Gemini surpasses ChatGPT in usage: Android integration is key

What happened?

A recent report from PhoneArena, cited by Hipertextual, reveals that Gemini, Google's artificial intelligence assistant, is used up to 10 times more than ChatGPT on Android devices. This figure is not due to Gemini's technical superiority, but to its deep integration into the Android ecosystem: it comes pre-installed on most new devices and has replaced Google Assistant as the default option. According to StatCounter data, Android holds approximately 70% of the global mobile operating system market, giving Gemini a potential user base of over 3 billion devices. In contrast, ChatGPT, although available as an app on Google Play, requires voluntary download and installation, adding friction. This phenomenon recalls Microsoft's strategy with Internet Explorer in the 1990s, when its integration into Windows allowed it to dominate the browser market, even though competitors like Netscape offered superior features.

Why is it important?

The shift in habits regarding the adoption of AI assistants has profound implications. OpenAI, which led the conversation with ChatGPT since its launch in November 2022, now faces Google's distribution advantage. Native integration reduces friction: users activate Gemini with a gesture or voice command without needing to open an app. According to a study by App Annie, the ChatGPT app has been downloaded over 500 million times, but its daily active usage is lower than Gemini's on Android, according to the PhoneArena report. This reflects a pattern already seen in the browser wars: default distribution can overcome perceived quality. Furthermore, Gemini's integration into services like Gmail, Maps, and YouTube (announced by Google in May 2024) could further increase its usage, creating a sticky ecosystem similar to what Microsoft achieved with Office.

Consequences for the market

For Google, this move consolidates its position in generative AI, but also attracts regulatory risks. The European Commission is already investigating potential anticompetitive practices in the voice assistant market, and the forced integration of Gemini could be seen as an abuse of dominant position, similar to fines imposed on Google for its price comparison service (€2.42 billion in 2017) or for Android (€4.34 billion in 2018). For OpenAI, the challenge is twofold: it needs to integrate more deeply into mobile ecosystems — perhaps through agreements with manufacturers like Samsung or Xiaomi — or differentiate itself with superior capabilities. In complex reasoning tests, such as the MMLU dataset, ChatGPT-4 still outperforms Gemini Pro 1.5, according to benchmarks from May 2024. For users, competition is positive, as it drives improvements in both products. However, there is a risk that Google will use its position to limit choice, as happened with Google Assistant, which marginalized alternatives like Alexa on Android.

“Gemini is used up to 10 times more than ChatGPT on Android phones, not because it's better, but because it's pre-installed and replaces Google Assistant.”

What should readers know?

• Gemini's advantage is primarily distribution, not quality. In blind tests, users preferred ChatGPT's responses in 65% of cases, according to a study by Artificial Analysis from March 2024.
• ChatGPT remains superior in creative and complex reasoning tasks, such as scriptwriting or solving advanced math problems.
• Gemini's integration could expand to more Google services, such as Google Docs or Calendar, increasing its dependency.
• OpenAI could respond with agreements with manufacturers (e.g., pre-installation on Samsung devices) or improvements to its mobile app, such as integration with the system's voice assistant.
• The PhoneArena report is based on app usage data on Android but does not include iOS, where ChatGPT might perform better due to the absence of Gemini as default.

In conclusion, Gemini's usage lead does not reflect technical superiority but an integration strategy that Google has exploited, reminiscent of the Internet Explorer era. The battle for mobile AI is just beginning, and the outcome will depend on whether regulators intervene, OpenAI's ability to close the distribution gap, and user response to potential lack of choice. What is clear is that in the world of technology, sometimes the best product doesn't win, but the one that is most accessible.