At MIT, Worms and Turtles Are Inspiring a New Generation of Robots

**What's up:** **Meta** is delaying the rollout of its flagship AI model and battling an ‘epidemic of scams’; Silicon Valley’s new hold on Washington; Trump chides **Apple**'s Tim Cook.

Sponsored by

At MIT, Worms and Turtles Are Inspiring a New Generation of Robots

By Tom Loftus

	What's up: Meta is delaying the rollout of its flagship AI model and battling an ‘epidemic of scams’; Silicon Valley’s new hold on Washington; Trump chides Apple's Tim Cook.

	MIT student VeeVee Cai holds Crush, a robotic sea turtle, at a pool on campus. Isabelle Bousquette / WSJ

	Good morning, CIOs. The obsession with humanoid robots notwithstanding, the future may belong to the weirder varieties, as the WSJ’s Isabelle Bousquette discovered on a recent trip to MIT’s Computer Science and Artificial Intelligence Laboratory.

	Think soft and squish robots. Flexible robots. Even edible ones capable of performing surgery.

	“I really wanted to broaden our view of what a robot is,” MIT CSAIL Director Daniela Rus tells the WSJ. “So if you have a mechanism that’s made out of paper and that moves, is that a robot or not? If you have an origami flower that you attach to a motor, is that a robot or not? To me, it’s a robot.”

	A tour of the labs confirmed this decidedly post-humanoid view of the field, where the WSJ encountered “bubble” robots and a swimming, sea-turtle-like robot in rapid succession.

	Now creative new uses of artificial intelligence are pushing her work to a new level. Rus, a recent recipient of the IEEE’s Edison Medal, last year helped spin out a company, Liquid AI, that used technology modeled on the neural activity of worms.

	Rus and her team have also designed a special AI system trained on the laws of physics that suggests robot designs from a prompt. It’s called “text to robot.” Read the story.

											Content from our sponsor: Deloitte

											Is It Time to Modernize Identity Security?

											Implementing a unified identity management platform can enhance incident response and reduce vulnerabilities associated with standing privileges and non-human identities. Read More

											More articles for CIOs from Deloitte 

Share this email with a friend.

Forward ›

Forwarded this email by a friend?

Sign Up Here ›

Delays Hit Meta's Flagship AI Model

	Meta chief product officer Chris Cox at LlamaCon 2025, an AI developer conference, in April. Photo: Jeff Chiu/Associated Press

	Company engineers at Meta Platforms are struggling to significantly improve the capabilities of its “Behemoth” large-language model, slated internally for an April release. The WSJ reports that senior executives are frustrated at the performance of the team that built the Llama 4 models and blame them for the failure to make progress.

	But Meta’s is not alone. Its recent challenges mirror stumbles or delays at other top AI companies that are trying to release their next big state-of-the-art models. Some researchers see the pattern as evidence that future advances in AI models could come at a far slower pace than in the past, and at tremendous cost.