@emilymbender@dair-community.social
@emilymbender@dair-community.social avatar

emilymbender

@emilymbender@dair-community.social

Professor, Linguistics, University of Washington

Faculty Director, Professional MS Program in Computational Linguistics (CLMS)

If we don't know each other, I probably won't reply to your DM. For more, see my contacting me page: http://faculty.washington.edu/ebender/contact/

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

Neverfadingwood, (edited ) to random
@Neverfadingwood@lingo.lol avatar

Last week, there was a story from AFP about AI "deceiving" people. Just now I've listened to an excellent podcast co-hosted by Prof. @emilymbender looking at the peer-reviewed academic paper that was the basis for this story. It is very insightful, and an excellent remedy to the AI hype.
EDIT: It's the May 20 episode.
https://www.twitch.tv/dair_institute

emilymbender,
@emilymbender@dair-community.social avatar

@Neverfadingwood @elmerot That was episode 33! So far, it's only in the intro, but once the episode is published it'll be part of the title.

You can find all back episodes here, once published:

https://www.dair-institute.org/maiht3k/

emilymbender, to random
@emilymbender@dair-community.social avatar

Here's a great reply to Sen Schumer's industry-centric, "innovation"-focused AI report from last week. For policymakers interested in people-centric, evidence-based policymaking, and those who would hold them to it, it's a great overview of the evidence:

https://senateshadowreport.com/

emilymbender, to random
@emilymbender@dair-community.social avatar

Hey folks, let's have a little chat about construct validity --- this is the concept that if you're going to use a psychological test, there should be good reason to believe that a) the thing it's meant to be testing is real and b) results on the test reflect something about that thing.

https://spectrum.ieee.org/theory-of-mind-ai

>>

emilymbender,
@emilymbender@dair-community.social avatar

So, even if a test has been established to have construct validity as a test relating to human cognition, you can't just throw it at a chatbot and take the results as meaningful.

What would it mean for a language model to have a theory of mind? How does string manipulation relate to that? Without answers to these questions, the tests are meaningless.

emilymbender, to random
@emilymbender@dair-community.social avatar

Colleague in globally dispersed organization asks for meeting, says it's time sensitive, wants 8am my time Monday. I arrange my day so I can be ready at that time. Colleague does not show up. Grrr.

emilymbender, to random
@emilymbender@dair-community.social avatar

Join us today!

On the next Mystery AI Hype Theater 3000, guest host @mmitchell_ai and I will look into claims that "AI has learned to deceive humans". Join us live at noon Pacific, May 20, 2024:

twitch.tv/dair_institute

emilymbender, to random
@emilymbender@dair-community.social avatar

This is tomorrow!

On the next Mystery AI Hype Theater 3000, guest host @mmitchell_ai and I will look into claims that "AI has learned to deceive humans". Join us live at noon Pacific, May 20, 2024:

twitch.tv/dair_institute

emilymbender,
@emilymbender@dair-community.social avatar

@kaveinthran You can find all of our episodes here!

https://www.dair-institute.org/maiht3k/

emilymbender, to random
@emilymbender@dair-community.social avatar

This paper and its predecessor that Chirag Shah and I wrote started off as a reaction to Google's plans to shove "GenAI" into search. It was clear that that was a bad idea as early as 2021 (when we started writing the first paper) and it's even clearer now.

https://dair-community.social/@emilymbender/112005034372139457

emilymbender, to random
@emilymbender@dair-community.social avatar

The folks at FTC Office of Technology are doing excellent work taking in research and using it to make a practical difference. See link for a range of topics they are interested in -- for anyone working in these areas, this is a great way to have impact!

https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2024/05/p-np-not-exactly-here-are-some-research-questions-office-technology

emilymbender, to random
@emilymbender@dair-community.social avatar

On the next Mystery AI Hype Theater 3000, guest host @mmitchell_ai and I will look into claims that "AI has learned to deceive humans". Join us live at noon Pacific, May 20, 2024:

twitch.tv/dair_institute

emilymbender, (edited ) to random
@emilymbender@dair-community.social avatar

As folks discuss the plundering of the open internet/sharing economy by the data-hungry LLM trainers, it seems like a good time to remind ourselves to find something other than "the tragedy of the commons" as a metaphor. On the racist, terrible origins of that phrase:

https://discardstudies.com/2019/07/15/the-tragedy-of-the-tragedy-of-the-commons/

karabaic, to random
@karabaic@mastodon.social avatar

So depressed by AI Hype...hey, here's a promising story in Nature about scientists being encouraged to report negative results...I'll read that to cheer me up!

