Sam Tobin-Hochstadt

Associate Professor, @IULuddy ·
Core Developer, @racketlang ·
Member, @TC39 ·
Handler, @gravemaker_ulti ·
@[email protected]

[Sam Tobin-Hochstadt @samth]

One week with an e-bike and I'm definitely an "e-bike guy" now.

[Sam Tobin-Hochstadt @samth]

@MattBruenig When you do percentiles like this you (Claude) should add some annotations about where 10k, 100k, 1 million etc is.

[Sam Tobin-Hochstadt @samth]

@tracewoodgrains @BenShindel But in fact you should _want to go_. It's not just that the kid needs to visit their grandmother, they also need to learn that the preference to play video games instead of doing it is bad.

[Sam Tobin-Hochstadt @samth]

@dylanmatt @Miles_Brundage But the only real problem with the model is that the billionaires aren't taking Miles' advice.

[NeurIPS Conference @NeurIPSConf]

This year, the NeurIPS 2026 Position Paper Track made the decision to require that all papers be substantially human-written, with AI used for only copy-editing or similar peripheral changes to the main text!

For more details, please check our blogpost: https://t.co/wrWuMQJwrx

[Sam Tobin-Hochstadt @samth]

@marcinpaprzycki @lukehoban Obviously but what does that have to do with what I said?

[Sam Tobin-Hochstadt @samth]

@marcinpaprzycki @lukehoban What are you talking about?

[Sam Tobin-Hochstadt @samth]

@lacker @QiaochuYuan Authorship and credit is about the relationship between people, and AI systems are not people.

[Sam Tobin-Hochstadt @samth]

@gabrielwinant There aren't even really any rich people that want to set up new universities for the glorification of rich people!

[Sam Tobin-Hochstadt @samth]

@ShriramKMurthi Now I want you to include runtime errors with stack traces and/or blame as a possibility.

[Sam Tobin-Hochstadt @samth]

@Keleesssss @davidjustodavid @vardi My understanding is that the false positive rate for Pangram is very low.

[Sam Tobin-Hochstadt @samth]

Why did CACM just publish a blog post by a B2B SAAS guy about how he's in the middle of vibe-proving P vs NP? What possible editorial process could this have gone through? @vardi

[Sam Tobin-Hochstadt @samth]

@ShriramKMurthi @dimvar I just wish they would actually do the work! I don't need 14 command line flags on the throwaway python script you wrote, I need you to actually answer the question I asked!

[Sam Tobin-Hochstadt @samth]

How can I make them never talk like this again?

"You're right to push on this, and I think you've caught me accreting special cases instead of finding the general mechanism. Let me answer the three questions honestly."

[Sam Tobin-Hochstadt @samth]

@arpitrage More than hard, these questions seem conceptually under determined and not necessarily admitting of the kinds of answers we want.

[Sam Tobin-Hochstadt @samth]

@MattZeitlin People's expertise is usually more detailed and more specialized than other people expect.

[Sam Tobin-Hochstadt @samth]

@maxvonhippel @kirancodes

[Sam Tobin-Hochstadt @samth]

@dimvar Yes that was Claude although GPT talks like that too sometimes.

[Sam Tobin-Hochstadt @samth]

@vardi @ShriramKMurthi You should try to restrict quotation marks to things that people have in fact said. The flippant remarks in this conversation have been yours alone.

[Sam Tobin-Hochstadt @samth]

@richardkelley Right I agree he's not being disingenuous, he just doesn't know very much. But that's why there should be some sort of review process so that it doesn't appear on the CACM website.