In a remarkable series of events, major players in the artificial intelligence (AI) arena, including OpenAI, Google, and the French startup Mistral, have all unveiled new iterations of their cutting-edge AI models within a span of 12 hours. This synchronized release signals an impending surge of activity in the AI industry expected throughout the summer.
The wave of releases commenced shortly after Nick Clegg’s appearance at an event in London, during which he confirmed the imminent launch of Meta’s latest AI model, Llama 3, slated for release within weeks.
Following Clegg’s announcement, Google swiftly rolled out Gemini Pro 1.5, its most advanced large language model, offering a free tier with limited daily requests.
Not long after, OpenAI introduced its own frontier model, GPT-4 Turbo, marking a significant milestone in the evolution of AI technology. Both GPT-4 Turbo and Gemini Pro 1.5 boast multimodal capabilities, capable of processing not only text but also images. Additionally, Gemini can handle audio and video inputs.
In the early hours of the morning, Mistral, a French AI startup founded by former Meta AI team members, unveiled its frontier model, Mixtral 8x22B. Unlike its American counterparts, Mistral opted for a simple download link to a 281GB file, adhering to an open-source approach that allows free access for developers to build upon.
While this open approach has drawn criticism for potential risks associated with unchecked usage, proponents argue it fosters innovation and collaboration, ultimately leading to better outcomes.
Meta’s release of Llama 3 is anticipated to roll out gradually, with smaller versions preceding the launch of its most advanced model later this summer. However, it faces formidable competition from OpenAI, which is reportedly gearing up to unveil GPT-5 within a similar timeframe.
Nevertheless, experts question whether the prevailing “large language model” paradigm might be approaching its limits. Meta’s chief AI scientist, Yann LeCun, cautioned against overestimating the capabilities of current AI systems, emphasizing their limitations in understanding the world beyond language manipulation. Despite advancements, the quest for artificial general intelligence remains elusive, underscoring the complexities and challenges inherent in AI development.