Posts by scott_bot@hcommons.social
(DIR) Post #ASMxWrmEx0fjQIbfn6 by scott_bot@hcommons.social
2023-02-04T12:17:56Z
0 likes, 0 repeats
Judge uses ChatGPT to speed up ruling over whether an autistic minor is on the hook for therapy fees.I'll be honest, as a kid I didn't predict Douglas Adams' Infinite Improbability Drive to be the closest model of the future's tolerance for chaos technologies.https://www.vice.com/en/article/k7bdmv/judge-used-chatgpt-to-make-court-decision
(DIR) Post #ASMxWsUYIG7LdjL3rM by scott_bot@hcommons.social
2023-02-05T12:02:04Z
0 likes, 0 repeats
⬆️ this is my third toot to receive a bunch of boosts and ⭐s, and I've got to say, the experience is still more pleasant here than on the bird site. Way less vitriol and unexpected 🤨-inducing replies.
(DIR) Post #ASaC8JOjHCLsLMjHUW by scott_bot@hcommons.social
2023-02-11T21:25:04Z
0 likes, 0 repeats
@zoeleblanc @TedUnderwood dunno, let's ask jeeves
(DIR) Post #ASdb4vDxmGsgpW3OEq by scott_bot@hcommons.social
2023-02-13T12:26:46Z
0 likes, 0 repeats
Inspired by my previous two boosts, on "adversarial collaboration" (https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0268577) and "adversarial review" (https://themarkup.org/hello-world/2023/02/04/journalistic-lessons-for-the-algorithmic-age), I'll share my preferred methodology for hypothesis-driven research, which I suppose can be called "adversarial triangulation":When I have a strong hunch, I design a hypothesis that is maximally likely to refute it. If it doesn't, I do the same thing again with a different angle of observation and a different method. If it still hasn't been refuted, I do the same thing *again* before I'm finally convinced. If any of it involves Bayesian stats, I use an informed-but-unfriendly prior.If you want to be sure your pre-conceptions don't inadvertently tug your results, you could do worse than stacking the deck against your beliefs while using three different approaches that rely on different types of observations.
(DIR) Post #ASdb4vxL3ZB36FHcxs by scott_bot@hcommons.social
2023-02-13T12:36:08Z
0 likes, 0 repeats
⬆️ This is a sure-fire approach to publish rarely and well, which is bad career advice and won't get you tenure in the fields this approach is most relevant. Which is a problem.If we're to believe the recently-controversial John Ioannidis that "most published research findings are false" (and two decades of evidence suggests we should believe him, in this limited capacity), that includes stuff you and I want to publish.
(DIR) Post #ASdb4x3kxBuAWRIGbg by scott_bot@hcommons.social
2023-02-13T13:01:22Z
0 likes, 0 repeats
(now wearing my humanities hat)My humanities colleagues understandably won't see much relevance in these toots, but we'd also benefit from more self-adversarial approaches when knowledge-making. Our hermeneutic processes aren't exactly immune from confirmation bias, even if we're usually more open about our situatedness.
(DIR) Post #ASntqtYf16TwJmMJqi by scott_bot@hcommons.social
2023-02-16T23:04:58Z
0 likes, 1 repeats
Student internship alert! Share widely.My brand-new National Endowment for the Humanities (NEH) initiative, the Office of Data and Evaluation, is hiring summer interns. We got in late in the game, so it's not in the advert.The office helps ensure the equitable distribution, reach, and impact of NEH programs, and we seek advanced undergraduate, master’s, or early-stage PhD students to help us reach this goal. We can offer $15.62/hour for 32 hours per week and in-depth professional development opportunities. 100% telework.Apply by 2/23/2023! **VERY IMPORTANT: make sure to mention the Office of Data and Evaluation in your cover letter**https://www.usajobs.gov/job/702481700Reach out to me (sweingart@neh.gov) with any questions.
