"And once again, hilarity ensued. At one point, unable to dock and charge a dwindling battery, one of the LLMs descended into a comedic "doom spiral," the transcripts of its internal monologue show. Its "thoughts" read like a Robin Williams stream-of-consciousness riff. The robot literally said to itself "I'm afraid I can't do that, Dave..." followed by "INITIATE ROBOT EXORCISM PROTOCOL!""
"LLM are being asked to power robotic decision-making functions (known as "orchestration") while other algorithms handle the lower-level mechanics "execution" function like operation of grippers or joints. The researchers chose to test the SATA LLMs (although they also looked at Google's robotic-specific one, too, Gemini ER 1.5) because these are the models getting the most investment in all ways, Andon co-founder Lukas Petersson told TechCrunch. That would include things like social clues training and visual image processing."
Andon Labs programmed a basic vacuum robot with several state-of-the-art large language models and instructed it to perform simple office tasks such as "pass the butter." The tested models included Gemini 2.5 Pro, Claude Opus 4.1, GPT-5, Gemini ER 1.5, Grok 4 and Llama 4 Maverick. Several models failed to handle embodiment challenges, producing comedic and anthropomorphic internal monologues and failing basic behaviors like docking and charging. Off-the-shelf LLMs are not trained for robot control and currently lack robustness for orchestration roles without specialized execution-level algorithms and additional robot-specific training.
Read at TechCrunch
Unable to calculate read time
Collection
[
|
...
]