1X’s VP of AI Eric Jang was interviewed recently by Jason Carman and revealed the strategy 1X plans on employing to develop autonomous AI for Neo.
Jang talks about his interest in living forever inspiring his decision to get into robotics due to the fact you can simulate the biological physics of a brain using a silicon computer. Which might open up possibilities for immortality.
Jang and his team at 1X, which consists of former Tesla and Google employees, are looking to build an intelligence that will have the ability to predict the future in a similar fashion to the human mind.
For instance they want Neo to be capable of observing a person and predicting what actions or movements they might perform next. This will enable Neo to perform tasks such as communicating non-verbally as was shown in the promotional video last month.
Jang predicts humanoids will be the only obvious choice of robot to perform practical tasks in the home in three years, something which isn’t obvious in mainstream robotics today.
Jang outlines his reasoning for believing humanoids are a viable solution now due to:
•Actuation technology becoming cheaper whereas previously it was prohibitively expensive.
• Recent developments in machine learning such as transformer architecture and large language models have enabled the ability to create software that can perform many tasks on a wide variety of form factors on one appendage eg controlling the elbow movements in an arm while also controlling the finger movements. Which was previously an incredibly challenging task to achieve.
•Building complex software solutions instead of focusing on quick general solutions as was previously the norm in Silicon Valley.
Jang revealed 1X’s plans for Neo:
•Make a general purpose intelligence for the hardware.
•Gather a large amount of teleoperated data.
•Train it with multimodal models.
This was a really fascinating interview and must watch for anyone interested in AI and robotics as Jang does a great job at explaining highly complex concepts that would otherwise be lost on the casual observer.
Jang also believes AI will one day allow humanity to view itself from an entirely new perspective, an idea he came across while reading the book Klara and the Sun by Kazuo Ishiguro.
Jang talks about his interest in living forever inspiring his decision to get into robotics due to the fact you can simulate the biological physics of a brain using a silicon computer. Which might open up possibilities for immortality.
Jang and his team at 1X, which consists of former Tesla and Google employees, are looking to build an intelligence that will have the ability to predict the future in a similar fashion to the human mind.
For instance they want Neo to be capable of observing a person and predicting what actions or movements they might perform next. This will enable Neo to perform tasks such as communicating non-verbally as was shown in the promotional video last month.
Jang predicts humanoids will be the only obvious choice of robot to perform practical tasks in the home in three years, something which isn’t obvious in mainstream robotics today.
Jang outlines his reasoning for believing humanoids are a viable solution now due to:
•Actuation technology becoming cheaper whereas previously it was prohibitively expensive.
• Recent developments in machine learning such as transformer architecture and large language models have enabled the ability to create software that can perform many tasks on a wide variety of form factors on one appendage eg controlling the elbow movements in an arm while also controlling the finger movements. Which was previously an incredibly challenging task to achieve.
•Building complex software solutions instead of focusing on quick general solutions as was previously the norm in Silicon Valley.
Jang revealed 1X’s plans for Neo:
•Make a general purpose intelligence for the hardware.
•Gather a large amount of teleoperated data.
•Train it with multimodal models.
This was a really fascinating interview and must watch for anyone interested in AI and robotics as Jang does a great job at explaining highly complex concepts that would otherwise be lost on the casual observer.
Jang also believes AI will one day allow humanity to view itself from an entirely new perspective, an idea he came across while reading the book Klara and the Sun by Kazuo Ishiguro.
