Elon Musk sets a date when artificial intelligence will surpass human intelligence

Elon Musk was one of the founders of OpenAI but, for various reasons, he abandoned the project, believing that its development was harmful to humanity. During an interview with Nicolai Tangen, executive director of Norges Bank Investment Management, the tycoon made a worrying prediction.

Musk’s vast knowledge on the subject – he created xAI a few months ago – has motivated a grim prophecy about the future of this advancement. According to the founder of SpaceX, the capacity of artificial intelligence models will surpass human intelligence end of 2025.

Furthermore, the tycoon predicted that an AGI or general artificial intelligence, It will not take long much to come. If generative AI already scares many, AGI multiplies that fear exponentially.

“I think we will have an AI that is smarter than any human probably by the end of next year. Artificial intelligence is the most advanced technology I have ever seen. Hardly a week goes by without some new announcement,” Musk said.

In turn, he highlighted the amount of hardware dedicated to artificial intelligence. “When you combine the available hardware with all the software advances and look at that curve, it seems crazy“.

Musk has already launched other prophecies that are still waiting to mature. For years he has been promising that Teslas could achieve fully autonomous driving and even man’s arrival on Mars. Although he has made progress on both goals, achieving them still seems far from his reach.

The challenge of a free AI

Elon Musk’s prediction coincides with that of Jensen Huang, head of NVIDIA, who believes that artificial intelligence will surpass human intelligence in five years.

The graphics card manufacturer played a decisive role in the development of the IAG thanks to its chips. At the same time, demand for dedicated hardware has caused a shortage that has slowed the momentum of OpenAI and other companies.

During the interview he took the opportunity to talk about the development of Grok, his artificial intelligence model.

The coach explained that the only limit at the moment is training the second version of Grok which will arrive next May. Musk went further and anticipated it will surpass GPT-4 once it is available.

“We’re trying to get enough GPUs to train it fast enough that we can show it in May. Which I think will probably happen. The next step would be Grok 3, which would be, I imagine, comparable to GPT-5 or better.”

It announced at the time that it would offer Grok to all X Premium subscribers. The idea of ​​opening it up to more people would be part of a strategy to increase subscriptions on its social network.

X AI is currently limited by country and is offered to those who pay for the most expensive plan (X Premium+).

Another point he talked about was AI safety, one of the key issues for the expansion of this system. One of the pillars of security is truth and he warns that if you do not program an artificial intelligence that is as truthful as possible, it will be even more complicated to manage it.