On Tesla AI Day 2022, Elon Musk was asked if he would be willing to open-source Optimus Robot to let the whole world explore humanoid research, rather than keeping it exclusively for Tesla. In his response, Musk expressed concerns about how Optimus can be potentially used in “some ways that are bad”. Though Tesla would allow users and researchers to provide instructions into the humanoid, those should be governed by some laws of robotics related to safety, like “not being harmful to others”.

Musk’s POV of the risks of robotic software getting into everyone’s hands probably comes from his perspective on the possible dangers of AI. In an interview in 2014, Musk had said that the developments in AI are akin to “summoning the demon” and also wrote on Twitter about getting “inspired” from The Terminator movie.

Why (not) open-source?

When it comes to the benefits of open-source, there are plenty and obvious ones. Developers do not have to start from scratch when experimenting and building their bots. This way, researchers and start-ups can focus on what they are trying to build rather than building the architecture from scratch. Though this can also be one of the reasons for big-league robotic firms like Tesla not open-sourcing everything they put into the field for grabs.

Furthermore, if the software is open-source, researchers and developers can contribute to the project directly and benefit the robotics community at large. 

The reasons to not open-source Optimus sound untrue when there are so many other robotic researchers and companies who have made their software publicly available. Probably the most notable, Boston Dynamics’ robot dog, Spot, was open-sourced back in January 2020. The very famous, promising, and 3D printable humanoid robot, Poppy, is also open-source and has a structure of a small version of human and biped locomotion, similar to Optimus. There are open-source humanoids including iCub, InMoov, and Thormang3, among others.

The much-awaited robot from Tesla received mixed reviews from across the robotics community. A large number of experts were excited as well as sceptical about Optimus as Musk did not hire academic experts of robotics. 

Animesh Garg, assistant professor at the University of Toronto, said that though Tesla has done an astounding job in a year to develop Optimus, the company could have achieved more by being more open to the community and receiving feedback while working on their projects. 

Gary Marcus from Robust.AI also showed disappointment about the vision of Optimus calling it “a bit of a dud”.

Open-sourcing seems good

Not just robotics, several fields related to AI have witnessed improvements after their developers decided to open-source them. For example Stability.AI made Stable Diffusion open-source and that led to huge innovations in image generation. Unlike DALL-E or Midjourney, Stable Diffusion can run on consumer GPUs now. Emad Mostaque, the developer behind the software, posted on Twitter, “Use this in an ethical, moral, and legal manner”.

DeepMind’s AlphaFold was also made available to the public. This protein-fold prediction software might probably also fall under the same “degree of danger” as robotics. But after the code was made available on GitHub, researchers have been making innovations like creating completely new folds in protein using AlphaFold, thus advancing the field even further.

On the Lex Fridman Podcast, Musk spoke extensively about how Tesla has the most-advanced real-world AI. “I haven’t really thought about this, but there could be a time when there are millions of Tesla robots on the street,” said Musk. “And our goal starts with building a general-purpose help robot. We have been trying to build cars for a while now, which are essentially a robot with four wheels. Building a robot that is beneficial to the world comes with a great responsibility.”

Before Tesla AI Day, Musk tweeted saying, “The point of the AI Day is to show the immense depth & breadth of Tesla in AI, compute hardware & robotics.”  Engineers and researchers willing to dive deep into the field are moved by Musk’s freedom and passion to work for the betterment of humanity. But keeping their developments and innovations behind closed doors will not attract talented people who are motivated to move forward in the field.

Musk believes in empowering more engineers, and making innovations beneficial for humanity is what drives him. A probable reason for not open-sourcing Optimus could be that Tesla wants privacy and complete ownership of development of robots in the world. Being the experts in the race, Tesla is making advancements, though not quickly, but with great precision and competence.