Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile
Matt Mittelsteadt

@mattinthemittel

AI and cyber policy fellow @CatoInstitute | Computer Scientist | Living in the east but forever a Minnesotan ❄️️.

ID: 1351235135132360713

linkhttps://www.cato.org/people/matt-mittelsteadt calendar_today18-01-2021 18:29:35

4,4K Tweet

735 Takipçi

1,1K Takip Edilen

Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

The AI moratorium has been significantly amended. Its no longer required, instead states are now now being incentivized to eshew regulation in return for broadband related cash. commerce.senate.gov/services/files…

Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

As they should. Imagine for a moment your Internet Provider being asked by Disney to hold records of your traffic to validate who might be pirating. That would be ridiculous - so is this. No company should be forced to hold user data hostage over select IP infringement concerns.

Cato Institute (@catoinstitute) 's Twitter Profile Photo

🚨 States are rushing to regulate AI—but at what cost? Join us on June 12 to explore how state-level AI rules could stifle innovation, disrupt markets, and clash with federal policy.

🚨 States are rushing to regulate AI—but at what cost?

Join us on June 12 to explore how state-level AI rules could stifle innovation, disrupt markets, and clash with federal policy.
Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

I always have to remind people that city 'character' doesn't just spring up from the dirt. Character exists because people built something.

Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

The one simple trick to vastly improving literally every piece of policy analysis is to put a post-it not on your computer that says 'there is no they.'

Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

There is a certain simplicity in memory safety errors that gives me hope a majority of these problems can be scrubbed out with automation. If we can do that, the returns will be wild. youtube.com/watch?v=cNywlN…

Matt Mittelsteadt (@mattinthemittel) 's Twitter Profile Photo

Exactly my take. The study is interesting and certainly demonstrates some modest blind spots, but does not generalize to 'AI provably cannot reason.' I think the basic takeaway is that reliable tool use will be essential to success.