UlrichJunker

@UlrichJunker@fediscience.org

Scientist in Artificial Intelligence and the Decision Sciences. Study of advanced methods for problem solving, which covers topics such as preference handling, constraint solving, and explanation generation.

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

UlrichJunker, to random

, March 5, 1934 - March 27, 2024 | Kahneman-Treisman Center for Behavioral Science & Public Policy

https://behavioralpolicy.princeton.edu/news/DanielKahneman

simon, to random
@simon@simonwillison.net avatar

If you haven't been following the wild news out of OpenAI (short version: the board of the controlling non-profit fired CEO Sam Altman yesterday) this piece by @benjedwards is a really great summary of everything that's known so far https://arstechnica.com/information-technology/2023/11/report-sutskever-led-board-coup-at-openai-that-ousted-altman-over-ai-safety-concerns/

UlrichJunker,

@simon @benjedwards very interesting. A power struggle between scientists and entrepreneurs with the scientists winning? This is unusual since scientists and technologists rarely sit in those boards.

SheDrivesMobility, to random German
@SheDrivesMobility@norden.social avatar

deleted_by_author

    •
    UlrichJunker,

    @SheDrivesMobility na ja, welche Politiker könnte die Kohlelobby möglicherweise beeinflussen wollen? Ich denke, dass die Auswahl nicht sehr gross ist.

    SheDrivesMobility, to random German
    @SheDrivesMobility@norden.social avatar

    deleted_by_author

    •
    UlrichJunker,

    @SheDrivesMobility was kann man an dem Rasen bloss so toll finden?

    simon, to random
    @simon@simonwillison.net avatar

    Embeddings: What they are and why they matter

    https://simonwillison.net/2023/Oct/23/embeddings/

    I took my recent PyBay talk and turned it into the most comprehensive answer I could possibly provide to the question "What are embeddings?"

    UlrichJunker,

    @simon imho the term “spatial representation” sounds much better than “embeddings”. For example, spatial representations are used to study voting problems

    https://en.m.wikipedia.org/wiki/Spatial_model_of_voting

    texttheater, to random German
    @texttheater@mastodon.social avatar

    Abgefuckt, dass es im Englischen kein Wort für Wissenschaft gibt.

    UlrichJunker,

    @emilymbender @texttheater just found out that there is an English Wikipedia article for “Wissenschaft”, which includes a comparison with “science”.

    https://en.m.wikipedia.org/wiki/Wissenschaft

    jbigham, to random
    @jbigham@hci.social avatar

    degenerative ai

    UlrichJunker,

    @jbigham how much regression does it involve?

    upol, to ai
    @upol@hci.social avatar

    🧵 [1/n]

    A Fortune 500 company hired me as an expert consultant to help them find out why their employees were not trusting their Explainable AI (XAI) system. My solution that worked really upset the VP of Engineering.

    He was angry because my solution didn't involve any substantial algorithmic changes. During the presentation, he said, "So where are the model changes in all of this?"

    Me: actually none.

    Him: so what did you actually DO?

    UlrichJunker,

    @upol answering the right question is the true power!

    ct_bergstrom, to random

    I wanted to consolidate a few thoughts on google, misinformation, large language models, enshittification, and the fate of the web as we know it.

    It started when Carl Zimmer shared this remarkable example of Google being fooled by machine-generated bullshit online.

    UlrichJunker,

    @ct_bergstrom the desertification of the web! So we need to create an oasis of authentic content!

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

    Yeah, our software sucks, but the world is just gonna have to deal, you know? Because this kind of software sucks but we're still gonna keep putting out there.

    https://www.vice.com/en/article/88xdez/generative-ai-is-a-disaster-and-companies-dont-seem-to-really-care

    UlrichJunker,

    @emilymbender apparently, many people like to do stupid things … and don’t care that others have to deal with the mess they are creating !

    alex, to random

    Lying in bed brainworm: why didn't the Arabs also create NULL while also creating zero.