(DIR) Post #ATo43u15FX31BzJ6cS by scott_bot@hcommons.social
2023-03-20T12:10:08Z
0 likes, 1 repeats
Prompted by @TedUnderwood's call to respond to what AI might look like in 3-5 years, @emckean & Vivek Seshadri's ideas of a data co-op, & @ct_bergstrom's discussion of LLM's propensity for bullshit, I've been reflecting on one possible future for large language models.Let's assume from the current state of text-to-image models that LLMs will keep improving. By improving, I mean at meeting benchmarks and passing for human, not at being better for society. The bullshit will become more difficult to identify.It's a problem for a million oft-discussed reasons. One of them is false information presented confidently through an AI assistant. The training set (the internet) is awash with enough false information that one of these models will eventually give a parent advice that will kill their kid.As they say: garbage in, garbage out. Companies may respond by doubling down on what Meta did with Galactica, training models on scientific articles and other trusted sources.1/n
(DIR) Post #ATo445j5iFmlVeyuEy by scott_bot@hcommons.social
2023-03-20T12:12:22Z
0 likes, 0 repeats
Galactica failed because LLM tech wasn't quite there yet, still making stuff up unwittingly, but I expect that will be fixed. The next Galactica will fail because because even if it doesn't outright make stuff up, the sum of scholarly & reference sources has its own issues.A model trained on scientific articles will lie succinctly, and one trained on humanities will bullshit like a high school debate champion; both will do so with the whiff of authority. Encyclopedias and other trusted sources will come with their own flavors of problems.As one example of many (pushing aside Big Stuff like geographic skew): science disagrees, science changes. Scientific consensus now is different from thirty years ago. A model trained on All Of It would lead us astray (see COVID and wrong ideas about airborne transmission).2/n
(DIR) Post #AXDj27pLxMoliKCxnM by scott_bot@hcommons.social
2023-06-30T11:19:13Z
0 likes, 1 repeats
Please indulge my two complaints about this week's science journalism:1. Every time I heard a journalist bring up the gravitational wave news, it was framed as: "Wow this is bending my mind! Dizzying! Let's bring in our two nerds to discuss it further."I guarantee you gravitational waves are less complex than, say, the Israeli/Palestinian situation. The performative blown minds aren't helping make people feel nature is normal and accessible.2. Dear journalists, the air quality isn't "purple" or "maroon," it's "very unhealthy" or "hazardous." There are actual recommended words and phrases to describe the severity and health impacts of a high Air Quality Index. Use them. The visualization isn't the thing; the bad air is the thing.I've just seen the sixth reference in the last 24 hours to the color of the AQI map. Tell people it's the equivalent to smoking half a pack of cigarettes, tell people with asthma to limit outdoor time and wear an n95.Thanks for coming to my TED Talk.
(DIR) Post #AXMjgd2P5JC0W9GVEG by scott_bot@hcommons.social
2023-07-04T13:32:23Z
0 likes, 1 repeats
Next week I'm posting a job ad ($100k-$180k) for the most impactful data scientist to the humanities in the US. Federal role, can shape funding landscape, public understanding of the humanities, etc. It's an opportunity to make a real difference.With this latest diasporic mastodon wave, and given the importance of the role and unique skills required, a humble request: please share this widely & urge Data/Viz/Stats/Humanities/SocSci folks to follow my account.The role requires a bit of a unicorn, and I need to reach as wide a network as possible to fill it.
(DIR) Post #AXMjgeP80IxqkuuAHw by scott_bot@hcommons.social
2023-07-04T13:37:10Z
0 likes, 0 repeats
The actual post will be out in the next week or so, and I'll make a similar request then. Thank you, friends!The job opening window will be somewhat brief, and I want to make sure everyone who could be interested has the chance to see it.
(DIR) Post #AaPOObcc6g7VOEGLrs by scott_bot@hcommons.social
2023-09-29T01:19:02Z
1 likes, 0 repeats
I don't know who needs to hear this but [the disconcerting silence you notice when a low machine hum you didn't notice suddenly goes quiet]
(DIR) Post #AaPOOd4ehu93tUOGDQ by scott_bot@hcommons.social
2023-09-29T11:08:19Z
0 likes, 0 repeats
☝️"‘Is there anybody there?’ said the Traveller, Knocking on the moonlit door;[...]But only a host of phantom listeners That dwelt in the lone house thenStood listening in the quiet of the moonlight To that voice from the world of men:Stood thronging the faint moonbeams on the dark stair, That goes down to the empty hall,Hearkening in an air stirred and shaken By the lonely Traveller’s call.And he felt in his heart their strangeness, Their stillness answering his cry,While his horse moved, cropping the dark turf, 'Neath the starred and leafy sky;[...]‘Tell them I came, and no one answered, That I kept my word,’ he said.Never the least stir made the listeners, Though every word he spakeFell echoing through the shadowiness of the still house From the one man left awake:Ay, they heard his foot upon the stirrup, And the sound of iron on stone,And how the silence surged softly backward, When the plunging hoofs were gone."--Walter de la Mare
(DIR) Post #Ast6XAox5UcNi3Qb7Q by scott_bot@hcommons.social
2025-04-08T13:01:49Z
0 likes, 1 repeats
DOGE fired me. Probably. Last Friday, the National Endowment for the Humanities put the majority of its staff on administrative leave. That's been the precursor to mass layoffs at other agencies, and I have no reason to expect our situation will turn out differently.This was the most fulfilling and important job of my career, and I'd hoped to retire there. It was an honor to serve my country alongside colleagues with such integrity, care, and wisdom. I don't know where my next adventure will be, and I'd appreciate your help figuring that out.At NEH, I was the Chief Data Officer and founded its Office of Data and Evaluation. Before that, I led initiatives at universities and libraries. For my next role, I'd like to help some nonprofit, philanthropy, university, or mission-driven organization make the world a little better.If you know anyone hiring a senior leader who cares deeply and broadly, please have them reach out to weingart.scott+future@gmail.com.