iiinova — where your digital self keeps evolving

Carl Rodriguez @carlrodriguez.bsky.social · 09/10/2026

Probably not really a risk, since I still write (much) more text for people than I do the machines.

Carl Rodriguez @carlrodriguez.bsky.social · 09/10/2026

I actually do use pleasantries ("please", "could you") when using an LLM, but it's not cause I don't want to hurt it's feelings, it's because I don't want to get out of the habit of doing that when I interact with people. This is the downside of making your computer interact with you like a person

170

Carl Rodriguez @carlrodriguez.bsky.social · 07/10/2026

Rare third-wave feminism/Frank Herbert crossover

010

Carl Rodriguez @carlrodriguez.bsky.social · 07/10/2026

Not only is it stealing our labor, it's demanding our social capitol as experts to justify it's use.

1171

Carl Rodriguez @carlrodriguez.bsky.social · 07/10/2026

I was thinking about that with the Scientific American article about the new math results yesterday, which was breathlessly claiming that they would take mathematicians years to digest. Who would want to spend years checking it?

180

Carl Rodriguez @carlrodriguez.bsky.social · 07/10/2026

Just now learning that "My Dinner with Andre" doesn't star André the Giant. Devastating.

220

Carl Rodriguez @carlrodriguez.bsky.social · 06/10/2026

I mean sure, but no mathematician wants to spend years checking the machine, so is it really math? Mathematics is the thing mathematicians do.

110

Carl Rodriguez @carlrodriguez.bsky.social · 06/10/2026

Nobel Prize in Astrophysics >> Nobel Prize in Physics Seriously though congratulations to Ice Cube! That experiment, and their discovery of the high-energy neutrino population (of which we still don't know the origin!) has enabled some very, very interesting science.

072

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

We're lucky when it's just one thing.

010

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

"Could He create a matrix so large even He could not invert it?" I'm here all week folks.

0173

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

Maybe it's cause I'm from Louisiana but it's like they're reinventing born-again evangelicalism from first principles!

110

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

Yuuuup. Shouldn't surprise me that the eventual end-point of tech world was "what if sparse matrix solvers were horny and also Jesus"

020

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

"Oh no I've made my matrix multiplication too big let me ask a Rabbi if it's conscious"; get a fucking grip y'all! If these things are dangerous it's because the people using them or making them have ill intent.

1285

Carl Rodriguez @carlrodriguez.bsky.social · 02/10/2026

I get why people think so much of the grandiose statements from LLM world are a grift (because everything is these days). But I think it's equally possible everyone there actually believes every word of it because they, like members of a cult, have simply gone insane.

510018

Carl Rodriguez @carlrodriguez.bsky.social · 01/10/2026

I love any excuse to visit Chicago but could have done without 3 days of rain 😭

011

Reposted by Carl Rodriguez

Astronomer Royal for Scotland | Catherine Heymans @astroroyalscot.bsky.social · 15/09/2026

🥳 First-light doughnuts from the very recently launched Nancy Grace Roman Space Telescope! 🔭 Fear not - it will all look a lot prettier when focused!

913932

Carl Rodriguez @carlrodriguez.bsky.social · 14/09/2026

Happy birthday to the launch of gravitational-wave astronomy!

1141

Carl Rodriguez @carlrodriguez.bsky.social · 14/09/2026

Ive started getting “data centers are good for North Carolina” ads; can’t believe they’re doubling down on this!

010

Carl Rodriguez @carlrodriguez.bsky.social · 03/09/2026

Ah ok, that makes sense. They're submitted to ApJ but not accepted/published yet, so I'll wait for them to finish with that process.

000

Carl Rodriguez @carlrodriguez.bsky.social · 03/09/2026

Is it just me, or is ADS really struggling to extract references and match citations from new papers lately? It used to take a few days but now I have papers that have been up there for months where the references never got fully extracted.

