Sam Altman anticipates novel insights in 2026

There is no doubt that artificial intelligence has advanced exponentially all this time. Engineering has perfected the creation of new AI-based processors, which we could say is a feedback. However, the costs could still be considerably high if we are aiming for a leap of dimensions that we would not be able to evaluate until this point. Although behind every idea, there is a clear intention to move towards the singularity or, at least, to begin to address this issue.

The singularity is about a hypothetical phase in which AI surpasses human capabilities, such that this event could drastically impact our society. If we reach that stage, it is likely that humans will begin to lose prominence, as well as the need for us to be required, as an AI could autonomously improve itself.

Today, artificial intelligence is used in almost all areas, but its impact is beginning to be seen most in companies that create new technologies. Today, the potential of AI agents is being used to help programmers build robust, fast and efficient applications in less time. This is possibly the reason for the exponential growth of several startups. Yes, it is true that there is a large pool of talent that makes this possible, but we must also consider the impact of AI in this new world that is being forged.

One thing we can confirm is that the step prior to the singularity itself could be based on intelligences capable of solving research questions, which would be reflected in new discoveries. An essay published by Sam Altman, CEO of OpenAI, called "The Gentle Singularity," addressed how AI will impact the human experience in the next 15 years.

This essay focuses on the approach towards a General Artificial Intelligence, although it has a certain tendency to exaggeration. This could be a positive thing when seen from a certain point of view, generating more enthusiasm towards those who want to be part of a historical event. It could also lead to greater investment, because we see AI as a clear bet on the future.

According to Altman, it is estimated that 2026 will be the year when this AI will have the capacity to discover new knowledge. This could be supported by considering current reasoning models as possible candidates for generating innovative ideas that will make a difference in the world. On the other hand, Greg Brockman, co-founder and president of OpenAI, argues that scientists have first used OpenAI's GPT o3 and GPT o4-mini AI reasoning models to generate new ideas that can be useful to the scientific field.

Apparently, and knowing that efforts are focused on developing an AI that is capable of building new knowledge, other companies besides OpenAI are with a clear focus on training AI models, in order to help scientists formulate new hypotheses, and subsequently, their findings.

Among the competitors we can find well-known companies such as Google, which has published an article regarding AlphaEvolve and its ability to tackle highly complex mathematical problems. Also Anthropic, which has promoted a program aimed at supporting scientific research. In addition, there are other companies, such as the startup FutureHouse, which envisions an AI agent with the ability to make real scientific discoveries.

The next few years, and I would be inclined to exaggerate to months, could be crucial for the first step in drug discovery, new materials, and how other industries will benefit from these advances. Success could define how automation of the scientific process would impact a discovery rate in considerably shorter times, as well as greater discovery potential.

However, the scientific community remains skeptical at the moment, considering that current AI models are not capable of generating genuine information. What is certain is that this will be the next step, and if Altman's words are reflected once again with the level of certainty that was had about AI agents, we may begin to question not only whether this will be possible, but how long it will take to make the first major discovery.