🌍🚀 Can We Scale AI Without Flattening Culture?
Here’s a question I keep asking lately: How do we structure AI to truly handle cultural & localization adaptation? Because let’s be honest —> no single model today gets it right, we have so many cultures, so many languages.
I still get treated as a man by AI systems, and many platforms don't even respect my native language. Imagine how far we still are from understanding culture. We see it every day: AI misses nuance:
- Images stay the same across cultures unless you know how to prompt. But do the general population know how to keep asking?
- Meanings get lost in translation. And to fix that, ML demands massive datasets, tons of compute, water, energy & trillions of dollars. All that power often chasing benchmarks that don’t connect to real people.
So I keep wondering:
👉 Are we building smarter machines or forgetting what makes humans brilliant? Because the opposite of human-in-the-loop is robotizing everything & losing what makes us beautifully creative, emotional, imperfect, and real.
As a tech executive I live this tension. Speed has always been our holy grail. We celebrate content flying in and out around the globe like lightning. It worked until we started ignoring country market managers, cultural experts, local customers… the people who actually understand what is needed in the real world.
Now I see a widening gap: between doing things fast and doing things right.
More and more I’m an advocate for human oversight, not because we’re falling behind, but because humans think, travel, empathize, and talk to real people. AI should handle patterns, humans should handle meaning. That’s the real partnership.
And while I am at it... I encourage prompt designers and AI trainers: teach models to be culturally aware.
Right now, we’re dividing into two camps:
🚀 “Full AI, speed above all.”
🤝 “AI + Human Interaction.”
I choose the second. It reminds me of the “Universal Spanish” experiment we did 20 years ago, a well-intentioned shortcut that ended up alienating entire countries in Latin America.
Just this week in talking to Simran Khanuja, who’s doing a PhD on cultural adaptation, we went over Pixar’s localization, and how universal stories are never one-size-fits-all. But why do teachers or parents think I cannot relate to an Indian grandmother the same way I relate to a Catalan one? Because context matters, BUT why could I not learn from an Indian grandmother? I'd love to!!! We had a long debate here, there is no easy answer! We both agree that we may be loosing sight of what is being created and that culture may be getting flat as a trend in the world.
At the end of the day, humans will decide if cultural adaptation thrives or disappears.
Let’s not get faster & culturally emptier. Let’s build AI that still feels human and lets the humans weigh in.
💭 What do you think? is AI making us culturally smarter or just faster?
#HumanInTheLoop #CulturalIntelligence #WeShapeAI
Link to revenue report: https://tinyurl.com/5n782ehj