OpenAI has already faced criticism for its ChatGPT app recommendations, which some users see as shilled in-app ads. The issue garnered national attention when Yuchen Jin, co-founder of AI startup Hyberbolic, shared a screenshot of ChatGPT advising a user to download the Peloton app. This kicked off a fierce debate all over the social media app X. While a screenshot may have received close to 462,000 views, it begs the question, why are these app suggestions even relevant or desired in the first place.
The conduct at issue was ChatGPT’s recommendation of the Peloton app to a consumer who subscribed to Apple Music. This led to furious backlash from users who began to notice that the recommendations no longer had any relevancy. In an attempt to clarify, Daniel McAuley, OpenAI’s data lead, explained that the Peloton recommendation was not a marketing solicitation.
“Hey, Kol. Thanks for flagging 🙏 This is not an ad (there’s no financial component). It’s only a suggestion to install Peloton’s app. But the lack of relevancy makes it a bad/confusing experience. We’re iterating on the suggestions and UX, trying to make sure they’re awesome.” – Daniel McAuley (@_dmca)
As for that app suggestion feature, OpenAI said this is currently in a private beta. It is EU-logged-in-user only, but available to logged in users from outside-of-EU, Swiss and U.K. The company emphasized that users could discover apps when ChatGPT suggests one at an appropriate moment or by calling them by name, integrating interactive interfaces directly within the chat.
Despite the clarification, concerns lingered among users. One person illustrated this annoyance by lamenting ChatGPT’s persistent suggestions that they use Spotify. This just increased the backlash, especially over the system’s inability to deliver appropriate, customized recommendations.
Sarah Perez, a reporter for TechCrunch, recognized the ramifications of these kinds of recommendations in her reporting. She then underscored that OpenAI wants to make app discovery the best possible experience for users. Yet the actual execution has faced some significant hiccups.
OpenAI is in the early stages of developing and iterating on its app integration feature. Yet it has the almost impossible task of increasing relevancy while increasing user trust. The reception from users will help inform what future versions of this new feature look like.
