Elon Musk is set to unveil Grok 3 which he hails as the “smartest AI on earth”.

The latest iteration of xAI’s reasoning model will be debuted in a live demonstration on Monday (17 January) at 8pm Pacific Time which corresponds to 4am GMT on the following day.

Speaking at the World Government Summit in Dubai, Musk said of xAI’s latest development: “Grok 3 has very powerful reasoning capabilities, so in the tests that we’ve done thus far, Grok 3 is outperforming anything that’s been released, that we’re aware of, so that’s a good sign.”

Little is known so far about the specifics of Grok 3, however, admissions made by Musk on X suggest that the model will introduce a number of advanced features, such as text-to-video conversion, and significant efficiency improvements. 

Forbes further reports that it will boast synthetic datasets, self-correction mechanisms and reinforcement learning.

Grok 3 is aiming to take its position as a competitor to the plethora of other AI models released in recent times. This includes OpenAI’s GPT-4, Google DeepMind’s Gemini and DeepSeek’s R1 model.

The latter of which shocked the global AI industry last month with the claims DeepSeek made over its low-cost of development and performance compared to its vastly more expensive western counterparts. 

‘Scary smart’

In his address, Musk described the capabilities of Grok 3 as “kind of scary”, revealing that it has formed solutions to prompts that were beyond the anticipation of its developers. 

He added that it also has the capability to reflect on mistakes to improve the accuracy of training using synthetic data.

“It goes back and forth through the data and tries to achieve logical consistency,” explained Musk. 

“So if it’s got data that is wrong, it can actually reflect upon that and remove the data that is wrong and does not concord with reality. So it’s base reasoning is very good.”

Musk formed xAI in July 2023 shortly after he signed an open letter calling for a pause in the development of powerful AI models. More recently he warned that AI has “exhausted basically the cumulative sum of human knowledge”.

As a result he suggested that synthetic data- which is itself generated by AI models – will now need to be used for AI training and has now been used in the development of Grok 3. 

However, research has suggested that synthetic data can lead to model collapse and Musk himself admitted that it is challenging to understand if AI has “hallucinated an answer or if it’s a real answer”.

The direction of travel for AI development and regulation was also discussed at last weeks AI Action Summit.

The US government, which Elon Musk is connected to as the leader of the newly-created Department of Government Efficiency initiative, was notable in its refusal, alongside the UK, to sign an international agreement on AI promising to ensure the technology is “open, inclusive and transparent”.

According to Politico, the Donald Trump administration had expressed concerns over language calling for “inclusive and sustainable AI”. In his keynote speech at the conference, Vice President JD Vance warned delegates that overregulation of AI “could kill a transformative industry just as it’s taking off” and “pro-growth AI policies” should be prioritised over safety.

Elon Musk to unveil ‘smartest AI on earth’