Humanoid Robot Hype: Why Yann LeCun Says the Bubble’s Ready to Pop

The booming humanoid robot race

One can't escape the hype of the humanoid robots all around the world. From start-ups that promise robots to clean our homes to giants that are after the automation of everything, the momentum seems to be dizzying. But Yann LeCun, chief AI scientist at Meta is pointing out that we might be facing a robotics bubble.

Hardware over brainpower?

LeCun's criticism is: a lot of robotics firms are investing in the production of the attractive physical appearance of the robot, while at the same time they are neglecting the hard task - to equip robots with general intelligence. In his opinion, it is not enough for a robot to be able to walk or do the lifting; it should be able to comprehend the world.

What really matters: world-models and sensory learning

In his opinion, the main thing the world model represents in the case of AI is an architecture that gets to know the environment from diverse sensory sources (for example, video, interaction, etc.) and is able to predict how the world will react. Only in this way can robots be truly "useful" outside of narrow tasks.

Why the bubble warning matters

The question is: quite a few billions have been invested in the ventures of humanoid robots. LeCun, however, thinks the technology is still in its infancy. Unless there are big leaps in AI, many of these investments will end up as a pile of ruins. The hype may go far beyond the reality.

What this means for businesses and investors

● In case you put your money into the development of robots with humanlike features: find out whether the company is setting to work on the problem of comprehension of the world, or they are only dealing with the mechanics.

● If your company is into AI or robotics services: the gap in software that LeCun talks about represents a potential area for your company world-model, self-supervised learning, sensory input.

● For the industry as a whole: slower progress, more modest deliverables, and possibly a market correction if the results are not up to the mark.

The takeaway

By the way of which LeCun sees it, the rise of humanoid robots is super exciting, but we are in the "hardware first, brain later" period. Without a significant leap in AI, only a handful of the current robot boom will be realized and the rest will turn into an expensive lesson. So to say: the bubble is huge but the burst is quite ​‍​‌‍​‍‌​‍​‌‍​‍‌possible.