Joe Levy

They do know that “SI” is basically the metric system, right? … probably not actually.

I know I’m supposed to mostly muse about drones & permafrost, but like everyone, I have been thinking a lot about AI lately. What strikes me most, & this is not an observation uniquely made by me, is the perversity with which we have invested our resources—what machines we’ve chosen to build and why

Or do we want to build machines that support the time-consuming processes of life, that foster learning, that make things, or that give us the time we need carry out work that cannot be automated by an LLM or a fleet of agents? If AI can enable human flourishing then it’s a redeemable technology.

If AI undercuts our efforts to grow food for all people, to keep air and water clean, and to build and maintain infrastructure that supports human life and human development, then it is a hurtling locomotive, throwing off exhaust and danger, but for all that we feed it, not doing work that matters.

It takes time to build a relationship, and time to make up your mind. The frustrations of AI are that the dominion machine we thought we had cannot speed some things up. Meat-world is going to always be slower than information-world.

So, the big question, I think, is what do we want our machines to do for us? Do we want machines that turn fossil fuels into the illusion of status and the illusion of dominion?

But that failure to understand, even more than the failure to control, points to the bigger issue. The bigger problem that many folks have noticed, not least the late Wendell Berry, is that many things that matter to us do not move at the speed of information.

It takes time to grow an apple, time to figure out how to get music from a ukulele, time to build a bridge, time to learn (to really learn— to observe, analyze, evaluate, & apply—not just to flash symbols in a regurgitation of information, no matter how well organized it might be by machine).

And it is shocking how quickly these dopamine hits can be delivered. AI gives us what we want, right now, or dispatches an agent into the web to deliver on our demands as soon as it possibly can. AI is like magic for anything that can move at the speed of information.

The first AI problem is that the dominion AI seems to grant us is, apparently, a risky delusion at best. That’s what Dario Amodei, Anthropic’s chief executive, was getting at when he said, “Left unchecked, it could outrun our ability to understand and control these systems,”

They say yes with enthusiasm, take criticism unflinchingly. They strive to deliver. No chorus of Likes can compete with the allure of the focused attention of someone who cares just about you and strives to fulfill your wishes.

So we have built status machines, and wish-fulfillment machines—a machine that gives us the appearance of consequences-free dominion over an artist, a writer, an advisor, an executive assistant for everyone.

Social media was a magic status machine. It cost less than the time and effort of buying a round at the pub, and produced a similar chorus of engaged and supportive voices. Also, I hope these posts are interesting to you (I’m a hominid, too, and seek status).

AI chatbots are a similar machine, but even better. Type a few words, and you get what appears to be dominion over a thinking being. Need a ghost-writer, an analyst, a graphic designer? A lover, a tutor, a research assistant? They can be yours, right now, attentive and at your beck and call.

We all got used to the machine of social media, a contraption that turns words into social status, or at least into the illusion of status. Type a few lines, and wait for the likes and reposts to tumble in. To our hominid brains, there’s little better than the approval of our peers.

17 Sept

I, too, yearn to be hotter than a croc, cooler than a chicken. I mean, I know some pretty cool chickens.

13 Sept

Maybe I’m cynical, but I can’t see announcements from AI companies as anything but marketing. Not enough pep in the stock? Share news about super tools that hack email. Worried about that Oct IPO? Suggest you need regulation & then blame the threat of regulation for poor profits & underperformance

13 Sept

Maybe it’s not news. Maybe it’s manipulation to paper over poor business strategy and smaller than anticipated economic impact. But if I’m wrong, no shade to our new robot overlords. People are cynical by nature. www.theatlantic.com/technology/2...

4 Sept

Happy fieldwork Friday! My team came upon some impassable chasms out in the Alvord Desert and needed to turn back. Actually, these are just mud cracks. I've been playing with macro photos of soils and sediments to better think about micro landscapes. Still a work in progress, but lots of fun! ⚒️🧪

23 Aug

Paleoclimate peeps - come work with us at Colgate! apply.interfolio.com/188708 Review of applications starts Sept. 1.

colgate.eduFaculty Position: Assistant Professor in Paleoclimate | Colgate University

23 Aug

Even if you're not a person focused on climate over time, check out our "how to apply" page to get some insights into what selective liberal arts colleges (SLACs) might be looking for in faculty applications: www.colgate.edu/academics/de...

17 July

Happ #FieldworkFriday from the Alvord Desert playa in far southeastern Oregon! Hot work on the Alvord Hot Springs discharge plume during the day, followed by hanging out on the playa at sunset! ⚒️🧪

3 July

A little World Cup planetary science... If the Earth were the size of a soccer ball, 🌎 = ⚽ then the Moon would be the size of a tennis ball: 🌔 = 🎾 The Earth-Moon distance, 🌎 to 🌔, would be about 30 soccer balls: ⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽⚽ or a little less than the width of the goal 🥅

3 July

h/t Hamilton village trivia for getting to think about these numbers the other night during the US-BH game!

Love exploring climate change in the geological record? Want to teach about climate change with fabulous undergraduates? Yearning to advance your paleoclimate research in a supportive, scholarship-forward environment? Then come work with us at Colgate University! ⚒️🧪❄️🧪🌊🧪 apply.interfolio.com/188708

We seek an individual who is committed to excellence in teaching and research at the undergraduate level. The area of specialization is open but should include an examination of climatic change in the geologic record.

apply.interfolio.com Apply - Interfolio {{$ctrl.$state.data.pageTitle}} - Apply - Interfolio

Applications should be submitted through Interfolio at apply.interfolio.com/188708. Review of applications will begin September 1, 2026, and continue until the position is filled.

