Since its release to the general public at the end of last year (November 30 to be exact) ChatGPT became the star of the moment. Let’s recall that until then the center of gravity of generative artificial intelligence was mainly focused on models generating images from text prompts, although it is true that another creation of OpenAI, GPT-3, was already on the lips beforehand. .

ChatGPT quickly became a reference thanks to the great work done by its creators. From the first moment, we were aware of the limitations and failures of the chatbot, which we subsequently saw, can be extrapolated to other similar services, in addition to gradually learning the reasons for these problems. On the other hand, with the arrival of the new Bing, we also found that services that have an Internet connection tend to be more accurate in their answers, but they still make mistakes.

Be that as it may, not even ten months have passed since the launch of this chatbot, The presence of this type of service can now be considered fully normalized in our lives, especially with their interfaces over the Internet and their applications, but also with the progressive integration of their functions into other products and services, something in which we will soon see a remarkable new evolution, with the general arrival of Windows 11 with Copilot, a step that will undoubtedly also see Apple and Google with iOS and Android.

As I said, the presence of chatbots has become normalized in a very short time, and whenever a service of this type ceases to be the “star of the moment”, we inevitably see their numbers dwindling. It is therefore necessary to monitor this decline in order to verify whether its “floor” is within the predictable (and therefore also expected by its creators) or whether we are facing a flower one day.

In this sense, it was reported by the Reuters agency ChatGPT has seen a drop in visits for the third month in a row. However, and according to the data that we can read in the said publication, there are indications that this decline would already be coming to an end, and that we may therefore already be quite close to knowing the real lower limit of the volume of traffic and users. the OpenAI service.

The basis for reaching this conclusion is that in the face of two rather sharp declines in June and July, each of 10%, in August there was also a decrease, but much more moderate, only 3.2%. So the data for September will be key, because if there is a lower decline, stagnation or even a slight rebound, we will be able to confirm this theory.

We understand that there are several reasons that have led to this decline in ChatGPT traffic. In the first place, of course, is what I already mentioned, the effect of the end of novelty, but in this case It seems no coincidence that the decline began shortly after Google released Bard worldwide.and we must add that users are increasingly aware that chatbot responses are not as reliable as one would like.