Yann LeCun, the head of AI research at Meta, recently stated in an interview that current AI systems are not as powerful as some researchers claim. He pointed out that true understanding, planning, and reasoning capabilities in artificial intelligence are still lacking and that multiple "conceptual breakthroughs" are needed to achieve human-like intelligence.
Image source note: Image generated by AI, image licensed by Midjourney
LeCun criticized the overly optimistic views of OpenAI and Google DeepMind, stating that human-level AI is clearly not achievable within the next five years.
He mentioned that current AI relies heavily on text training, but this method provides very limited information. He proposed that the next generation of AI systems needs to have emotional capabilities to better set goals and understand consequences.
This is not only to simulate human intelligence but also to effectively address complex issues such as climate change and disease control. LeCun also predicts that in the future, everyone will have AI assistants, which will greatly simplify interactions in daily life and the digital world.
Additionally, LeCun supports the development of open AI systems, emphasizing that this helps promote competition, prevents monopolies by large tech companies, and allows more people to access AI technology. He also believes that building AI systems with "moral character" can control AI in a way similar to how laws regulate human behavior, thus achieving a safe and controllable development path.
Key points:
🧠 Current AI lacks understanding and reasoning abilities and requires multiple conceptual breakthroughs.
🌱 The next generation of AI needs emotional capabilities to set goals and understand consequences.
🌍 LeCun supports open AI systems to prevent monopolies and predicts that everyone will have AI assistants in the future.
