Sam Altman, CEO of OpenAI: "Regulation for small models makes no sense, it should be for large ones"

   MADRID, 22 May. (Portaltic/EP) -

The CEO of OpenAI, Sam Altman, believes that regulation for small language models powered by Artificial Intelligence (AI) "does not make sense" and is committed to regulating large models such as, in this case, OpenAI's ChatGPT, because they are "the ones who can really do harm".

This was expressed by the executive director of the company that creates tools such as Whisper and Dall-E2, in a colloquium held this Monday at IE University, where he was accompanied by OpenAI research scientist Mo Bavarian; Owners Scaleup Program Executive Director Joe Haslam; and the co-founder and CEO of Clibrain, Elena González-Blanco.

During the presentation, the CEO of OpenAI has remarked that regulation is "very important", but has also emphasized that over-regulating "is not good" since, in the case of small language model companies, it would cut their capacity of innovation and creation.

"We don't think regulation for small models makes sense," Altman observed, while noting that regulation has to be designed for large models, and provide that we really "can't do any harm in the world." Thus, he has suggested that small models should be "allowed to grow" and that the logical thing to do is to focus regulatory attention on large models, since "we are the ones who can manage them."

Likewise, the CEO of OpenAI has stressed the importance of privacy and the concern that exists around this issue at all levels. In this sense, he has assured that the company will continue "working with other governments around the world" to improve the language models driven by AI and protect privacy.

In addition, he has pointed out that from OpenAI they want to carry out technology with which "people are comfortable and safe". "We want to do what people want. We want to create the product that people want to use," said Altman, referring to the work they have been carrying out with ChatGPT to improve aspects such as privacy, data security and security. Quality of the information.

"It is fair and reasonable to be skeptical," he said, while admitting that technology's work with AI "has not been perfect." In this sense, he has pointed out that although there have been "many errors", all the work carried out to solve them and create improvements must also be taken into account.

In the same way, Altman has explained that they have pushed these technologies, in this case ChatGPT, to be "open to the world", even knowing that they are not perfect because "people have to try them". In other words, as he has developed, he considers it better that people, institutions and governments test a technology of these characteristics even if it is not yet completely ready, to see "how it can be used in a good and bad way and, based on this, adapt it to society.

For all these reasons, he has reiterated the importance of regulating AI-powered language models such as ChatGPT, but he has also emphasized that what concerns the company is, above all, "doing the right thing" and, therefore, has put into value that they have been "making ChatGPT more secure" for eight months.

Asked by the European Union (EU) in the fields of technology and AI, Sam Altman has indicated that the talent and quality of technology development in Europe is "brilliant" and has been very important for this field. In fact, he has stressed that OpenAI "would not have happened without Europeans" and that a large part of his staff is from different EU countries.

Following this line, he has added that it is a space where "all the ingredients are ready to be a place of innovation" and that you only need to have "the commitment to do it". However, Altman has pointed out that, from his point of view, he would make "some legal changes", and that Europe should not "hold back", but rather "lead" with its technology.

As you explained, the success of ChatGPT and the OpenAI language models is based on three characteristics. First of all, it is an algorithm "that can really learn", that is, it can make relationships between a large amount of data and that "was not obvious to people".

Also, the algorithm is predictable; "We can predict how good the results are going to be," Altman explained. Finally, he has detailed that with "very little data", it is an algorithm capable of behaving like "any system of values". For all this, he has indicated that "he has all the ingredients" to go far.

Specifically, Altman and Bavarian have explained that they use ChatGPT personally "very often" and that, above all, within OpenAI they use it to code and connect all the technology within the company. In this sense, they have remarked that, above all, the intention of ChatGPT is to become a super AI assistant that can "help with everything". Thus, they have indicated that the idea is that it can help to carry out small tasks and that they are increasingly complex tasks.

On the other hand, the executive director has also pointed out that another of the objectives of technologies such as ChatGPT is to balance society. With this, he has made reference to the possibility of helping anyone to have access to information or tools to be able to develop their initiatives, "regardless of who they are".

Thus, the CEO of OpenAI has also referred to education and how tools such as ChatGPT interfere. In this regard, he has pointed out that it can mean a "super tool" for both teachers and students and that, therefore, everyone should "embrace" this change in the model of education that is to come. "Educational institutions need to be in on this," Altman has said.

In addition to all this, he concluded by recalling that AI-driven technology, such as OpenAI, is the "most empowering technology that human beings have seen" and that, thanks to it and the creativity and innovation capabilities it presents, " The world is going to be tremendously different and better." "If you give people better tools, you will have a better future", has sentenced the CEO of OpenAI.