In the world of social networks, we discuss the “network effect”, a phenomenon in which the value of a platform increases as more people use it. It’s so difficult to displace Facebook, Instagram and X because the value is in the ecosystem of connections between people and the content they host.
Something similar could happen in the AI universe. It’s the “personalization effect.” People will become dependent on AI customizations that they will be too lazy to use — or migrate — to other chatbot platforms.
OpenAI is moving to enable more and more customizations through Memory functionality and custom GPTs. Google has also been looking to strengthen the ecosystem around Gemini. This week, the Mountain View company presented the integration of its AI with several YouTube extensions, Google Flights and more.
AI personalization is the next big movement we should watch in the short term. This type of innovation brings improvements in model performance and user experience. But there are points of attention that we need to discuss.
Privacy is one of them. Although OpenAI has implemented control mechanisms so that people can erase their memories, the company makes it clear in its announcement that it will be able to use them to train upcoming models.
The shrinking of plurality over reality is another concern. If the algorithm that assembles the “feeds” on social networks is a sensitive mechanism for the formation of bubbles that shape people’s perception, what can we say about a customized AI that can, at the limit, respond only to what people want? read and listen?