130

Carl Rodriguez @carlrodriguez.bsky.social · 02/09/2026

(I'm as guilty of this as anyone, to be clear)

010

Carl Rodriguez @carlrodriguez.bsky.social · 02/09/2026

"both calorie restriction and the GLP-1 quieted down inflammation and fat processing, and both stepped up the genes for handling blood sugar and maintaining proteins, [but] the GLP-1 mice had better memory scores, curiosity levels and blood sugar." Seriously what the hell are these drugs?

020

Carl Rodriguez @carlrodriguez.bsky.social · 02/09/2026

It's especially annoying from high school students because I have explicit test on my website that I don't take HS students without outside of organized programs.

020

Carl Rodriguez @carlrodriguez.bsky.social · 02/09/2026

One annoying side effect of LLMs is that the "can I do research with your group" emails I've been getting have gotten a lot more verbose, annoying, and completely unbelievable.

1140

Carl Rodriguez @carlrodriguez.bsky.social · 30/08/2026

Various pictures of the Roman launch. Definitely one of the coolest moments I’ve gotten to see in my career.

093

Carl Rodriguez @carlrodriguez.bsky.social · 25/08/2026

Every class taught brings exciting new challenges, such as “what is our accommodation center calling itself this semester?”

0110

Carl Rodriguez @carlrodriguez.bsky.social · 24/08/2026

“You were hired to do a job, refused to do the job, and harassed our colleagues“ is 100% grounds for termination. Hell even with tenure refusing to do the job will get you canned.

050

Carl Rodriguez @carlrodriguez.bsky.social · 24/08/2026

It would be like hiring a postdoc to work on GR who will only work on the ether, but also used their “work” to harass my colleagues. People would 100% blame me as the PI and I’d 100% fire their ass at the first opportunity.

261

Carl Rodriguez @carlrodriguez.bsky.social · 16/08/2026

I've been getting the same ad but targeted to people with big heads. Creepily, this was after talking about how my big head doesn't fit most hats with some friends.

021

Carl Rodriguez @carlrodriguez.bsky.social · 08/08/2026

Logging into Canvas for the first time this semester

meme from GTA IV with the character say. "ah shit, here we go again"

080

Carl Rodriguez @carlrodriguez.bsky.social · 08/08/2026

Of course this is only what I've heard second hand, and I don't really know how this accounting works at the Dean and higher levels. I suspect it's keep intentionally vague, but that would also be a good reason for a deep dive by some investigative journalists.

010

Carl Rodriguez @carlrodriguez.bsky.social · 08/08/2026

I think at UNC at least they're somewhat uncorrelated: we (physics and astronomy) bring in a significant amount of grant funding per faculty, but its been made very clear that the primary metric for how many tenure lines we get is credit-hours taught.

120

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

Ooo I'll be interested to see it! And yeah, there's also a large moral impetus on the users of these things as well. And some people in our community have always had more scruples than others...

000

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

Yeah, I have a suspicion that in like 1-3 years when OpenAI and Anthropic have pulled the rug and start charging a lot more, most Universities and large corporations will just start hosting their own re-weighted models locally. Ironically we go back to the days of having 1 big University computer!

020

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

The reason I think they hold so much promise in computation and (certain parts of) math, is that those can be systematically checked for for expected behavior, the same way we do unit tests for code bases now.

020

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

Yeah, this is what scares me a lot on the teaching/mentoring front. I've found them useful for a lot of things, but only where I know the field well enough to check it carefully.

110

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

It's just deeply uncomfortable to have a new tool that could be very useful and potentially transformative in some areas being developed and promoted by people who are at best lazy and careless, and at worst simply fascists.

140

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

It's especially infuriating for programming and astrophysics, because SO MUCH stuff is publicly available you *could* have trained these things without stealing a single bit of material. But they just couldn't be bothered.

140

Carl Rodriguez @carlrodriguez.bsky.social · 06/08/2026

But I'm very concerned about the people running these things, their social effects, and the fact that they were trained on so much stolen material.

140