OpenAI and Figure join the race for humanoid robots

The shocking emergence of ChatGPT has made this clear: AI is advancing at a breakneck and accelerating pace, and it is beginning to transform desk-based industries that typically harness human intelligence. They will begin to take over portions of many white-collar jobs in the years to come, leading initially to huge increases in productivity, and ultimately, many believe, a huge increase in unemployment.

If you’re out of school right now and looking to be helpful, a blue-collar job that involves real physical labor might be a better bet than anything that puts you behind a desk.

But on the other hand, it’s starting to look like a general-purpose humanoid robotic worker might be closer than anyone thinks, imbued with light-speed, swarm-based learning abilities to go along. with GPT-version-X communication capabilities, a whole lot of internet knowledge, and all the physical characteristics you need for a given job.

Such humanoids will start out as dumb construction apprentices with no common sense, but they will learn – at a frightening rate, if the last few months in AI have been any indication. They will be available 24/7, power sources permitting, gradually expanding their abilities and overcoming their limitations until they begin to displace humans. They could potentially drive down the cost of labor, leading to huge productivity gains – and a fundamental disruption of the blue-collar labor market at a size and scale limited primarily by manufacturing, materials and job types. that they are able to take on.

The best-known humanoid to date is Atlas, by Boston Dynamics. Atlas has made impressive progress over the past 10 years. But Atlas is a research platform, and Boston Dynamics founder Marc Raibert made it clear that the company is in no rush to bring humanoids to the mass market.

Atlas takes control | Boston Dynamics

“We need to remove the pressure to make things more reliable, more manufacturable and cheaper in the short term,” he said. said IEE Spectrum last year. “These are things that are important, but they get in the way of trying new things. The pitch I made for [parent company] Hyundai says this explicitly and offers financing long enough that we are not distracted in the short term.”

Hanson Robotics and Engineered Arts, among others, have focused on human/robot interaction, focusing on making humanoids like Sophia And Ameca, with realistic and expressive faces. In particular, Ameca communicates using the GPT language model.

Others are absolutely focused on using these things en masse as quickly as possible. Elon Musk announced Tesla’s entry into humanoid robotics in 2021and the company seems to have made pretty decent progress on the Optimus prototypes for his Tesla Bot in the months that followed.

“That’s probably the least understood or least appreciated part of what we do at Tesla,” he told investors last month. “But it will probably be worth a lot more than the car side of things in the long run.”

Tesla’s Optimus robot now walks and uses its humanoid hands for basic tasks

You’re here

Again this year, entrepreneur Brett Adcock, founder of Vettery’s “online talent marketplace”, and more recently Archer Aviationone of the main contenders emerging electric VTOL aircraft movement, announced that his latest venture is focused on humanoid robots.

The first phase of a Musk-like “master plan” for the new company, Figurefirst aims to “build a complete electro-mechanical humanoid”, then “perform human-like manipulation”, then “integrate humanoids into the workforce”.

The figure aims to create “the world’s first commercially viable general-purpose humanoid robot” – at this stage referred to as Figure 01. This guy will be 5ft 6in (168cm) tall and weigh 132lbs (60kg) according to the company. It will be able to lift a targeted “payload” of 44 lbs (20 kg), walk at speeds of up to 1.2 m/s (2.7 mph/4.3 km/h) and operate up to five hours on one charge.

As a two-legged humanoid, he’ll be able to climb stairs, walk around, and generally operate in most environments humans can – at least, eventually. With human-like hands and body mechanics, he should be able to use the tools we use and do a variety of our tasks.

Presentation of the figure

Adcock approaches this project with speed and aggression. When Archer threw, he hired a bunch of talent away from other big eVTOL companies and made very fast progress both technically and fundraising. Indeed, according to the AAM Reality Index, Archer is now the third best-funded eVTOL company in the world and is expected to have aircraft in service by 2025, around the same time as Joby Aviation. Joby was founded in 2009 and Archer came out stealth in 2020.

Figure seems to take the same approach; according to a interview with IEEE Spectrum, the company had more than 40 engineers on board in early March, with key talent from the Florida Institute for Human and Machine Cognition, Boston Dynamics, Tesla, Waymo and Google X, among others. So while the company doesn’t seem to have built anything yet, it’s a serious business that is going full speed ahead.

“We face high risk and extremely low chance of success,” reads the Figure website. “However, if we are successful, we have the potential to positively impact humanity and build the greatest company on the planet.”

Meanwhile, OpenAI has signaled that it is once again interested in humanoids. OpenAI wants to create an “artificial general intelligence” or AGI – a machine that can do almost any task better than a human. Thanks to the GPT language model, the company expects to eliminate a lot of this list in the short term, but it will have to put its smarts into a body to handle the rest.

OpenAI had its own robotics division for many years and indeed built a humanoid hand capable of fine manipulation and sensing that used neural networks and reinforcement learning to figure out how to solve a Rubik’s cube, with just one hand.

Solve the Rubik’s Cube with a robot hand

But the company will close its robotics team in 2021, telling VentureBeat that he was focusing his efforts on where he was making the fastest progress: generative AI and big language models like GPT. The main difference, according to co-founder Wojciech Zaremba, is data and computing power. The AI-powered robots have been shown to be able to learn physical tasks “extremely well” – there’s just a lot more text-based training data than video-relevant data for the robots, and text is much faster to learn. process for a neural model.

It is therefore on the text that they have concentrated their efforts, and the results are difficult to dispute; OpenAI now has the fastest growing platform in history, a groundbreaking groundbreaking breakthrough in the GPT language model.

But thanks to an investment in Norwegian company 1X, formerly known as Halodi Robotics, it looks like OpenAI is ready to step back into embodied intelligence. Details are scarce at this point; 1X announced the closing of a US$23.5m Series A2 Funding Round on March 23, led by the OpenAI Startup Fund.

androids for real world manipulation | 1 TIME

Halodi has made some progress with its robot eve, a humanoid on a rolling stand, a version of which can be seen in the video above. But a wheeled robot is limited, in terms of the human tasks it can take on, and 1X is now focusing on a bipedal platform called Neo, to “explore how artificial intelligence can take shape in a body of human type”.

He somewhat resembles Figure’s design, in that he appears to use electric rather than hydraulic actuators, he runs a screen for a face, and he uses humanoid hands.

The company doesn’t have much more to say at this point other than the words “Summer 2023”, which seems to indicate that we’ll be meeting Neo-on-legs very soon. It will be fascinating to see what OpenAI contributions bring to a project like this.

This space may well become the next high-tech gold rush, because as both Tesla and Figure point out, whoever gets a general-purpose humanoid first, capable of doing real work in a fairly broad, is in an incredible position to generate almost unlimited money and value.

Workers tend to be much cheaper to buy than to hire. Some of the most shameful chapters in human history demonstrate just how much wealth can be generated if you own a submissive, non-union workforce that operates under zero OHS requirements, requires no pay no sick leave and may be responsible for managing a range of different tasks.

A humanoid robotic revolution could deliver many of the same benefits in a much more morally defensible way – although if it succeeds enough (and it appears it will, it’s only a matter of time) – it will open its own Pandora’s box of societal consequences.

Of course, the human form is definitely not optimized to do today’s work, so at some point it will start to make sense to create superhuman robots that are faster and stronger than the best of today. between us, with greater reach, extra legs and arms, and completely different form factors. And that’s when things start to get really weird and scary.

Sources: 1 TIME, Figure

Leave a Comment