The AI race is about to enter a new phase. With the launch of the new o1 models, OpenAI marks a turning point in the evolution of AI systems. This is not just an incremental update: o1 represents a paradigm shift, moving from simple prediction based on training data to a system that attempts to emulate human reasoning. The era of GPT, in short, is already coming to an end: but before we talk about the future, a little refresher in the box below: do you know what GPT means?
GPT (Generative Pre-trained Transformer) is an artificial intelligence model that generates text by predicting which word might follow in a sequence, thanks to pre-training on huge amounts of text. As many “studied” have always said, it is a bit of a T9 with superpowers. It is the “brains” behind tools like ChatGPT. And now it's about to be supplanted.
New models under the magnifying glass
OpenAI's announcement has shaken up the technology landscape once again. In the words of Sam altman, CEO of the company, o1 represents “the smartest model in the world”. What is the truth behind these claims? First of all, the fact that the system (as we anticipated months ago) operates fundamentally differently from its predecessors, using an approach that is more reminiscent of a reasoning process than simple pattern replication.
The main innovation lies in the way the new o1 models process information. Unlike GPT models, which generate answers based primarily on statistical patterns, o1 actively evaluates several possibilities before providing an answer, just as a human would do when faced with a complex problem. The computing power required for this type of processing is impressive. The system requires computational resources and power consumption (already very high, speaking of AI) significantly higher than previous models, highlighting how ambitious this new direction is.
A turning point for AI
Mark Chen, former vice president of research at OpenAI, emphasizes the fact that the new o1 models have for the first time begun to develop a characteristic until now considered exclusively human. This evolution already shows profound implications in various sectors. For example, in programming, o1 excels at testing code to find bugs, thanks to its ability to systematically evaluate different possibilities and verify solutions. A significant step towards what OpenAI calls “superintelligence,” although on the timing to reach the finish line there is still disagreement.
Market impact and future prospects
The introduction of the new models comes at a strategic time. The generative AI market shows signs of slowing down, and this innovation could revitalize investors' interest, avert the risk of a "bubble" that would be bigger than the first "internet bubble" at the beginning of the millennium, opening up new development prospects. OpenAI has built an ambitious narrative around o1, which sees AI as a tool to address global challenges such as climate change and space exploration. The company's valuation, is reached 157 billion dollars, reflects these aspirations.
However, the real test will be the practical application of this technology. I emphasize the fact that the demand for enormous computational resources could limit its accessibility and immediate impact on the market.
New Models and Goodbye to GPT: The Transformation of the Technological Landscape
The GPT era may indeed be coming to an end, but that doesn’t necessarily mean we’re entering an age of superintelligence. The new models certainly represent significant advances, but the gap between AI capabilities and human intelligence remains considerable. The secrecy surrounding o1’s development, in my opinion, fuels both interest and skepticism. No one outside of OpenAI knows exactly how the technology works, which makes it difficult to objectively evaluate the company’s claims.
The future will tell us whether o1 truly represents the beginning of a new era of artificial intelligence or whether, as I believe, it is just another step, albeit a significant one, in a long evolutionary path.
