Editor-DN
The next generation of artificial intelligence systems will be able to perform tasks on their own without human input, and they are being enabled by this. OpenAI’s new o1-like modelsAccording to company CEO Sam Altman.
Speaking during a recent fireside chat. T-Mobile Capital Markets Day, Altman praised the merits of o1 models. Their ability to ‘reason’. They say this will open up entirely new opportunities with AI that were previously impossible with GPT-class models.
Altman says that these reasoning models are capable of working through a problem before proposing a solution. Level 3 AI developmentwhich he describes as an agentic system.
Where are the agent systems? Chat GPT Effectively you will be able to work on your own to get the best response, including going to work on other services. This will then lead to Level 4—systems that can innovate.
What is the big change with AI?
During the fireside, Altman acknowledged the merits of the existing GPT series of models. It includes GPT-4o which gives power. Chat GPT And high-end sound. It is natively multimodal but works like any previous AI—token-by-token.
He said: “The GPT series models were amazing in terms of ‘System-1’ type thinking, but what we wanted were systems that could reason. If AI can solve problems, has great value. o1 is the first system that can do complex reasoning, and if you give it a challenging problem, you can get extraordinary results.”
Here, System 1 refers to fast, intuitive, and automatic cognitive processes. System 2, where models are going with o1, refers to more deliberate, logical reasoning but doing so slowly.
In time, it will look as important as the GPT series. Think of it as the GPT-2 stage of these new reasoning models. You’ll see this creep up to GPT-4-level models in the coming years.
Sam Altman, CEO of OpenAI
Altman emphasized the importance of this development: “Over time, it will look as important as the GPT-series. Think of it as the GPT-2 stage of these new reasoning models. You see. In the coming years it will be up to GPT-4-level models.”
It’s still very early days. o1, with the current preview, is roughly equivalent in performance to what will come with GPT-2 in the coming years, a full model generation before ChatGPT launches in November 2022.
Although early, he expressed confidence in the rapid progress: “But even in the coming months, you will see upgrades as we go from o1-preview to o1. The improvement curve is very steep. , and things that models can’t solve today can be solved in a few months.”
Altman also highlighted the potential for new and innovative applications: “We’re going to see a whole new set of ways to use these models… we’re at the beginning with o1, new ways to use it. There will be ways, and it will take some time for us and the users to figure out how to use it.”
Why is o1 a big deal?
Open AI o1 is a completely new class of large language models. Previous generations and approaches, including the GPT family, respond to user prompts with tokens, often resulting in misleading or outright misinformation.
There are several techniques to solve this problem, including large contextual windows that allow the AI to access previously edited details and memory functions that do the same across multiple chats. However, these are both sticking plasters, and a paradigm shift was needed.
If AI can help us invent new things, cure diseases, come up with better sources of energy, it will be a huge win.
Sam Altman
With o1, OpenAI changed the approach, moving to a concept of chain thinking where, once you give the AI model a prompt, it goes away, and works on the problem step by step. does, the way a human being can act. One problem before presenting the answer. I’m sure we’ve all lost grades in school for not showing our work properly. Well, now the AI has to show its work too.
Speaking to T-Mobile, Altman cited healthcare and education as great areas where reasoning models like o1 could have a significant impact. “If you imagine that every student gets personalized tutoring, tailored to them along with other learning experiences,” that’s important. His big hope is that AI will help advance scientific discovery. “If AI can help us invent new things, cure diseases, come up with better energy sources, that would be a huge win.”
Altman concluded by reiterating OpenAI’s commitment to deep learning and his belief in the path to AGI and beyond while being open to adapting his approach based on continuous learning. He envisions a future where agentic experiences enabled by this technology will have a profound impact.
He says we should expect the first full version of o1, not a mini or preview, in the coming months, with o2 and next-gen versions in the coming years. What is not clear is whether the ‘o’ family of models will become agentic, or whether this will be another paradigm shift like the shift from GPT to o.
