Why your robot butler isn't here yet | Ken Goldberg

UC Berkeley robotics professor Ken Goldberg explains how far away we are from a Jetsons-like future with robot helpers in every aspect of society, and whether the current buzz around robotics is realistic or just wishful thinking.

Ken and host Luisa Rodriguez talk about:
• Why training robots is harder than training large language models like ChatGPT.
• The biggest engineering challenges that still remain before robots can be widely useful in the real world.
• The sectors where Ken thinks robots will be most useful in the coming decades — like homecare, agriculture, and medicine.
• Whether we should be worried about robot labour affecting human employment.
• Recent breakthroughs in robotics, and what cutting-edge robots can do today.
• Ken’s work as an artist, where he explores the complex relationship between humans and technology.

Chapters:
• Cold open (00:00:00)
• Luisa's intro (00:01:19)
• General purpose robots and the “robotics bubble” (00:03:11)
• How training robots is different than training large language models (00:14:01)
• What can robots do today? (00:34:35)
• Challenges for progress: fault tolerance, multidimensionality, and perception (00:41:00)
• Recent breakthroughs in robotics (00:52:32)
• Barriers to making better robots: hardware, software, and physics (01:03:13)
• Future robots in home care, logistics, food production, and medicine (01:16:35)
• How might robot labour affect the job market? (01:44:27)
• Robotics and art (01:51:28)

----

_The 80,000 Hours Podcast_ features unusually in-depth conversations about the world’s most pressing problems and what you can do to solve them.

[Xaphedo]

The marketer vs the engineer. In 2 years vs in 2 to 50 years. The classic dichotomy of high uncertainty predictions. Thank you Ken for sharing your expertise and time!

[idrlracing]

Models are able to analytically think now not just do permutations of 20000 common words. Since they self reason and imagine its not hard for them to come up with most commonly used motions by a human being no matter how many exponential degrees of freedom those imply. What matters is robots being able to do a certain number of generalized tasks well not that each robot butler also needs to be a trained guitarist or gardener.

[Geosynchronus]

Sounds like a major step forward towards full immersion virtual reality with haptic and other modalities..I want to be a bunny 😊

[Yaddlezap]

the ability to gather and learn from the necessary real & synthetic data is already here, engineering has already had decades of progress ready to line up with it, early-adopter robot assistants might take 3 or 4 years but they'll be here before you know it

[michelleelsom]

Maybe the answer is to implant a chip into a human body so that the AI can learn directly from the movements of the human as they go about daily tasks, surely this is the only way for it to learn such complex movements

[vrushankdesai]

How is language 1-dimensional? You have a choice between ~50k tokens at each step. Writing a paragraph involves choosing from a set of 50, 000^(num_words) unique trajectories

[polymathpark]

What you are discussing on the many commands to make the hand move, the consideration of ligaments, joints, etc. that seem to be so simple for us, John vervaeke calls this "relevance realization" in humans, our brains learn how to ignore massive amounts of information and hone in on the necessary, or "relevant" information and behaviors.
Yall should have him on!

[alejandrorp]

Moravec's paradox. That's why

[TheExodusLost]

Bottom line, companies are going to automate every task they reliably can, and that number will only increase over time. It doesn’t take 100% of jobs able to be automated to create unemployment and pressures like we have never seen. Free market + acceleration + slow regulation, I wouldn’t bet against it.

[rolandgibson-murphy]

What he is saying is that the hype is preceding capability, and the payoffs will not be in the short term.... Short-term gains are exactly what people in the U.s. want and expect.
It's unavoidable. The money could be better places for short-term monetary gains.

Robots are coming, just not soon enough to pay off dividends.
People want robots to work for the greater good. A home robot to help with everything, while the corporations want to replace you.

[CandidDate]

Because we are robots the only kind that work.

[stormveil]

Opening spiel immediately sets off red flags. Language isnt 1 dimensional..
I want 10 apples.
Ken needs 1500 gpus.
Sakura requires a hug.

There are 4 dimensions there in just a simple sentence in 1 language.