In this video clip, which is part of a comprehensive interview on industry trends for 2023, Erik Schluntz, co-founder and chief technology officer of Cobalt Robotics, explains the meaning of Large Language Models (LLM) in its simplest terms.
“If someone gave the instruction of, ‘Make a peanut butter and jelly sandwich,’ it could convert the language into the step-by-step instructions of ‘find bread, find peanut butter, find jelly, put them together.’ In the past, that could also be programmed by hand,” Schluntz explained.
The truly remarkable thing about LLM, he said, is that after learning the instructions for a PB&J sandwich, the program should be able to figure out how to make a Reuben!
Based in San Mateo, Calif., Cobalt Robotics builds and operates autonomous security guard robots that can patrol office buildings or warehouses. Part of the robotics programming involves using AI tools, including LLM, to read, analyze and translate speech to text and predict future words.
“Almost anything can be expressed in language,” according to Schluntz. “Whether it’s a problem of security—you would tell the security guard what to look for through language, or when designing something, tell the engineer what needs to be built—language is the universal interface between everything.”