A lot of people assume technological success is measured in sophistication. The bigger, the faster, the smarter—those are the metrics that get attention. But what if that’s the wrong way to think about it? What if the real question isn’t how advanced is this technology? but rather how much does it actually help people?
This isn’t a new mistake. History is full of examples of people marveling at complexity while missing the point. The early personal computers were like this. Engineers obsessed over hardware specs and memory limits while regular people just wanted something useful. The PC only became mainstream when it stopped being about technical prowess and started being about what people could do with it. AI is at the same crossroads now.
The Wrong Metrics
If you ask AI researchers how they measure success, they’ll talk about model size, training efficiency, or benchmarks like accuracy on some dataset. But these are internal measures—like a chef rating a dish based on how difficult it was to prepare rather than how good it tastes. No one cares if you use the most complex neural architecture if the end result is a chatbot that’s slightly better at answering trivia but still bad at understanding what you actually want.
The best technologies fade into the background. The ones that truly change the world are the ones that integrate so seamlessly into people’s lives that they stop feeling like “technology” at all. The internet was once a clunky, dial-up experience. Now, most people don’t think about it. They just use it. AI should be aiming for that—not just impressing researchers with its sophistication.
Breakthroughs vs. Impact
This raises a tough question: should we be chasing technological breakthroughs, or should we be optimizing for human impact? It sounds like a false choice, but it’s not. There’s a real tradeoff. If you’re in a lab, obsessed with making a model that’s 5% more accurate on some benchmark, you might be missing an opportunity to make something useful with an existing model.
A lot of startups get this wrong. They assume that if they just build a better model, the market will follow. But the biggest AI successes so far—Google Search, Alexa, ChatGPT—weren’t the result of the most advanced models; they were the result of models that were just good enough to be useful.
The Real Measure of Success
The real test of an AI system isn’t whether it’s smarter. It’s whether people choose to use it. This is the metric we should be optimizing for: does this AI make someone’s life better in a way they care about?
Think about the most impactful AI tools today. Grammarly helps people write better. Google Photos finds your memories. Tesla’s Autopilot reduces driving fatigue. None of these are the most advanced AI models in the world. They’re just the ones that deliver real value.
It’s tempting to idolize raw progress—faster chips, bigger networks, more data—but ultimately, the best technology isn’t the one that looks most impressive on paper. It’s the one that, quietly and efficiently, makes life better for the people who use it.
So maybe the real breakthrough isn’t about AI getting smarter. Maybe it’s about us getting better at measuring what actually matters.