By Michael Scott
Days after Figure chief executive Brett Adcock confirmed $675 million financing from Microsoft, OpenAI, Nvidia, Intel, Jeff Bezos, and Ark Invest, the AI robotics company had unveiled a human-robot exhibiting common sense.
The AI robotics developer made waves following the introduction of a humanoid robot that infuses OpenAI technology to engage in real-time conversation and execute tasks simultaneously.
Figure revealed in a post published on Twitter that infusing OpenAI technology allows Figure 01 hold entire conversations with humans. The humanoid has a demonstrative capacity to understand and instantly react to human interactions.
Figure Taps OpenAI Tech in Conversational Robot
The Figure post hailed its recently unveiled alliance with Sam Altman-led OpenAI to utilize the latter’s language intelligence and high-level visuals for the robots. The integration allows its products to take fast, low-level, and dexterous actions.
The clip reveals Figure 01 interacting with Seniro AI Engineer Corey Lynch. The creator engages the robot in executing several tasks in what appears to be a makeshift kitchen, which it successfully undertakes.
Figure 01 quickly identifies apples as food when Lynch asks to give something to eat. Figure 01 would later collect trash into the appropriate basket while simultaneously responding to questions as evidence of its multitasking capabilities.
Lynch took to the X post to explain Figure 01 capabilities in detail. The robot can describe the visual experience and schedule future actions. The AI expert revealed that it reflects on the memory blend with verbal reasoning.
Lynch confirmed that the development team feeds images to the cameras besides transcribing text tapped from speech that onboard microphones capture to the large multimodal model that Microsoft-backed OpenAI trains. Multimodal AI involves AI capable of understanding and generating various data types, including images and text.
Figure 01 Humanoid Robot Holding Conversation
Lynch confirmed that the behavior exhibited by Figure 01 is learned and executed at normal speed, thus ruling out remote controls. The AI engineer explained that the model was processing the conversation history to generate language responses that the robot would speak to the humans through a text-to-speech procedure.
Lynch added that the model is utilized to decide whether the learned and closed-loop behavior will run on the robot so that it can fulfill the given command. Also, the model facilitates loading neural network weights onto the respective GPU to allow it to execute the policy.
Lynch hailed the Figure 01 design, enabling it to describe the surroundings concisely and apply common sense in executing decisions. It infers actions such as placing dishes in a rack and parsing vague statements.
The unveiling of Figure 01 triggered a passionate response in the X community, with the majority hailing its capabilities. A significant number added it to the mileposts of realizing singularity.
Lynch informed AI developers and researchers that Figure 01 leverages neural network and visuomotor transformer policies to influence behaviors. Also, it maps the pixels directly to the actions. The networks are responsible for receiving onboard images at 10hz and generating 24-DOF actions at 200hz.
Is Figure 01 Debuts Fulfilling Utilitarian Objective?
The debut of Figure 01 coincides with global leaders and policymakers deliberating on addressing AI proliferation in the mainstream sectors. However, a substantial amount of the ongoing discussion is centered on the large language models such as ChatGPT, Anthropic’s Claude AI, and Gemini from Google.
Tesla Inc. chief executive Elon Musk urged the integration of utilitarian objectives that US Berkeley’s professor Ken Goldberg acknowledged as a priority.
The industrial engineering scholar considers that the basis of accelerated investment in companies such as Figure is motivated by the hope that such could become compatible, particularly with space exploration initiatives.
Figure AI is set to witness rivalry from Hanson Robotic in its quest to merge AI with its products. The firm has sought to enhance the Desdemona AI robot since its debut in 2016.
Lynch admitted the accelerated changes thanks to AI facilitating innovations that make holding entire conversations with the humanoid robot a reality. Today, the senior AI engineer is behind the humanoid robot, capable of planning and executing fully learned behaviors.