Mastodawn

Suresh Venkatasubramanian Jan 24, 2023

Never been happier to hear a complaint from House committee chairs.

Conflicting guidance from the White House on #AI can hurt our ability to grow this technology. Read more from @[email protected] and @[email protected] of @[email protected]: https://republicans-science.house.gov/press-releases?id=8694F377-4FD1-4F0D-BF9A-33DDCFDE2B06

🐦🔗: https://twitter.com/housescience/status/1616489053519478788

Lucas and Comer: White House is Giving Conflicting Guidance on Artificial Intelligence

Lucas and Comer sent a letter to the White House Office of Science and Technology Policy (OSTP) raising concerns about conflicting guidance from the administration on Artificial Intelligence (AI).

House Committee on Science Space & Tech - Republicans

Suresh Venkatasubramanian Jan 3, 2023

This is a very nice analysis of what LLMs can and cannot do right now.

RT @[email protected]

Everyone are writing about LLMs so I did too. (It became longer than I expected. and also not long enough.)

https://gist.github.com/yoavg/59d174608e92e845c8994ac2e234c8a9

🐦🔗: https://twitter.com/yoavgo/status/1610055975389614080

LLMs.md

GitHub Gist: instantly share code, notes, and snippets.

Gist

Suresh Venkatasubramanian Dec 31, 2022

RT @[email protected]

@[email protected] But he also provides context about Signal specifically, right? Not keeping metadata to avoid court orders to collect metadata seems notable. Curious about the details; does that interfere with a pen/trap order? Or only historical? Seems pretty important to me.

🐦🔗: https://twitter.com/OrinKerr/status/1608207346274283521

Orin Kerr on Twitter

“@matthew_d_green But he also provides context about Signal specifically, right? Not keeping metadata to avoid court orders to collect metadata seems notable. Curious about the details; does that interfere with a pen/trap order? Or only historical? Seems pretty important to me.”

Twitter

Suresh Venkatasubramanian Dec 30, 2022

Given that R members of Congress have been threatening hearings over Exec use of signal, this is unsurprising.

RT @[email protected]

Cassidy Hutchinson told the select committee that "dozens" of lawmakers she communicated wit hused Signal — including GOP Leaer McCarthy.

🐦🔗: https://twitter.com/kyledcheney/status/1608881870107811841

Kyle Cheney on Twitter

“Cassidy Hutchinson told the select committee that "dozens" of lawmakers she communicated wit hused Signal — including GOP Leaer McCarthy.”

Twitter

Suresh Venkatasubramanian Dec 21, 2022

Seen on my desi WA groups:

Show thread

Suresh Venkatasubramanian Dec 19, 2022

"it's not a USB drive, it's a ginger mint". An important distinction for sure.

Suresh Venkatasubramanian Dec 19, 2022

Just started to read the Jan 6 committee executive summary: https://january6th.house.gov/sites/democrats.january6th.house.gov/files/Introductory%20Material%20to%20the%20Final%20Report%20of%20the%20Select%20Committee.pdf

It's gripping and alive in a way that you don't really expect a government report to be. And has dry gems like this one:

Suresh Venkatasubramanian Dec 19, 2022

In which my hands are an independent participant who occasionally coordinate with my head.

RT @[email protected]

‼️ New episode ‼️ Tune in to discussion between @[email protected], @[email protected] and special guest co-host @[email protected] of @[email protected] to learn how we can (1) reimagine AI design (2) implement best practices from the #AIBillofRights BluePrint and (3) much more!

https://share.transistor.fm/s/9fb770f2

🐦🔗: https://twitter.com/ai_equal/status/1604868423955406848

In AI We Trust? | Dr. Suresh Venkatasubramanian (White House OSTP/Brown University): Can AI be as safe as our seatbelts?

In this episode, we are joined by Dr. Suresh Venkatasubramanian, a former official at the White House Office of Science and Technology Policy (OSTP) and CS professor at Brown, to discuss his work in the White House developing policy, including the AI Bill of Rights Blueprint. Suresh also posits on the basis for current AI challenges as failure of imagination, the need to engage diverse voices in AI development, and the evolution of safety regulations for new technologies.

Suresh Venkatasubramanian Dec 16, 2022

My work here is done, part I: 1/2

Suresh Venkatasubramanian Dec 7, 2022

This is utterly bonkers. And why I'm failing to resist the siren call of ChatGPT as #ANewThing

RT @[email protected]

ChatGPT can solve novel, undergrad-level problems in *computational complexity* 🤯

"Please prove that the following problem is NP-hard..."

Solution in next tweet -->

Credit: @[email protected]
(1/3)

🐦🔗: https://twitter.com/OfirPress/status/1600434732705415171

Ofir Press on Twitter

“ChatGPT can solve novel, undergrad-level problems in *computational complexity* 🤯 "Please prove that the following problem is NP-hard..." Solution in next tweet --> Credit: @TzvikaGeft (1/3)”

Twitter