A recent TechCrunch article reminded startup founders that merely incorporating ChatGPT into their products does not make them revolutionary, even in the eyes of investors caught up in the latest trend. This is a crucial alert for large companies as well.
Having previously written in these columns about the innovative nature of the platform OpenAI introduced a year ago, my current views might seem contradictory (although my opinion hasn't changed since January). It's vital to differentiate between the tool made for the general public, sparking various original use cases, and the generative AI that startups and multinationals now seek to utilize.
To understand my perspective, it's essential to recognize what this widely discussed, fascinating entity is: an algorithmic system based on statistical mechanisms, capable of generating rational content (texts, sounds, images, videos) from a large data pool that feeds it and whose characteristics it replicates. Therefore, it lacks real creativity, with innovation arising only from its implementation context.
Even without focusing on its largely experimental stage in the professional environment, it's evident that implementations mainly involve rather basic applications. Interactive information search and decryption, or virtual assistants (sorry, co-pilots) holding sensible conversations, to mention the most common examples, are indeed useful advancements but only bring incremental improvements for users.
Those more daring, wishing to leverage ChatGPT or its equivalents to develop a truly distinctive solution, must acknowledge and accept that this component will only contribute to the tip of the iceberg (and even then, it must be correctly "fed," but that's another debate). The innovation's core engine will rely on other technologies, especially machine learning and other AI approaches, while staying within the field.
In other words, and revisiting a theme dear to me, generative AI is not a magic wand that instantly turns an existing service into something extraordinary. It's just an additional tool within a vast toolbox, primarily enhancing communication between a human and data sources. It's up to each individual to envision the circumstances in which it will have the most impact…