A Godfather Of AI Has Called Out Elon Musk’s Bulls**t

🌈 Abstract

The article discusses Elon Musk's claims about the progress of artificial intelligence (AI) and the skepticism expressed by AI expert Yann LeCun regarding Musk's predictions. It also explores the challenges facing the development of advanced AI, such as the energy and data requirements.

🙋 Q&A

[01] Musk's AI Claims vs. Reality

1. What are some of Musk's past claims about AI that did not age well?

• In 2016, Musk claimed that Tesla's vehicles could drive autonomously with greater safety than a human.
• In 2019, Musk claimed Tesla would have a million robotaxis by the end of 2020.
• Recently, Musk claimed that AI will be smarter than any human by next year and smarter than all humans combined by 2029.

2. How did Yann LeCun, one of the "godfathers of AI," respond to Musk's recent claims? LeCun pointed out that if AI were truly that advanced, it should be able to teach itself to drive a car in 20 hours, like a 17-year-old human, but current AI systems still struggle with fully autonomous and reliable self-driving.

3. What is LeCun's view on the current capabilities of AI systems? LeCun stated that current AI systems have about as much computing power as a common housecat's brain but are much less clever, as they still cannot understand the physical world, plan complex actions, or have a level of reasoning comparable to humans.

[02] Challenges Facing Advanced AI Development

1. What are the key challenges in developing Artificial General Intelligence (AGI) according to experts?

• Energy requirements: Simulating a whole human brain using AI neural networks would consume orders of magnitude more power than the entire US currently produces.
• Data requirements: Making AI more advanced requires training on ever-larger datasets, which can be unethical and expensive to obtain legally.

2. What is the overall scientific consensus on when high-level machine intelligence (as smart as a human) might be achieved? According to surveys, the majority of AI researchers believe that superhuman AI will either never happen or happen in the next century. The overall consensus is that high-level machine intelligence has a 50% chance of happening before 2059, but there are many uncertainties that could delay progress.

3. Why do experts think AGI might never happen? The key reasons are the immense energy and data requirements needed to develop AGI, which may be insurmountable with current technology and ethical constraints.