Hacker News new | ask | show | jobs
by minimaxir 1800 days ago
The cynical-but-likely-accurate take is that researching language modeling has a higher ROI and lower risk than researching robotics.
8 comments

Yes.

Also, regular ML researchers sit at tables with laptops. Robotics people need electronics labs and electronics technicians, machine shops and machinists, test tracks and test track staff...

If you have to build stuff, and you're not in a place that builds stuff on a regular basis, it takes way too long to get stuff built.

> If you have to build stuff, and you're not in a place that builds stuff on a regular basis, it takes way too long to get stuff built.

I wonder why they don't invest in establishing the competency for robotics. The potential return might seem enormous, though their choices might signify that they don't agree.

Or maybe they just aren't willing to leave their comfort zone. 'Software will eat the world' is a convenient idea for people who want to stay in that comfort zone.

It's just higher risk and lower bang for buck right now.
My prediction is that dropping the real-world interactions will severely slow down their progress in other areas. But then again, I'm super biased because my current work is to make AI training easier by building specialized hardware.

Reinforcement learning can work quite well if you produce the hardware, so that your simulation model perfectly matches the real-world deployment system. On the other hand, training purely on virtual data has never really worked for us because the real world is always messier/dirtier than even your most realistic CGI simulations. And nobody wants an AI that cannot deal with everyday stuff like fog, water, shiny floors, rain, and dust.

In my opinion, most recent AI breakthroughs have come from restating the problem in a way that you can brute-force it with ever-increasing compute power and ever-larger data sets. "end to end trainable" is the magic keyword here. That means the keys to the future are in better data set creation. And the cheapest way to collect lots of data about how the world works is to send a robot and let it play, just like how kids learn.

I dont think this is cynical and I don't think it's a bad thing. OpenAI is not a huge org. The truth in 2021 is that not only is robotics 'just not there yet' in terms of being a useful vehicle for general intelligence research (obviously robotics research itself is still valuable), there is also nothing really pointing at this going to be the case in the next 5-10 years.

Given that, unless they want to commercialise fruit picking or warehouse robots, it seems sensible.

Sure. Yet consensus among "brain scientists" has long been that locomotion and the ability to explore the physical world is essential (or even central) to how consciousness develops and works in humans. Which in turn would seem pretty important for an institute working on cutting edge AI?

One of the reasons ML-based AI is pretty dumb still is possibly that this autonomous exploration side of AI is largely ignored.

It all seems to tie back into what Judea Pearl talks about in his "book of Why" (how you can't model intelligence without modelling learning of causal inference) or what Jeff Hawkins explores with his "reference frames of reference frames of the world" 1000 brains theory.

> Given that, unless they want to commercialise fruit picking or warehouse robots, it seems sensible.

How successful do you think attempts to monetize this will be? Apart from Kiva at Amazon, I'm not even sure most shelf-moving robots are profitable enterprises (GreyOrange, Berkshire Grey, etcetera). I'm very skeptical of more general purpose warehouse robots such as you see from Covariance, Fetch, etcetera. I don't really know too much about fruit-picking other than grokking how hard it would be and how little it would pay.

To be clear, I'm not saying these companies make no money or have no customers. But it's not clear to me that any of them are profitable or likely will be soon, and robots are very expensive. I'm happy to learn why I'm wrong and these companies/technologies are further ahead than I realize.

Yep. Hardware is hard and expensive. The real world is messy and complicated. Focusing on information wrangling currently has a much higher payoff.
My guess is they see they are close to something very big with language models and want to invest everything there
Wojciech stated this pretty explicitly on his Gradient Dissent podcast a few months back.
After a bit of Googling are you referring to Wojciech, the head of YouTube?
Presumably they are referring to the OpenAI co-founder Wojciech Zaremba mentioned in the first few sentences of the article.

It seems madisonmay didn't read the article either, or they would have known that the podcast they were referring to was the exact source used by the article.

thanks for the link internet stranger!
Order picking in e-commerce warehouses seems a potentially profitable market.
Definitely! Pieter Abbeel (who was working with OpenAI at some point) and others realized this and founded https://covariant.ai/.
I was wondering why OpenAI's gym was archived on GitHub this pivot seems more sense.
GYM is not exclusively for robotics - it’s for reinforcement learning in simulated environments, which I’m sure they will keep doing. Also it looks like it is still being maintained, so not really sure what you mean.
Can you explain what that means? I'm familiar with OpenAI Gym, I've been away from Github for a long time.
Read only