AI, not fear it.
Transformation is not a tech project. It’s a leadership decision. Trust is what makes the shift sustainable.
5. What Duolingo Could Have Done Differently
How to lead AI transformation without sparking revolt
Duolingo had the right destination. AI tutors that personalize learning at scale is a powerful vision. But the way they handled the turn cost them trust, talent, and momentum.
It didn’t have to.
They could have started by using AI to support human experts, not replace them. Let translators work faster. Let educators focus on creativity. Preserve cultural nuance and insight.
They could have trained their people to work with AI as a partner. And they could have communicated the shift early, with clarity and care.
Instead, they treated AI as a cost-cutting tool. Their people felt it. And they pushed back.
Vision matters. But vision without trust doesn’t move people forward. It drives them away.
6. A Moment of Choice: Your Roadmap Starts Now
What leaders must do as GenAI and agents reshape business
AI is here. The question is whether your leadership is ready for it.
Start with your data. If it's untrusted or fragmented, your AI will be too. Clean data isn't a technical task. It's a leadership priority.
Next, your processes. Automating broken workflows only makes things worse. AI doesn’t fix bad design. It amplifies it.
Finally, your ambition. The best leaders aren’t chasing use cases. They’re rethinking how their business works. Where can agents create leverage? Where can people do more of what only people can do?
Agents Unleashed is your roadmap. Chapter 5 lays out how to redesign your organization with AI and your people at the center.
And if you want to talk about what that looks like in practice, book a 30-minute call with me. No pitch. Just a conversation about how to move from theory to execution—without losing your team along the way.