Tessera

Updated yesterday's Substack after news broke of Trump's teleprompter making $100k on Kalshi. This administration has certainly set a new low bar for self-dealing, but prediction markets are supercharging anonymous corruption at the highest levels of government. open.substack.com/pub/jonippol...

Screenshot from ABC News with photo of Trump at podium and headline "White House teleprompter operator made more than $100K betting on Trump's speeches: Sources"

Insider trading steering military decisions and politicians betting on their own campaigns. Prediction markets could be the most dangerous new technology since ChatGPT—especially for this administration. My new Substack at jonippolito.substack.com #Elections #PredictionMarkets #IranWar

Screenshot of a Substack post titled “Corruption gets its own platform” by Jon Ippolito, dated July 15, 2026. Below the headline is a photo of a roulette wheel surrounded by stacks of multicolored casino chips, visually linking gambling with the article’s discussion of prediction markets and corruption.

Librarians wondering how AI affects media literacy/acquisition policies/patrons' needs should consider joining UMaine's Digital Curation program, which tackles AI's impact on collecting institutions. I'm quoted in this related story from local NBC affiliate www.newscentermaine.com/article/news...

News anchor in a newsroom in front of a "News Center Maine" sign, with an inset image showing a person holding an open book beside library shelves.

Jon Ippolito @jonippolito.net · last mo.

One of the biggest election stories no one's talking about: the US Democratic primaries are becoming a proxy war between rival AI factions, with candidates tuning their strategies accordingly. My latest Substack digs into how that happened. open.substack.com/pub/jonippol...

The AI backlash is already on the ballot

TL;DR: For US politicians running in 2026, embracing AI has become a liability.

open.substack.com

If we don't get this right, we will have completely fumbled AI's arrival in the classroom. Quote from MYFest26 webinar co-led with @leaton01.bsky.social and organized by @mahabali.bsky.social and co. myfest.equityunbound.org/events/struc...

Quote: "Students do not experience learning separately from belonging" — Rhoan Garnett

Jon Ippolito @jonippolito.net · last mo.

Librarian, archivist, or curator looking to level up your digital game? You might be interested in our online courses that teach #DigitalCuration skills. Cherrypick your favorite or take the full grad certificate digitalcuration.umaine.edu #GLAM #libraries #museums #archives #data

Infographic titled "Digital Curation skills" showing interlocking gears labeled with digital curation tasks, including choose metadata, create database, digitize analog media, put exhibition online, build website, visualize data, copyright, backup data, migrate files, archive documents, 3D scan, save source files, cloud, RAID, Omeka, and emulation.

Jon Ippolito @jonippolito.net · last mo.

Interviewed for this "Spotlight on America" story about data center pushback, I said towns should compare what they're promised with the actual effects reported by communities that have had to live with these things mynbc15.com/news/spotlig...

Graphic stating that 14% of Americans would support a data center being built in their community, with a Reuters/Ipsos poll citation and a small interview inset.

Chatbots and AI summaries are making access to content increasingly personalized. But what if AI rewrites websites themselves to match your biases? My latest artwork explores what that shift could mean for public knowledge and cultural memory blog.still-water.net/what-you-thi...

Tablet centered in a softly lit blue-and-peach corner displays the text “(I do not think it means) what you think it means,” while shadows of a flower and a dagger appear on opposite walls, suggesting meanings beyond the words on the screen.

Another way to see data centers aren’t just about AI: even by EPRI’s most generous forecast, AI accounts for less than half of the load added since ChatGPT. You may not like what the rest of the buildout is being used for. blog.still-water.net/what-if-data...

Pie chart titled “Loads on data centers built since ChatGPT.” It shows AI workloads accounting for an estimated 35% (conservative) to 44% (liberal) of load, while non-AI workloads account for 56–65%. The chart emphasizes that a majority of load on recently built data centers is still non-AI, despite AI representing a substantial share. Source: EPRI estimates (2024, 2026); calculations by Jon Ippolito.

Guggenheim Collection@guggenheim.bsky.social · 2mo ago

I remember the day a sharp-eyed art handler and I discovered that unlabeled piece amidst correspondence in the bowels of the Guggenheim. I'd never seen a Ramirez like it...led to an exhibition of hitherto unknown coloristic drawings by this outsider artist.

Untitled by Martin Ramirez, 1954 https://botfrens.com/collections/212/contents/137290

Solomon R. Guggenheim Museum, New York
Gift of the Estate of Martin Ramirez, 2008

© 2016 Estate of Martin Ramirez / Courtesy Ricco/Maresca Gallery

If AI accounts for only 20% of data center use, why is it the justification for nearly every new data center? I did the math and suspect Big Tech is using AI as a cover for unpopular uses like surveillance advertising. blog.still-water.net/what-if-data...

A chart showing a log-log scatter plot that shows big tech is spending far more on AI than it earns from it. The x-axis shows annual data-center and compute spend in billions of dollars on a logarithmic scale from $25B to $300B. The y-axis shows AI revenue run-rate in billions of dollars on a logarithmic scale from $0.35B to $320B. Seven companies are plotted as purple dots. xAI sits at $31B spend and $0.5B revenue, the lowest revenue point. Oracle is at $60B spend and $12B revenue. OpenAI is at $120B spend and $25B revenue. Meta is at $135B spend and $0.5B revenue, far below all others at its spending level. Alphabet is at $185B spend and $30B revenue. Microsoft is at $190B spend and $37B revenue, the highest revenue point. Amazon is at $200B spend and $15B revenue. A green dashed diagonal line marks break-even where revenue equals spend; all companies sit far below it. A red dashed diagonal line marks 10% revenue recovery; most companies cluster near or below it. Vertical error bars on Oracle, Amazon, Alphabet, and Microsoft show the estimated range between reported AI revenue and a conservative first-party-products-only floor, with Oracle having the longest bar, dropping from $12B to approximately $1.5B.

