After shocking the world just recently with its hyperrealistic image generator Sora, OpenAI is back at it again – with something even more dystopian.
The Microsoft-backed AI startup has partnered with robotics firm Figure to create a humanoid robot powered by its GPT family of language models.
The result is a robot that can “see”, respond to external stimuli, and converse fluently with humans in its vicinity – it’s equal parts fascinating and frightening.
What is Figure 01?
Figure 01 is a humanoid robot developed by robotics firm Figure. The company posted a video to YouTube this week displaying the robot’s capabilities for the first time.
During the video, a man asks the robot to pass him something edible from the table.
“On it – so I gave you the apple because it’s the only, um, edible item I could provide you with from the table” the robot responds in an eerily human way.
Figure 01 is then instructed to explain why it did that while packing trash away, which it does successfully. In the process, it showcases an ability to simultaneously complete distinct physical and mental tasks.
This just in! View
the top business tech deals for 2024 👨💻
Crucially, Figure 01 is not being operated by anyone. It’s making decisions based on the external stimuli being presented to it. This puts it far, far ahead of OpenAI’s Optimus, another humanoid-like robot developed by Tesla, which has proved less impressive thus far.
What’s Going on Inside Figure 01’s Head?
What’s doing the mental legwork in the video above is a Visual Language Model (VLM) which was developed by Figure in collaboration with OpenAI.
“With OpenAI, Figure 01 can now have full conversations with people” the company explained in a recent tweet.
The robotic company says that the OpenAI models provide “high-level visual and language intelligence” while its own neural networks “deliver fast, low-level, dexterous robot actions”.
Figure 01: What’s Next?
Figure AI was recently valued at $2.6 billion and has attracted significant investment from Amazon founder Jeff Bezos, as well as OpenAI backers Microsoft and fellow tech giant Intel. So, it’s highly unlikely this will be the last we hear from the robotics firm.
Figure has stated previously that its goal is to produce robots that are capable of completing everyday tasks autonomously.
And, while startups, universities, and research groups have made great strides with the physical mechanics in the last decade, we’ve not seen such deep collaboration and integration with an advanced language model thus far.
It’s unlikely we’ll see Figure 01, or any other robotic humanoid, commuting into the office and having chats at the water cooler any time soon. But humanoid robotics factories are on the horizon, with Amazon going to great lengths to make this a reality sooner rather than later.
What’s more, building costs for this sort of technology have dropped sharply in recent years, and with more and more robotics companies receiving millions of dollars in funding, it’s impossible to really gauge how quickly things are truly moving.
