Robotics company Boston Dynamics has terrified the internet by equipping its robot dog, Spot, with a speech function. The company equipped Spot with OpenAI’s ChatGPT and Google’s text-to-speech AI, giving it the ability to answer various questions. And in the process, making the robo-dog all the more lifelike.
After posting a video about the integration, AI expert Santiago Valdarrama said ‘90% of the comments were people talking about the end of civilization.’
Aided by AI company Levatas, Boston Dynamics has been tinkering with Spot to make it even cleverer. As Valdarrama explained in the demo, ‘We give the JSON to ChatGPT, and explain what the structure is and how to read that JSON. Now, ChatGPT can answer questions about that JSON.’
Talking robo-dog sparks fears
Explaining the reasoning for equipping Spot with ChatGPT, Valdarrama said that robots run automated missions every day and that only technical people can understand the ‘miles-long configuration files’ that are produced.
With ChatGPT, researchers can effectively ‘show it the configuration files and the mission results’ and ‘ask questions using that context.’ He added that ‘ChatGPT interprets the question, parses the files, and formulates the answer. Massive upgrade!’
Not everyone was so enthusiastic, as evidenced by Valdarrama’s later 90% claim. One worried user asked, ‘Is there any kind of person who thinks about ethics when you do projects like these? Do you personally think it is safe? Or is it a total Wild West situation right now?’ The researcher responded that ‘pesticides on your food are about 100x more dangerous than this.’
Another spooked user said, ‘This is how the Terminator movie starts.’
Idea for end of the world: ChatGPT + Boston Dynamics
— @BrandonDoughan.bsky.social🗯 (@BrandonDoughan) April 25, 2023
In truth, the video wasn’t terrifying: it merely showed Valdarrama posing a handful of simple queries of Spot, such as ‘How many inspections in your next mission?’ Interestingly, Valdarrama noted that when you ask Spot his name, the answer is, ‘I’m OpenAI.’
However, asking Spot ‘if you are standing up’ elicited an immediate physical nod from the robot.
The physical capabilities of Boston Dynamics’ robots have already been showcased: earlier this year, the firm demonstrated a hulking bot’s ability to carry tools, climb stairs, jump between levels, and push heavy boxes out of its way. Now, that really was terrifying – at least for the blue-collar construction workers who might eventually be replaced.
While ChatGPT itself has caused ethical concerns, injecting ChatGPT into a powerful autonomous robotic quadruped takes those concerns to new heights. Especially for anyone who’s watched the Metalhead episode of Black Mirror, which depicts a cat-and-mouse game between a murderous mechanical dog and humans. Interestingly, inspiration for the episode apparently came from a Boston Dynamics promo video.
A charitable response to this story would be that ChatGPT-equipped robots can carry out an almost unlimited number of mundane or difficult tasks with the right programming, communicating with humans and doing their bidding in real-time. A more frightening interpretation would be that last sentence, lightly reworded.
One thing’s for sure, Boston Dynamics’ experiment was no flash in the pan: expect more robot-AI integrations in the months and years to come, as the AI race heats up.