GitHub

IT House News of June 4, at OpenAI At a business event, the company CEO Sam Altman presented a three-stage theory for the development of the AI product.

In his view, after chat robots and AI smarts, the focus of the next phase of industry development would be “active AI”, an AI system that could continue to operate backstage and offer help to users.

Altmann divides the development of current AI products into three phases. The first stage is... ChatGPT A chat robot for delegates; the second stage is an AI smart system capable of carrying out specific tasks, such as Codex; the third stage is a “active AI” that further enhances automation.

I bet that's what's coming next. If there is anything to be prepared for in the coming year, that is the best way to go.

Altman refers to the current stage of the smart body as “the most interesting IA product category” to date, the development of which has been largely driven by the real needs of business clients. However, as the range of products continues to grow, users are beginning to face new confusion. Many people don't know when to use chat robots, when to use Cordex or APIIt is also difficult to integrate the context information required and various plugin tools.

To this end, OpenAI hopes to further strengthen the capacity of AI ' s intelligent bodies and build a more complete product system on this basis. Altman revealed that the company was developing an integrated product that would bring Codex's intelligence capabilities to Chat.GPT Together with other tools, a unified platform similar to the “super-apply” was formed to lay the foundation for “the realization of all the other things we want to do”.

At the same time, AI costs are becoming a focus of business attention. Altman stated that, at the beginning of the year, businesses would hardly have taken the initiative to discuss costs, but now they have become “a huge problem”, perhaps “the second most important topic of the day”.

He revealed that Uber had indicated that it had exhausted the AI budget for the whole year in the first quarter alone. In response, OpenAI hopes to create more value by improving model efficiency while reducing costs. "I think we'll have a lot of ways to help our clients gain value with lower input."

In addition to the cost, Altman mentioned another real challenge: “Most people” do not know how to use AI efficiently. He stated that many people had realized that they were under-utilizing AI and were not fully valued, but that there was a threshold for learning about new ways of working.

“It is not easy to learn new ways of working, and start-up costs are a little too high”, Altman said. Even though OpenAI can display many complex and powerful applications, most clients do not use AI as they do in practice.

Active AI is also OpenAI's response to this -- if users are not willing to learn how to use AI, the future AI system should be actively integrated into work learning, running backstage and helping users.

Ortman's idea is that users don't even need to understand what AI can do. "Can OpenAI, as an intelligent body, always run backstage, connect to all the contexts of my company? Don't let me try to understand what it can do, it'll help me."

He noted, however, that such a model would pose new challenges to the existing IT architecture of enterprises. Business needs to redesign AI deployment modalities, security strategies and mechanisms for the allocation of power resources. Unlike the traditional chat window, which responds to user requests on demand, an AI system that continues to operate and has access to a large number of enterprise context data places higher demands on data security, access control and infrastructure.

As more and more enterprises begin to try to integrate AI in depth into business processes, OpenAI is gradually expanding its product focus from simple dialogue to a smart body with a capacity for continuous implementation. According to Ortmann's plan, a “active AI” that can be operated on a long-term, proactive mission or will be the next stage after chat robots and AI Agent.

Read about:

Advertising statements: The external jump links (including not limited to hyperlinks, 2D codes, passwords, etc.) contained in the text are used to convey more information and save time for selection purposes only for reference purposes, which are included in all IT House articles.