The Department of Earth and Environmental Geosciences at @colgate.edu invites applications for a tenure-stream position in the area of Paleoclimatology at the rank of Assistant Professor, beginning fall '27.

29 May

Road salt? In my drinking water? If you live in central NY, the answer might be yes, and a lot of it. Read all about @colgate.edu Earth & Environmental Geosciences student Morgan Cecil's findings here bit.ly/hamiltonsali... - fresh out in Case Studies in the Environment. ⚒️🧪💧🧪❄️🧪

22 May

Happy #FieldworkFriday! Exploring lava tubes from above and below at Tabernacle Hill volcano outside of Fillmore, UT. 🥾🌋⚒️🧪

19 May

You know, we don't see the "before" part of this scene, but I think Stickman was probably provoking the rocks on the North Kaibab trail before they turned on him ⚒️🧪

18 May

Today was a particularly effective day to teach about aeolian processes at Coral Pink Sand Dunes

nytimes.comOpinion | Why the A.I. Job Apocalypse (Probably) Won’t Happen

An interesting (& maybe not quite right) read from Ezra Klein about AI and knowledge work. I agree with him that good media literacy would treat breathless news reports about AI's power and pending replacement of work as ad copy for AI companies. www.nytimes.com/2026/05/03/o...

But some knowledge work is about low-probability/high-impact results: finding a turn of phrase that delights a reader, finding an obscure precedent that turns a legal case, seeing a similarity with a prior patient that leads to diagnosis, intuiting a hypothesis that leads to a new experiment.

These tasks seem more like labor to me-a human preoccupation with experience and novelty. And I think that's the real test of AI knowledge work-does it outpace our best human workers on meaningful low-probability/high-reward cases.

nytimes.comOpinion | Why the A.I. Job Apocalypse (Probably) Won’t Happen

EK is probably right that human contact and relationships will matter more in the future, as knowledge work is displaced to machine-made. Connections and relationships and stories tied to knowledge products are place where "human-made" might market like "homemade" does today.

But the other half of this might be about the product of knowledge work. Some tasks are about getting the most probable answer--AI is built for that. There are many cases where the most probable answer is the best.

nytimes.comOpinion | Why the A.I. Job Apocalypse (Probably) Won’t Happen

The crux is whether knowledge is a resource like coal, which capital (an engine or an AI) consumes to produce produce for profit, or whether knowledge is part of labor, a skill that is applied to do work. If knowledge is just data, then more productive use of data means more use of knowledge (data).

But if knowledge work really is information applied through skill, then AI is about replacing labor, & machines that can produce similar knowledge products will reduce demand for knowledge labor. Is knowledge produce more like denim fabric (better machine made) or like a sandwich (human-centric)?

nytimes.comOpinion | Why the A.I. Job Apocalypse (Probably) Won’t Happen

But he invokes Jevon's Paradox-favorite of geoscientists-which, based on coal as an example, says that the more efficiently a resource can be used, the more it is consumed (not the less). EK says that knowledge will be like coal-more efficient use means more demand for knowledge.

29 Apr

For all who are also deep in review of undergraduate student theses, I salute you. My students actually do get to practice Haiku Voice when we submit LPSC abstracts. The rest of the time, I push for Active Voice.

Sarah Whorley, Ph.D. 🦬⚔️@dralgaegirl.bsky.social

Stolen from FB....

22 Apr

Happy Earth Day! I'm giving a talk on campus, the gist of which is that geoscientists don’t just care about the Earth for the Earth’s sake—we care about it because it’s our home and people who matter to us live on it. ⚒️🧪 h/t NASA’s Suomi NPP satellite for this stunner.

21 Apr

Experiencing a little #moonjoy outside of the Ho Science Center @colgate.edu. A late evening on campus helping students put the finishing touches on their senior thesis presentations.

15 Apr

Kind of weird to have an LPSC2027 save the date with no location. Is it in Texas again? Will abstracts that suggest broadening participation in planetary science is a good thing be discouraged again? If the answer to both is yes, I think my response might be, "When is DPS again?"

12 Apr

New Commission on Fine Arts renderings just dropped for the National Participation Trophy, er, #DCarch. Blank on left has a nice big frieze you can write whatever you like on. I recommend Times New Roman (the boringest font) and the Photoshop bevel effect. #monuments #DC #Architecture

10 Apr

Happy little mud cracks from the Alvord Desert playa at the end of a cracked-up week. #drones

31 Mar

Spring is coming, change is coming. Audio on for peepers! 🐸

24 Mar

It feels like the subtext of recent reporting on long TSA lines is that it’s the Dem’s fault for holding firm on unmasking ICE agents and requiring judicial warrants for searches before agreeing to fund DHS. That reporting slant is just gross.

24 Mar

It’s an abdication of reporting responsibility when @NPR reported yesterday that Americans “most affected by the shutdown” are people stuck in line at airports. I think it might be the dead, wounded, terrified folks in cities targeted by ICE & CPB who were are at the front end of the shutdown.