    Alternative formulation: what's the history of NULL values

    UlrichJunker,

    @alex before NULL there was NIL (not in list). Now I wonder what NULL is standing for …

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

    On 9/28, I had the opportunity to speak at a virtual roundtable convened by Congressman Bobby Scott on “AI in the Workplace: New Crisis or Longstanding Challenge?”. The roundtable was a closed meeting, but sharing our opening remarks is allowed. Mine are here:

    https://medium.com/@emilymenonbender/opening-remarks-on-ai-in-the-workplace-new-crisis-or-longstanding-challenge-eb81d1bee9f

    UlrichJunker,

    @emilymbender that’s an excellent way of presenting machine learning systems. I hope that the roundtable participants got a better understanding of the technology in turn.

    jaztrophysicist, to random French

    deleted_by_author

    •
    UlrichJunker,

    @BrKloeckner @jaztrophysicist donc la méritocratie se lit comme médiocratie? Enfin je comprends …

    SheDrivesMobility, to random German
    @SheDrivesMobility@norden.social avatar

    deleted_by_author

    •
    UlrichJunker,

    @SheDrivesMobility hier aktueller Blog von @garymarcus , einem der prominentesten Kritiker des aktuellen KI-Hypes https://garymarcus.substack.com/p/face-it-self-driving-cars-still-havent

    thomasfuchs, to random
    @thomasfuchs@hachyderm.io avatar

    Started to mute and block the "AI can code"-stans and "LLMs-can-program-like-humans" enthusiasts because it's like talking to a wall.

    The response to "LLMs is using statistics to autocomplete code that may or may not work, they don't actually understand or can reason about problems (the number one thing programmers do)" is always "but the next version will fix this". Since like 3 versions ago.

    It's a category error to assume AI will ever be able to write complex software.

    UlrichJunker,

    @thomasfuchs I would not completely exclude that certain kind of systems will be able to write software, but the path to this will be long and it will not be opaque . There are so many things that we are just starting to understand.

    AndyPerfors, to random

    One of the things I love about being old is that work conferences get more and more purely fun with age. It's a combination of multiple factors, I think:

    1. The biggest by far is just that you know so many more people, and many of them are old friends (or at least friendly acquaintances) that you haven't spoken to in a while. So you're discussing science and exchanging ideas -- which itself is super fun -- but also just catching up, and relaxed and calm because they are just people and not scary famous dudes you need to impress.

    2. Relatedly, there's so much less anxiety. I don't feel like I need to prove myself, Some of that feeling I had when I was younger was self-imposed (nobody was judging me as much as I was judging myself) but some of it was a real reflection of the fact that I was a young scientist, sometimes on the job market, but even if I had a job still needing to earn a reputation etc. Now -- well, I have enough reputation that I can get the ears of interesting, smart people, and do the work I want to do, and not have to worry about providing for my family -- and those things are all I really care about. There's so much less pressure.

    3. Having been around for decades gives one a global sense of the field and where it's going, and it means I enjoy conferences on two levels at once: learning about the specific findings, but also in seeing the shape of what we're learning and where we've come from. It is humbling and rewarding to feel a part of this bigger thing, and makes me actually like humanity instead of despair for it :)

    UlrichJunker,

    @AndyPerfors @the_heruman what do you mean by “old”? I thought scientists don’t turn old (except for their bodies may be)

    janellecshane, to random
    @janellecshane@wandering.shop avatar

    Training large language models on the outputs of previous large language models leads to degraded results. Less diversity, more bias, and …jackrabbits?

    https://www.scientificamerican.com/article/ai-generated-data-can-poison-future-ai-models/

    #IsThisTheSingularity

    UlrichJunker,

    @janellecshane is this a way of getting the monster to eat itself? So it will vanish in a cloud of meaninglessness in the end?

    mmitchell_ai, to random
    @mmitchell_ai@mastodon.social avatar

    Guess what country I'm in!
    I wish the implicit assumptions & priority status of this sign were laughable. =/

    UlrichJunker,

    @mmitchell_ai in the same country in which you find signs like this.

    the_heruman, to random

    A few days ago I suggested that a conference be hybrid, but it seems that nobody supports these things anymore...

