@jonippolito@digipres.club avatar



UMaine Digital Curation https://DigitalCuration.UMaine.edu • Re-collection http://re-collection.net • Variable Media Network http://variablemedia.net/e • Full-text search via tootfinder

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

jonippolito, to random
@jonippolito@digipres.club avatar

3 ways to search Google without being told to eat rocks or glue cheese on your pizza:

  • Add "udm=14" to your search string. It just adds an instruction on the end of the URL that tells Google you want web rather than AI results, but you'll feel like a hacker.

  • Use https://udm14.com, which will always add the udm14 cheat code.

  • Use StartPage https://www.startpage.com or TenBlueLinks https://tenbluelinks.org, which query Google without AI or tracking of any kind.

jonippolito, to journalism
@jonippolito@digipres.club avatar

What jobs are we preparing students for by boosting their writing productivity with AI? After shedding 40% of its workforce, the gaming site Gamurs posted an ad last June for an editor to write 250 articles per week. That’s a new article every 10 minutes, at $4.25 per article.

As @novomancy has noted, AI is only the accomplice here. This clickbait nightmare is the logical conclusion of the ad-supported web.


molly0xfff, to web
@molly0xfff@hachyderm.io avatar

If you've ever found yourself missing the "good old days" of the , what is it that you miss? (Interpret "it" broadly: specific websites? types of activities? feelings? etc.) And approximately when were those good old days?

No wrong answers — I'm working on an article and wanted to get some outside thoughts.

@jonippolito@digipres.club avatar

@molly0xfff Personally I miss the anarchic days of 1990s net art. However, for people with shorter Internet memories, I'd point to RSS (1999) as a democratic protocol that let anyone be a publisher. While the shuttering of Google Reader (2013) marked the encroachment of publishing monopolies online, the benefits of Really Simple Syndication persist to this day—for example making it easy for anyone to distribute a podcast on multiple platforms.

jonippolito, to random
@jonippolito@digipres.club avatar

This demo of a tool that can compare the output of a dozen AI models at once results in surprisingly common answers to prompts like "tell me a joke" or "give me a random number." When an average is well established, LLM results are reliable—and uncreative https://youtu.be/C3H4LKmcw1M?si=5tT7-fCdIobLQw79

jonippolito, to generativeAI
@jonippolito@digipres.club avatar

Google's Education VP wants us to believe AI is the classroom's new calculator, but this is a terrible analogy:

  1. We know how calculators produce their results.
  2. You can check a calculator's answer using pretty much the same algorithm it uses.
  3. Rare floating point errors aside, calculators do not invent false answers.
  4. Calculators are based on math principles; LLMs are based on no principles.


jonippolito, to Cybersecurity
@jonippolito@digipres.club avatar

A cybersecurity researcher finds that 20% of software packages recommended by GPT-4 are fake, so he builds one that 15,000 code bases already depend on, to prevent some hacker from writing a malware version.

Disaster averted in this case, but there aren't enough fingers to plug all the AI-generated holes 😬


jonippolito, to random
@jonippolito@digipres.club avatar

I'm glad a few patients benefit, but saying a neural link "reads your mind" is ontologically equivalent to saying that ChatGPT "understands how an air fryer works." These shorthand phrases imply some level of software / wetware engineering that just hasn't been invented yet 3/3

jonippolito, to Futurology
@jonippolito@digipres.club avatar

As impressive as Musk's solitary NeuraLink demo is, beneath the hype lies a misperception with a disturbing parallel to large language models. Both operate on statistical inference rather than scientific models of the brain, and garner attention for cherrypicked successes 1/3

jonippolito, to generativeAI
@jonippolito@digipres.club avatar

AI companies to universities: Personalized tutors will make you obsolete

Also AI companies: Thanks for recording your lectures so we can sell them on the open market to train personalized tutors


#Data #HigherEducation #OnlineLearning #AIethics #AIinEducation #GenerativeAI #LLM

jonippolito, to llm
@jonippolito@digipres.club avatar

"Aftermarket" fixes applied after training, like injecting diversity terms into prompts, don't fix the underlying model and can even exacerbate harmful fabrications. If the training set is biased—and the Internet is—it's really hard to correct that after the fact.


jonippolito, to DigitalArt
@jonippolito@digipres.club avatar

This exquisite collaboration between artist Auriea Harvey and curator Regina Harsanyi is a gift to anyone who wants to learn more about digital art. Unwrapping it reveals virtuoso lessons in Internet art, emulation, 3-D scanning, indie video games, and augmented reality


jonippolito, to random
@jonippolito@digipres.club avatar

US copyright law doesn't allow enough leeway for emulation, so I joined Harvard Law School, Software Preservation Network, and Library Copyright Alliance in arguing for a DMCA exemption to expand access to copyrighted tech for education and preservation.


