The future of artificial intelligence is on the table at OpenAI, the company behind ChatGPT. Its CEO, Sam Altman, has presented his team with three possible paths to safely manage the advancement of this technology, a decision he considers urgent and which could be defined within the next year.
The debate comes at a time when several voices in the sector agree that the room for maneuver to establish safeguards is shrinking. How this dilemma is resolved will directly affect how millions of people use AI in their work and daily lives.
A global supervisory body, in the style of atomic energy
The option for which Altman shows the greatest preference is the creation of an international supervisory body. This proposal, which OpenAI has advocated for since 2023, would be inspired by the model of the International Atomic Energy Agency, the entity that monitors the peaceful use of nuclear energy worldwide.
Its function would be to coordinate the main artificial intelligence laboratories, including OpenAI and Google DeepMind, to ensure responsible development and minimize risks. The logic is similar to the cooperation between countries to prevent the proliferation of nuclear weapons: having technology companies work together instead of competing without control.
The merger of laboratories: the so-called Demis Plan
The second alternative proposed by Altman is known as the Demis Plan, in reference to Demis Hassabis, co-founder of DeepMind, Google's artificial intelligence division. This idea consists of uniting the most relevant AI laboratories into a single large entity.
As reported by the outlet The Decoder, this organization would function similarly to CERN, the European particle physics laboratory where researchers from different countries collaborate on large-scale projects. The goal would be to concentrate talent and resources to advance in a more coordinated manner, working alongside allied countries.
The scenario that Altman calls terrible
The third path is the one that generates the most concern. It is a scenario in which a single laboratory manages to develop an artificial intelligence capable of continuously improving itself, thus achieving almost absolute dominance over the rest of the competition.
As reported by outlets such as Cryptobriefing and posts on the social network X, Altman warned that a company with that level of technological advantage could become so powerful that it would render the processing capabilities of any rival obsolete.
The Decoder noted that Altman dubbed this scenario the Eliezer Plan, alluding to Eliezer Yudkowsky, a recognized theorist in the field of artificial intelligence safety. It is an extreme possibility, raised to complete the range of options for internal debate.
Other voices in the sector warn about the time available
This discussion is not happening in isolation. Jack Clark, a representative of Anthropic, another of the sector's major companies, has warned that the world has only a few years to solve the security challenges posed by artificial intelligence.
The concern is not new. In July, Dario Amodei, CEO of Anthropic, published an essay in which he argued for the need to establish coordinated security practices across the industry. This reflects a growing consensus on the importance of creating protocols and independent evaluations for AI systems.
What this debate means for users
The decisions made regarding the control of artificial intelligence are not exclusively a technical or business matter. If the idea of a supervisory body or the merger of laboratories succeeds, it is likely that future AI systems will be more transparent and reduce risks such as bias or misuse.
Conversely, if a single company comes to dominate the development of artificial intelligence, users could find themselves with fewer options available and a greater dependence on that company's rules and services. The direction this discussion takes will determine whether AI advances as a shared and regulated tool or as power concentrated in very few hands.
Frequently Asked Questions
Who is Sam Altman?
Sam Altman is the CEO of OpenAI, the company that develops ChatGPT and one of the most influential laboratories in the development of artificial intelligence worldwide.
What are the three paths that Altman proposes for the future of AI?
The three scenarios are: the creation of an international supervisory body similar to the International Atomic Energy Agency, the merger of the main AI laboratories into a single entity known as the Demis Plan, and an extreme scenario in which a single company dominates the sector thanks to an AI capable of improving itself.
What is the Demis Plan?
It is a proposal inspired by Demis Hassabis, co-founder of DeepMind, which proposes uniting the large artificial intelligence laboratories into a single organization that would function similarly to CERN, collaborating with allied countries.
Why does Altman consider the third scenario terrible?
Because it would imply that a single laboratory achieves such a great technological advantage that it would leave the rest of the companies in the sector unable to compete, concentrating disproportionate power in a single company.
What other companies have warned about AI safety?
Anthropic, through its policy head Jack Clark and its CEO Dario Amodei, has insisted on the need to establish coordinated security measures across the industry before it is too late.
When is a decision expected to be made on these scenarios?
According to what Altman conveyed to his team, the decision on the path to follow should be defined within approximately one year.