    I understand that it is technically complex (well, for us a little less, since we are computer scientists), but I feel that this is the only ethical way to do things. Hybrid events are good for climate change, they allow people with little budget or health problems to participate, etc.

    It's a shame there are virtually no hybrid events anymore, not even among the top conferences 😞

    UlrichJunker,

    @FMarquardtGroup @the_heruman when I ran a workshop at a virtual conference in 2020, I bought a good microphone. Good sound really is important and I don’t regret of having spent some money on this. It’s a real game changer for .

    UlrichJunker,

    @HannahCelsius @bookmarchitect @StephZihms @gpollara @FMarquardtGroup @the_heruman @academicchatter but posters are not necessarily a great experience either. Would be better to a have web conference where people can ask questions (and the poster online).

    lysander07, to ai

    The 2nd #AI winter had several causes: the collapse of the LISP machine market, a Slowdown in deployment of expert systems, the end of the Japan's Fifth Generation project, and DARPA's subsequent cutback of Strategic Computing Initiative...
    Slides: https://drive.google.com/file/d/1LUOA-NiE4nJ4sn5elbXUN4c7YDtK9uXB/view?usp=drive_link
    #artificialintelligence #aiwinter #LISP #machinelearning #lecture @fizise @enorouzi #aiart #stablediffusionart #creativeAI

    UlrichJunker,

    @lysander07 @amszmidt @enorouzi @fizise the stop of funding came quite abruptly. For example, Darpa stopped all projects using in the begin of the 1990ies. Many AI projects have been financed due to exaggerated promises and industry expected to get some magic out of this, whereas AI researchers have been more interested in scientific questions and in establishing a scientific field. There was a clear gap in expectations!

    futurebird, to random
    @futurebird@sauropods.win avatar

    ~one billion
    ~1,000,000,000

    That's how many parking spaces there are in the USA. For every car? About 4 empty spaces... just sitting there, not absorbing the rain, making flooding worse, making cities hotter as they bake in the summer sun.

    Our built environment and laws bend over backwards to make driving the only viable transportation option in nearly every imaginable context.

    You need to pay for healthcare, but almost never parking.

    (Pointing this out makes libertarian heads explode.)

    UlrichJunker,

    @katja_diehl

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

    Important reporting from @daveyalba at Bloomberg on the working conditions of the people hired to be "trainers" for Google Bard.

    >>

    https://www.bloomberg.com/news/articles/2023-07-12/google-s-ai-chatbot-is-trained-by-humans-who-say-they-re-overworked-underpaid-and-frustrated?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTY4OTE2NDQ1NiwiZXhwIjoxNjg5NzY5MjU2LCJhcnRpY2xlSWQiOiJSWE9INERUMVVNMFcwMSIsImJjb25uZWN0SWQiOiJFNzAxNENGQzIzNTI0MzU0QTVENUY2QkREMDAxOEU3NiJ9.CJRveZ4G3hcFeuImGivhVDZZJjz5g7W1G8Ihrehq6x8

    UlrichJunker,

    @emilymbender so it was never about truth, but about avoiding critical reactions from the users? Apparently, does not please everybody.

    tzimmer_history, to random
    @tzimmer_history@mastodon.social avatar

    A major German newspaper asked me to write an opinion piece on the Supreme Court - then refused to publish it because they regard my argument that the rightwing majority stands in direct opposition to multiracial, pluralistic democracy and the modern state as too extreme. 🤷

    UlrichJunker,

    @tzimmer_history most German newspapers are rightwing as well. So not very surprising.

    b0rk, to random
    @b0rk@jvns.ca avatar

    today I'm thinking about the "don't use floating point for money" advice I hear all the time. It obviously has a lot of truth to it.

    But -- Excel/Google Sheets uses floating point for all of its calculations, people use spreadsheets for money calculations all the time, and it generally seems to work just fine -- the results get rounded for display.

    So I'm trying to figure out if there's a more nuanced guideline than "never use floating point for money".

    UlrichJunker,

    @b0rk according to what I heard, Excel uses integral floating-point numbers plus a separate scale (number of positions by which the number needs to be shifted). So 0.33 would be represented as 33 and a scale of -2.

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