Typical university press release: this technology can revolutionize our energy grid by storing energy in a building’s foundation, because we powered an LED with some tiny cells we made.
The first sustained artificial nuclear reaction [1] managed to produce...half a watt. I'm glad they weren't overly cynical about future possibilities.
I think they kind of lost out in Deep Learning to Berkeley and Stanford but they’re slowly building their expertise in it (frankly most of Academia doesn’t compare to industry outputs in Deep Learning so I don’t think it’s that big a deal)
Yeah, so good they were extensively faking their research and demos for a "personal food computer"...and even selling the fucking things to unsuspecting schools. Literally snake oil scamming.
https://www.google.com/search?client=firefox-b-1-d&q=MIT+med...
Then there's your robotics team, backed by millions of dollars in free products and "consulting" (ie doing work for them) in sponsorship by defense contractors...beaten by a bunch of high school kids who raided a BoatUSA store and Home Depot to build an underwater robot with bilge pumps and a bunch of PVC pipe.
The most exciting thing to come out of MIT Media Lab in two decades is an alarm clock wrapped in carpet that rolls off your nightstand when it goes off.
MIT is a defense industry thinktank masquerading as a university.
Russ’s student Scott later went on to make Atlas (Boston Dynamics Humanoid) backflip. They were the first to come up with a backflipping quadruped. Alberto’s team won the Amazon pick and place challenge (also demonstrated a robot playing Jenga). Recently Pulkit showed how to rotate any arbitrary object with a dexterous robotic arm. I’d honestly find it hard to name universities better than MITs at robotics
What human being would even be qualified to judge the research output of a large research university across all fields? A research university is an umbrella under which hundreds of independent teams perform research. You can do journal metrics, but those have a long latency, you are measuring performance 2 to 10 years in the past, and such metrics are not without bias.
of q^2/2C