@jbigham@hci.social
@jbigham@hci.social avatar

jbigham

@jbigham@hci.social

Professor of HCII and LTI at CMU SCS. Manage human-centered ML at . #hci, #a11y, #dialog, #nlp. AI/ML. #yinzfluencer of 4 kids. aspire to exceed my attention-seeking pattern regurgitating machine colleagues.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

jbigham, to random
@jbigham@hci.social avatar

has anyone done an analysis of how various conference locations change the average travel distance of CHI attendees?

i thought part of Hawaii rationale was that it is sort of in between Asia and the USA, even though relatively few attendees are super close to Honolulu.

given that flight times over the curved globe often don't match our intuitions, i'm not sure Hawaii helps that much.

shaves off only a couple hours from Asia compared to west coast of the US, for instance.

jbigham, to random
@jbigham@hci.social avatar

perhaps a question with an obvious answer, but i don't think so, what is the point of rejecting papers that aren't "flawed"?

i'm in the middle of 4 overlapping PC processes, … it's a lot of work, and most of that work comes down to debating whether to reject papers that are "borderline", where there's no fatal flaws, just maybe they're not quite clear enough or maybe the delta over prior work isn't big enough etc.

is this a good use of our time?

jbigham, to random
@jbigham@hci.social avatar

who wants to put together the Top 100 HCI … we construct it to be as amazing as possible, but then just publish it and then run away. if we do this right, we can fund CHI for the next 40 years off the click revenue.

jbigham, to random
@jbigham@hci.social avatar

some of the papers i was a coauthor on were accepted, and some were rejected. i can honestly say i would not have predicted which ones fell into which categories correctly, which is probably a problem.

jbigham, to random
@jbigham@hci.social avatar

Does anyone personally find the Nest thermostat’s “learning” to be useful?

I had thought it was well accepted that the learning was mostly for marketing, and the real selling points were (i) a nice and different looking thermostat, (ii) easy instructions for replacing your ugly thermostat with a Nest, and (iii) remote control.

But, I’ve heard a lot of people say that the learning is important? Really? I have three and I’ve never even noticed it and it seems like it would be annoying.

jbigham, to random
@jbigham@hci.social avatar

remember how threads was a thing?

jbigham, to random
@jbigham@hci.social avatar

isn't the rational option for people who think Trump is uniquely bad for the country to register as a republican to vote for whoever is polling #2 in the primaries?

jbigham, to random
@jbigham@hci.social avatar

one of the problems with mastodon is that people don't actually like to boost things, and so the lack of an algorithm means a lot of cool stuff doesn't have an easy way to spread

jbigham, (edited ) to random
@jbigham@hci.social avatar

compared to today (August 2023), will the way we interact with computers in 10 years be:

jbigham, to random
@jbigham@hci.social avatar

all conferences should be located within 5 miles of a hub airport in a low-cost city. sorry pittsburgh, i love you, but i don't make the rules. except this rule, which i'm trying to make.

jbigham, to random
@jbigham@hci.social avatar

we are experimenting in my new startup with controlling a highly dangerous LLM with a low-cost logitech game controller.

see my profile for how to apply to work with us. stanford affiliation is required. we allow dogs in some offices.

jbigham, to random
@jbigham@hci.social avatar

it's kind of amazing that the only requirements to capture Twitter's market are: (i) work exactly like twitter, and (ii) owners don't be super public about being awful… and, only bluesky seems to be even trying?

jbigham, to random
@jbigham@hci.social avatar

word is the number of CHI paper submissions is up like 25% as compared to last year?? 🙃

jbigham, to random
@jbigham@hci.social avatar

my university has these amazingly complicated PDF documents for various purposes that let you attach other PDFs to them. but, nobody is able to figure out how to reliably add alt text to PDFs. that is the hardest problem. apparently.

jbigham, to random
@jbigham@hci.social avatar

what other programmers out there use print statements instead of debug tools 99.99% of the time? what messages do you put in the print statements? print("yo dawg, made it to the end of this toot!")

jbigham, to random
@jbigham@hci.social avatar

alignment is so bizarre but so straightforwardly the way ML folks would think about values … assume you have figured out how to write down all your values in an unambiguous policy, now work on the hard problem of getting the model to follow those things!

jbigham, to random
@jbigham@hci.social avatar

maybe we should have other moratoriums, but can we at least have a moratorium on saying “AI did stuff”?

jbigham, to random
@jbigham@hci.social avatar

i think i had enough submissions to say with some statistical backing and not entirely sour grapes that review outcomes felt pretty random. that can't be good, right? most of my papers made it to RR, a primary reason one didn't was b/c a new method we introduced for self-training small models didn't outcompete GPT4… NAACL/EMNLP might have their problems, but …

jbigham, to random
@jbigham@hci.social avatar

we have chosen to put most of our research into documents in PDF format.

PDFs are a huge pain to make accessible.

most scientists write their papers in Latex, overleaf, etc., which cannot produce accessible PDFs.

to make such PDFs accessible, one uses Adobe Acrobat, which is expensive and proprietary.

increasingly, we post our PDFs to arXiv, which ~forbids accessible PDFs b/c they can't be compiled from source.

~none of our science is accessible.

artifacts (and file formats) have politics.

jbigham, to random
@jbigham@hci.social avatar

i am now returning from an in-person workshop halfway around the world with great people. it was amazing. we should really make a grand challenge of HCI being to replicate this kind of experience virtually because nothing is anywhere in the ballpark now.

jbigham, to random
@jbigham@hci.social avatar

sometimes i think i should get even more into rural hobbies to bring these things to the city, which is missing out. like hunting. but i don't want to hurt animals. has anyone tried hunting with like those plastic darts i had to use as a kid so i wouldn't poke my eyes out / damage the walls? it's either hunting or raising animals for food or seniors in high school driving tractors to prom.

jbigham, to random
@jbigham@hci.social avatar

this is the schedule for being a AC… the amount of steps is too damn high! :)

it's too many steps to fit within the generous tooting limits, hence the picture (with alt text).

jbigham, to random
@jbigham@hci.social avatar

nice thing about being a professor is you can look at work and non work things on the internet all the time

jbigham, to random
@jbigham@hci.social avatar

i'm going to say something that i think isn't controversial -- if a NYTimes reporter has to spend 3 days trying every possible way to "trick" your LLM into saying the thing they're desperately asking it to say… that's not a safety concern.

jbigham, to random
@jbigham@hci.social avatar

review that recommended RRX for us dinged us for novelty compared to an arxiv paper that was publish about 20 days before the CHI deadline.

assuming that's even a good criticism -- do we just cede to whoever posts on arxiv first now?

what do other communities do?

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • megavids
  • kavyap
  • DreamBathrooms
  • thenastyranch
  • magazineikmin
  • InstantRegret
  • GTA5RPClips
  • Youngstown
  • everett
  • slotface
  • rosin
  • osvaldo12
  • mdbf
  • ngwrru68w68
  • JUstTest
  • cubers
  • modclub
  • normalnudes
  • tester
  • khanakhh
  • Durango
  • ethstaker
  • tacticalgear
  • Leos
  • provamag3
  • anitta
  • cisconetworking
  • lostlight
  • All magazines
    •