#Copyright #Libraries #Museums #ComputerHistory #Digipres #DigitalPreservation #MediaArcheology

jonippolito, to Emulation
@jonippolito@digipres.club avatar

@euanc 's research shows why we need preservation alternatives to migration like emulation. Look at how something as important as exponents in equations changes in various environments. Imagine if this was the formula for a drug or a flight plan to the Moon!


jonippolito, to ChatGPT
@jonippolito@digipres.club avatar

AI metaphors lurk even in seemingly innocuous statements like "chatbots don't understand what isn't programmed into their datasets."

Words like "understand" and "programmed" don't reflect what's going on under the hood with large language models.

OK, we can't all say "an overfit embedding yields misleading centroids," but we all do need to recognize metaphor's power and perils. This paper is a good start: https://arxiv.org/abs/2401.08711


jonippolito, to ukteachers
@jonippolito@digipres.club avatar

I'm proposing that all educators confronting AI—even writing teachers—ask students to generate an image. Unlike ChatGPT, which comes off as some kind of robot oracle, text-to-image generators show AI capabilities and limits in vivid color 🧵 1/4


@jonippolito@digipres.club avatar

Unlike ChatGPT's simple text field, an interface like Leonardo.ai pulls back the curtain on LLM machinery. Let students choose the temperature (.3 or .8?) or model (PhotoReal or Pastel Anime Dream?) and see the complexity behind the black box in action 3/4

@jonippolito@digipres.club avatar

AI's error-prone nature is blatant in image generation. Students may not know Robespierre from Richelieu but they know neither had 13 fingers. Getting multiple outcomes by default and "rerolling" for a new set helps elucidate AI's probabilistic nature 2/4

@jonippolito@digipres.club avatar

Images also illustrate the critical concept of AI averaging. Use it to explain the Pope in Balenciaga. Notice how attractive everyone is, then explore biases in training data as well as the more subtle "midpoint hottie" problem. Way easier to see this bias in images than text 4/4

jonippolito, to design
@jonippolito@digipres.club avatar

UMaine New Media and Computer Science just finished a groundbreaking 4-month experiment comparing AI with traditional digital methods in creative tasks. From essay writing to game design, 50 students tested AI's capabilities head-to-head with conventional tools https://blog.still-water.net/ai-versus-old-school-creativity 🧵1/7

jonippolito, to Futurology
@jonippolito@digipres.club avatar

When I first introduced DNA as a storage medium to my students in 2014, it seemed like a radical idea. It's still radical, but no longer just an idea.

We look at the pros and cons of all storage media in my online course starting 17 January (via https://DigitalCuration.UMaine.edu)

"Want to Store a Message in DNA? That’ll Be $1,000"


jonippolito, to random
@jonippolito@digipres.club avatar

From the What a Surprise Department:

"Judge: Amazon “cannot claim shock” that bathroom spycams were used as advertised"


jonippolito, to ai
@jonippolito@digipres.club avatar

Harvard's metaLab has launched https://aipedagogy.org, a resource chock full of tasty assignments by trailblazers of generative AI in the classroom. (My own "AI Sandwich" is also on the menu.)

I've already stolen Juliana Castro's "Illustrate a Hoax" for my own class!

textfiles, to random
@textfiles@digipres.club avatar

Kind of defeats the purpose, really

@jonippolito@digipres.club avatar
jonippolito, to SelfDrivingCars
@jonippolito@digipres.club avatar

In a recent guest lecture I noted that self-driving cars are still a dicey proposition despite an investment of 30 years and $100bn. Now San Francisco has suspended Cruise operations after its robotaxi drove over and pinned a pedestrian hit by another car to the ground. I know Cruise gives their cars cutesy names, but did they really have to name this one "panini"? 😬


jonippolito, to generativeAI
@jonippolito@digipres.club avatar

Repeated calls by artists for transparency in this NYU/USC symposium ignore the fact that generative AI is opaque by nature. Generative AI exploits the most transparent digital medium of all time—"view source" is literally built into every web browser—to make a technology so opaque even even its creators don't understand how it works.


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