The next model designed by OpenAI to succeed GPT-4 would offer an incremental improvement. Far from the announced revolution.
Since the arrival of ChatGPT in 2022, generative AI algorithms are advancing very quickly. You only need to see the developments of a system like Midjourney for image generation to see this.
Will this development slow down in the future? This is what we believe we understand The Information about the next developments of OpenAI.
La suite de GPT-4
In recent weeks, Sam Altman's company has made big promises around the next evolution of GPT-4, the model that drives ChatGPT.
This new version of the model, known under the code name Orion, would only offer an incremental improvement compared to GPT-4. In other words, the experience offered by ChatGPT would not be disrupted.
Furthermore, OpenAI would face the same problem as for GPT-4: the heaviness of the model and the computing power required to run it. Generative AI and in particular GPT-4 require very large resources, which implies a cost in both infrastructure and electricity for the tech giants who equip themselves with it, such as Microsoft.
Our Twitch show SURVOLTÉSit's every other Wednesday from 5 p.m. to 7 p.m.: meet us to talk about electric cars or electric bikes, around debates, interviews, analyzes and games!