5 grafs later...

https://www.nature.com/articles/d41586-024-01389-7

emilymbender,
@emilymbender@dair-community.social avatar

@karabaic @alex sigh

emilymbender, to random
@emilymbender@dair-community.social avatar

I would like to remind the world that you actually don't have to get into bed with OpenAI. StackOverflow was a beacon of resistance, but I guess their principles were for sale after all.

https://www.tomshardware.com/tech-industry/artificial-intelligence/stack-overflow-bans-users-en-masse-for-rebelling-against-openai-partnership-users-banned-for-deleting-answers-to-prevent-them-being-used-to-train-chatgpt

emilymbender, to random
@emilymbender@dair-community.social avatar

Mystery AI Hype Theater 3000 Episode 31: Science is Human Endeavor in which Drs. Molly Crockett and Lisa Messeri join me & @alex to explore why the dream of artificial scientists is not only impossible but actually a nightmare.

https://www.buzzsprout.com/2126417/15020029-episode-31-science-is-a-human-endeavor-feat-molly-crockett-and-lisa-messeri-april-15-2024

Thx to Christie Taylor for production!

emilymbender,
@emilymbender@dair-community.social avatar

Also available as video on PeerTube!

https://peertube.dair-institute.org/w/nrjPApZ1pvhh59Ja31wP6p

emilymbender, to random
@emilymbender@dair-community.social avatar

Latest Mystery AI Hype Theater 3000 newsletter, on AI hype infecting a key player in the pharmaceutical sector.

Wouldn't it be great if the FDA were as on top of the impacts of AI hype as the FTC?

https://buttondown.email/maiht3k/archive/wheres-the-vaccine-against-ai-hype/

emilymbender,
@emilymbender@dair-community.social avatar

@zenkat Yes -- as I understand it, AlphaFold is supervised machine learning over a very carefully curated dataset. Nothing out of OpenAI suggests "carefully curated dataset" is anywhere in their methodology.

emilymbender,
@emilymbender@dair-community.social avatar

@zenkat And thanks for that link! A good article indeed.

emilymbender,
@emilymbender@dair-community.social avatar

@zenkat In that case, calling this a "tailored version of OpenAI’s ChatGPT" would be outright lying.

emilymbender, to random
@emilymbender@dair-community.social avatar

Bender & Shah 2024 is now officially published as open access at TWeb.

For everyone who is using or promoting LLM driven chatbots as information access systems: this one is for you

https://dl.acm.org/doi/10.1145/3649468

emilymbender, to random
@emilymbender@dair-community.social avatar

TFW you've selected and organized too many links for an all hell episode and it's still less than half of the backlog and yet more links are coming in....

Anyway, join us today!!

Ready for more Mystery AI Hype Theater 3000?

Climate change has reached AI Hell (frozen over last we checked in December) and now we're flooded with nonstop nonsense. Join me and @alex as we wade our way through on Monday April 29, noon Pacific

https://www.twitch.tv/dair_institute

emilymbender, to random
@emilymbender@dair-community.social avatar

Not just corporate capture, but TESCREAL corporate capture. Ugh.

https://www.wsj.com/tech/ai/openais-sam-altman-and-other-tech-leaders-to-serve-on-ai-safety-board-7dc47b78

emilymbender, to random
@emilymbender@dair-community.social avatar

Ready for more Mystery AI Hype Theater 3000?

Climate change has reached AI Hell (frozen over last we checked in December) and now we're flooded with nonstop nonsense. Join me and @alex as we wade our way through on Monday April 29, noon Pacific

https://www.twitch.tv/dair_institute

arjen, to random Dutch
@arjen@idf.social avatar

Envisioning Information Access Systems: What Makes for Good Tools and a Healthy Web?

Must read article in ACM Transactions on the Web on challenges in information access and whether LLM might play a role (or not!), by Chirag Shah and @emilymbender

"Information access is not merely an application to be solved by the so-called ‘AI’ techniques du jour. Rather, it is a key human activity, with impacts on both individuals and society."

Better design that right!!

https://dl.acm.org/doi/10.1145/3649468

emilymbender,
@emilymbender@dair-community.social avatar

@arjen thank you!

  • 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
    •