Yann LeCun, Meta’s chief scientist, made a groundbreaking statement at Meta’s launch event in Paris, declaring that generative AI, the technology behind ChatGPT, had reached a dead end. Instead, he promised to develop new artificial intelligence models that exhibit human-like rationality.
LeCun expressed frustration with the current state of AI and machine learning, emphasizing that humans have common sense, but machines do not. He introduced his Meta’s latest AI project called Image-Based Joint Embedding Prediction Architecture (JEPA). This ambitious effort aims to go beyond the capabilities of generative AI and enable machines to conceptualize abstract ideas rather than simply spewing existing online information.
Generative models are now a thing of the past, according to LeCun, as Meta plans to shift its focus to collaborative embedding prediction architectures. He said confidently: “My prediction is that in a few years generative large-scale language models will no longer be in use. He has since been critical of the hype surrounding generative AI models such as ChatGPT and image-based Dall-E.
LeCun believes the fear and excitement surrounding generative AI creates exaggerated perceptions of its capabilities. Meta CEO Mark Zuckerberg announced on Facebook that the JEPA tool will be open source for researchers to experiment with. The goal is to develop AI that closely matches human understanding of the world. “We need models to perceive and make predictions about the world,” Zuckerberg said. “This study is another step in that direction.”
In contrast to its competitors, Meta has taken a more cautious approach to ChatGPT-style AI for social media platforms Facebook and Instagram. Meta incorporates generative AI into their products, but it’s not as well-hyped as companies like Microsoft and Google. At the same time, Meta released an open-source AI model that requires less computing power than the technology behind ChatGPT.
LeCun’s declaration marks a major shift in the AI landscape, with Meta spearheading the development of new AI models that seek to mimic human rationality. The JEPA project represents a revolutionary step towards pushing machine intelligence beyond the limits of generative AI.
