4 min read
Reuters obtained an internal business letter from Meta Platforms on Friday, which stated that the company is creating a new section within its Reality Labs arm to create humanoid robots driven by artificial intelligence and able to help with physical activity.
With the rise of sophisticated AI models driving innovation in automation and robotics, Facebook parent company Meta is joining the ranks of competitors in the humanoid robots space, which includes companies like Tesla and Nvidia-backed Figure AI.
According to the message from Meta’s Chief Technology Officer Andrew Bosworth, the robotics product division will be concentrating on developing “consumer humanoid robots with a goal of maximizing Llama’s platform capabilities.”
Meta has a developing suite of generative AI products on its social media platforms, and the name of the main series of AI foundation models is Llama.
“We believe expanding our portfolio to invest in this field will only accrue value to Meta AI and our mixed and augmented reality programs,” he said.
Leadership and Strategic Appointments
Meta has appointed Marc Whitten, the former CEO of self-driving car company Cruise, as vice president of robotics, according to Bosworth’s memo. Whitten will be leading the group. The news was initially reported by Bloomberg News.
John Koryl was also appointed Vice President of retail, according to Bosworth. Meta employs Koryl, the former CEO of the second-hand e-commerce firm The RealReal, to increase sales of its Quest mixed reality headsets and artificial intelligence wearables, including Ray-Ban Meta smart glasses, which are a joint venture between Meta and eyewear manufacturer EssilorLuxottica.
Meta Lab’s temporary store in Los Angeles inaugurated in November is one example of how the company plans to establish “more direct expertise in the space” through retail sites. However, according to Bosworth, retail partners will still contribute significantly to hardware sales.
Reuters attempted to reach Meta for comment, but the company did not act quickly.
Challenges Facing AI Robotics
This news is a huge gamble on Meta’s Reality Labs division, which has lost money since its inception despite its stated goal of creating growth-driving technology. In the last three months of 2017, the division lost about $5 billion.
The development of AI-powered gadgets for manufacturing, logistics, and home functions is receiving massive funding from major tech giants and startups. Researchers have discovered that the advancements in artificial intelligence (AI) that have driven chatbot development have not necessarily helped in understanding the actual world, which has slowed progress on robots.
Meta has long supported studies in “embodied AI,” intending to create AI personal assistants with vision, hearing, and movement capabilities in the physical, three-dimensional environment. Yann LeCun, the company’s top AI scientist, has also been very open about pointing out the problems with language models and working on new models that can better understand space.
With the debut of her “spatial intelligence” startup World Labs last year, prominent computer scientist Fei-Fei Li entered the fray, and Elon Musk, CEO of Tesla, has stated that his company’s humanoid robot “Optimus” can carry out a wide range of everyday chores and would then be offered to customers.
Apptronik, a company that makes humanoid robots powered by artificial intelligence and made for use in factories and warehouses, raised $350 million on Thursday from investors including Alphabet’s Google in order to increase production.
According to Bloomberg, Meta intends to build its own hardware for humanoid robots, with an initial focus on domestic tasks. The company also aspires to develop artificial intelligence (AI), sensors, and software that can be used to power robots produced and sold by other companies.
Meta is reportedly in discussions with robotics firms including Unitree Robotics and Figure AI about a potential partnership, but the company has no imminent plans to introduce its own brand of robot, according to the source.
