According to PYMNTS.com, new survey data shows that among U.S. consumers who’ve used a dedicated AI platform, a whopping 83% have tried OpenAI’s ChatGPT. That compares to just 48% for Google’s Gemini and 30% for Microsoft’s Copilot. The data, which measures whether someone has used a tool at least once, captures early habit formation. On smartphones, the story changes due to defaults: 40% have used ChatGPT’s mobile assistant, 40% Google Gemini, and 37% Google Assistant. The report highlights that ChatGPT is becoming the primary reference point for conversational AI, while Google’s upcoming shift from Assistant to Gemini and Apple’s struggles with Siri show how crucial this early lead is.
First Impressions Last
Here’s the thing about this data: it’s not measuring daily active users or deep engagement. It’s measuring exposure. And that might be even more important. In consumer tech, the first platform you use for a new category often shapes your entire understanding of it. It sets the expectations. So, when 83% of people dipping their toe in the AI water try ChatGPT first, that’s a monumental head start. They’re learning what AI “is” from OpenAI’s interface and capabilities. For Gemini and Copilot, this creates a huge problem. They’re not just competing on features; they’re fighting against an already established mental model. Why would the average person even bother trying another one if the first one they used seems to work just fine?
The Default Trap
Now, the mobile data is fascinating because it shows the power of the default setting. The near-equal usage between ChatGPT, Gemini, and Google Assistant on phones isn’t about choice. It’s about what’s shoved in your face. Google Assistant’s numbers are basically a legacy report—a testament to being pre-installed on a billion Android devices for years. But that advantage is fragile. Google is replacing Assistant with Gemini, and the data suggests most of Gemini’s future mobile usage will just be that default inheritance. Users won’t actively choose it; they’ll just get migrated. That’s powerful for Google, but it’s a different kind of power than ChatGPT’s. One is about habit through discovery, the other is about habit through inertia.
What About Apple?
And then there’s Apple. This data should be a five-alarm fire in Cupertino. The iPhone is arguably the most powerful distribution platform on the planet, yet Siri is completely absent as a primary AI entry point. Let that sink in. Default placement, it turns out, isn’t enough if the capability isn’t there. People aren’t using Siri for this new wave of AI tasks because, well, it’s not really built for them. It’s why Apple is scrambling, talking to Google and others, for a lifeline. They have the users, but they’ve completely missed the initial habit-forming window. Playing catch-up in AI isn’t just about building a better model; it’s about convincing users to break a habit they’ve already formed elsewhere.
The Stakes For Everyone Else
So what does this mean for the rest of us? For users, it probably means less choice in the long run. Markets that consolidate around a single early leader are hard to disrupt. For developers and businesses building on AI, it makes the platform choice clearer. Betting on the ecosystem where the users already are is usually the safer play. This early data suggests the conversational AI landscape is hardening faster than anyone expected. We’re not in a period of wide-open experimentation anymore. Habits are locking in. And for any company trying to compete, whether in software or even in hardware that integrates these systems, the question isn’t just “Is our AI good?” It’s “Can we get people to care enough to switch?” That’s a much, much harder problem to solve.