Need to put your collection in a database? On the web? There's a course for that! Join us online this fall for Digital Collections & Exhibitions, part of UMaine's all-online Digital Curation program digitalcuration.umaine.edu #DigitalCuration #libraries #collections #archives #museums #GLAM

Prompt engineering is a moving target. Here's my rundown on how GPT-5.5 and Claude 4.7 reward clarified output over magical incantations, and what it means for AI literacy www.linkedin.com/posts/jonipp...

The latest OpenAI and Anthropic models made “best practices” prompting techniques obsolete. What does this mean for AI literacy? Claims of downgraded performance with GPT-5.5 and Claude 4.7 last… | J...

linkedin.com

Gallup finds Americans would rather live near a nuclear plant than an AI data center. This resistance seems largely independent of AI disapproval. And respondents didn't even cite less publicized hazards, like e-waste/noise/exhaust/job ephemerality www.linkedin.com/posts/jonipp...

#datacenter #nuclearpower #environment #aiethics | Jon Ippolito

Americans would rather have a nuclear plant in their backyard than a data center. A Gallup poll conducted in March out this week found 71% of Americans oppose the construction of new data centers in ...

linkedin.com

anil oza @aniloza.bsky.social · 3mo ago

At today's UMA Faculty Institute @leaton01.bsky.social urges faculty to steer higher Ed through the AI disruption rather than sit back and watch

wow — the preprint host, arxiv, is banning authors for a year if they submit papers with hallucinated citations 🤖

Twitter thread from Thomas Dietterich, reading "Attention
@arxiv
authors: Our Code of Conduct states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated. 1/
3:03 PM · May 14, 2026
·
66.7K
Views
Relevant
View quotes

Thomas G. Dietterich
@tdietterich
·
1h
If generative AI tools generate inappropriate language, plagiarized content, biased content, errors, mistakes, incorrect references, or misleading content, and that output is included in scientific works, it is the responsibility of the author(s). 2/
Thomas G. Dietterich
@tdietterich
·
1h
We have recently clarified our penalties for this. If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can't trust anything in the paper. 3/
Thomas G. Dietterich
@tdietterich
·
1h
The penalty is a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted at a reputable peer-reviewed venue. 4/
Thomas G. Dietterich
@tdietterich
·
1h
Examples of incontrovertible evidence: hallucinated references, meta-comments from the LLM ("here is a 200 word summary; would you like me to make any changes?"; "the data in this table is illustrative, fill it in with the real numbers from your experiments") end/"

Helen Branswell 🇨🇦@helenbranswell.bsky.social · 3mo ago

I find this unsettling: A new study in #Lancet shows that the number of faux citations in scientific articles is on the rise, and #AI is likely to blame. @aniloza.bsky.social reports. www.statnews.com/2026/05/07/l...

UMaine's summer Digital Curation classes are full but we have space in this fall's online course DIG 540 (Digital Collections and Exhibitions), aimed to help everyone from labs to libraries organize their material and put it online digitalcuration.umaine.edu

Word cloud for DIG 540 course, including "collection", "database", "web", etc.

This week GitHub starts training on your code by default. For me, code reuse is probably the most ethical AI use case. But turn off training if you're worried about others using your code unethically: Settings > Copilot > Features > Privacy

Turn off GitHub’s default training on your code via Settings > Copilot > Features > Privacy

A study of websites created since 2022 found 1 in 3 are AI-generated text. And their characteristics follow my prediction that AI will act as compression algorithm on cultural heritage. www.linkedin.com/posts/jonipp...

#aiethics #generativeai #digitalcuration #archives #ailiteracy #libraries | Jon Ippolito

A study of websites created since 2022 found 1 in 3 are AI-generated text. And their characteristics follow my model of AI as compression algorithm. Researchers from Imperial College London and Stanf...

linkedin.com

A dozen states have tried so far, but Maine is now on the verge of becoming the first in the US with a state-wide data center ban. www.linkedin.com/posts/jonipp... #MEpolitics #AIethics #DataCenter #Environment #Climate #Sustainability

#mepolitics #aiethics #datacenter #environment #climate #sustainability | Jon Ippolito

Maine is on the verge of passing the first state-wide ban on data centers in the US. Similar measures have failed in at least a dozen states this year, so Maine’s bill is poised to set a national pre...

linkedin.com

Ben Brumfield @benwbrum.bsky.social · 4mo ago

Curious about AI’s real environmental impact? In this recorded webinar we debunk overblown claims from both sides, share new comparisons, and demo a tool that tracks your AI footprint in real time. digitalcuration.umaine.edu/teleconferen...

Faces of Ben Brumfield, Jon Ippolito, Greg Nelson, and Ethan Morin on screen during the teleconference

In just under two hours, we're hosting @jonippolito.net in a webinar presenting his research on the environmental impact of #AI use compared with other common digital technologies. Register for "How Green is your Prompt" webinar here: content.fromthepage.com/april-2026-w...

How Green Is Your Prompt? (April 2, 2026) - FromThePage Blog

content.fromthepage.com

At 6:50am last Monday, markets spiked despite no news on the war. 15 minutes later, the US president announced he's reversing course to de-escalate. Someone's cashing in on vacillation over the war, at a time when Trump's circle is hobbling the SEC blog.still-water.net/its-650am-do...

Suspicious trades before Iran war announcements on the West Texas Intermediate oil futures exchange, 23 March 2026. Source: CNBC.