Nobody in their right mind will use genAI, LLMs in the next 5 years ...
During a session at the World Economic Forum in Davos, Yann Le Cunn, the chief AI scientist at Meta, has foreseen a significant shift in AI architectures. He believes that the current generative AI and large language models (LLMs) are limited in their capabilities and that in the upcoming years, they will become obsolete.
The Future of AI Architectures
Le Cunn predicts that the current paradigm of AI, which relies on generative AI and LLMs, has a short shelf life of about three to five years. He anticipates the emergence of a new paradigm for AI architectures that will overcome the limitations of existing systems.
The Decade of Robotics
Moreover, Le Cunn envisions the next few years as the "decade of robotics," where the fusion of robots and AI will lead to innovative applications of technology. He points out four reasons why the current AI models fall short: lack of awareness and understanding of the physical world, limited memory capacity, absence of reasoning ability, and incapability of performing complex planning tasks.
Advancements in AI
As the conversation about the deficiencies of current AI models continues, OpenAI and Perplexity recently unveiled new agentic AI that excel in handling complex, multi-step tasks. For example, OpenAI's AI agent called Operator is capable of carrying out tasks like ordering groceries, booking flight tickets, and creating memes by leveraging the web.
In conclusion, the landscape of AI is evolving rapidly, and the future holds promise for more advanced and capable AI architectures that will revolutionize the way we interact with technology.
