xAI promised a revolution with Grok 3 before the end of 2024, but its absence on the social network X raises questions about the future of next-generation AI models.
Tl;dr
- xAI’s Grok 3, announced for the end of 2024, has still not been launched, reflecting delays in artificial intelligence projects.
- These reports illustrate the limits of current approaches, where efficiency gains through increased resources are reduced.
- With a small team and growing technical challenges, xAI embodies the obstacles faced by the entire AI industry.
An ambitious promise not kept
Last July, Elon Musk, founder of xAI, announced that Grok 3, his company’s flagship model, would be available by the end of 2024. This model promised significant advances, in particular thanks to training on an impressive GPU infrastructure. However, to date, no official announcement confirms its imminent deployment. This is not the first time that Musk has exceeded his initial deadlines, his projects often being marked by overwhelming but unrealistic optimism. Despite everything, expectations around Grok 3 remain high, particularly on its ability to compete with GPT-4o from OpenAI or Gemini from Google.
A general trend in the industry
The Grok 3 delay is part of a larger trend. In 2024, Anthropic failed to launch the expected successor to its Claude 3 Opus model, even canceling all mention of the latter in its documents. Google and OpenAI have also reportedly encountered similar obstacles in their respective developments. These shifts seem to indicate that the laws of scale in AI, based on a massive increase in resources and data, are reaching their limits. Progress from one generation of models to the next is becoming increasingly difficult to achieve, forcing companies to explore other techniques.
Internal factors specific to xAI
Compared to giants like Google or OpenAI, xAI operates with a much smaller team. This lack of human resources could partly explain the delays. Additionally, the potential launch of a mid-tier model, Grok 2.5, spotted on xAI’s site, could mean the company is taking a phased approach. Despite everything, this situation raises questions about xAI’s ability to realize its ambitions in the face of better-equipped competitors.
The technical challenges of the new generation of AI
Grok 3’s delay also reflects broader technical challenges. Current model training methods, relying on massive resources, seem to be reaching a ceiling. Elon Musk himself acknowledged this difficulty, admitting that the goal of making Grok 3 a cutting-edge model may not be achieved. This observation fuels debates about the need for radical technological innovations to overcome current limitations and maintain momentum in the development of AI.
Belgium
