AI Models Have Yet to Reach Human-Like Reasoning, Meta AI Chief Says
At the recent Hudson Forum, chief AI scientist at Meta, Yann LeCun, addressed the state of AI and how today’s models do have not yet reached a stage where they can be deemed as human-like AI in reasoning, thinking, or even memory.
LeCun highlighted that while it is true that nowadays, AI can perform specific tasks, the models have yet to reach a level where the model is an AI that acts like human beings with the same complexity that we have in terms of thoughts and behaviors.
Can AI Act Like a Human?
Even though current systems do not have the ability to develop AI that works like a human brain, things are promising, expressed Le Cun, and machines with human-like understanding and planning might take several years, if not decades to bring to life.
This is in very strong contrast to the opinions of people like Elon Musk and Shane Legg, who believe that human-like AI might be just around the corner.
Meta’s AI chief said that “we need machines that understand the world; that can remember things, that have intuition, have common sense, things that can reason and plan to the same level as humans”, sharing with no fear his vision into developing AI human-like robots.
Current AI systems, including the large language models (LLMs) like OpenAI’s ChatGPT, are still fundamentally so limited in most instances that they perform their work as mere predictions of the next token or pixel rather than as thinking beings with an understanding of a three-dimensional world around them.
LeCun further pointed out that while humans pick these tasks up in a snap, the most advanced AI systems require enormous amounts of data and do not achieve comparable performance in physical environments.
To get around these limitations, LeCun suggests the creation of what he calls “world models” which can comprehend and interact with the physical world.
“A world model is your mental model of how the world behaves,” said LeCun and by that he believes these models will let AI robots look like human beings.
As the trend is toward greater integrations of human-like AI models, this attracts the interest and investment of major cloud providers and AI startups alike. Recently, a team of top AI researchers raised $230 million for their startup, World Labs, which is focused on developing AI that sounds like human type of models.
World Labs, co-founded by Fei-Fei Li, managed to raise $230 million in funding for the development of its advanced Large World Models (LWMs) that will mimic human-like reasoning by allowing machines to perceive, generate, and interact with 3D environment, in a similar manner to human cognitive behavior.
Final Thoughts
It’s not a matter of doubt whether the emergence of intelligent models such as GPT has made the world more efficient, but the real issue lies in the balance between ethicality and innovation.
In the past two years, many Big Tech companies have shown severe reluctance to acknowledge the possibility of detrimental consequences for humanity with the development of an AI that looks like a human being and even thinks like one.
No doubt that the world is easier with the presence of AI; however, the problem lies within the fine line between ethicality and technological innovation. Unfortunately, big tech leaders are refusing to acknowledge the consequences of developing AI that looks like a human and how far it will affect human beings, because they don’t see it unethical, but more like technological achievements.
The industry’s tech leaders are more often failing to see their creations as ethical monstrosities but rather perceive said creations as mere technological achievements. An AI robot like human beings will jeopardize the very notion of human identity, privacy, originality, possibly leading to the worldwide misuse of this intelligent technology.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Tech sections to stay informed and up-to-date with our daily articles.