Dr. Justin Baumann

AI’s water problem is worse than we thought A new investigation reveals how Amazon is amplifying Oregon’s nitrate pollution crisis.

heated.world/p/data-cente... from @emorwee.bsky.social

Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.

Fully protected Marine Protected Areas do not displace fisheries | PNAS Fully protected Marine Protected Areas do not displace fisheries

Fisheries: Creating marine protected areas only displaces fisheries and does not benefit the ocean

Costello: No they don't and yes they do

www.pnas.org/doi/10.1073/...

Hi folks! Is anyone recruiting masters students that pay a livable stipend? I have an excellent senior who is an incredible researcher and seeking a position. She is top notch and really interested in mechanistic questions- currently working on Aiptasia, Astrangia and Nematostella.

Beginning of semester me: *sets aside Tuesday and Thursday for research*

Me now: Spends all of Tuesday and Thursday grading and writing emails.

:(

NSF Graduate Research Fellowship Program (GRFP)

NSF GRFP solicitation out.
App deadlines now in November.
As others noted, eligibility rules for current graduate students now exclude 2nd year grad students. See the solicitation - they’ve emphasized “first year” and “less than one academic year” in grad program.

www.nsf.gov/funding/oppo...

OpenAI admits AI hallucinations are mathematically inevitable, not just engineering flaws In a landmark study, OpenAI researchers reveal that large language models will always produce plausible but false outputs, even with perfect data, due to fundamental statistical and computational limi...

Reason No. 324 why no-one should think that "AI" (artificial but certainly not intelligent) is an unmitigated Good Thing: it's mathematically inevitable that AI will hallucinate.

We need far more discussion of the risks of all this and less frantic hype. I mean, really, what's the hurry?

CH₄ has a lifetime of about a decade in the atmosphere, whereby it oxidizes to CO₂ and remains in the atmosphere for tens of thousands of years. More than 99% of the cumulative climate impact (i.e., the radiative forcing) is from its life as CO₂. In the end, it's still CO₂ that gets you. 🥲

Pulling mussels: Bates researchers spend the summer studying Maine’s disappearing shellfish A research team from Bates spent the summer investigating Maine's disappearing shellfish, the blue mussel.

New press from Bates on what my summer students were up to
www.bates.edu/news/2025/09...

Read all about our journey across the Maine intertidal :)

I present to you- my lab as trees.
My students said we needed “more whimsy”, so we have it now. Great summer working in mussel physiology with these awesome students

Sounds good. My own experience is that green crabs often have some blue coloring on claws and elsewhere. I’d guess is not a hybrid. But the blue is very blue and there are people who might be keen to test it out!

Can you elaborate on the location (what bay, for example) and how common this blue morph is?

Any more photos?

We can’t even get everyone to read the syllabus 😭

I used to tell biology/forestry/ecology graduate students that being a Federal Agency Research Scientist was a great career to shoot for, but they might not even exist next year.

Everyone is acting like US scientists will just go get science jobs elsewhere and sure some will but there are not anywhere close to enough science jobs elsewhere.

The end result of this will be much, much, much less science, not science happening in different places.

I wrote about chemtrails, Lee Zeldin, and how his EPA routinely brags about its participation in a conspiracy to let corporations poison us
newrepublic.com/article/1978...

Read news, op-eds, and analysis, not AI summaries. Create your own art (even if it's messy) or hire an artist to do it. Do your own homework. Talk to people, not chatbots. Keep your thinking and skills sharp and cherish our messy humanity. That's the new punk rock.

Governments must enable faster wind growth and increase ambition National targets are not ambitious enough for a global tripling, and most countries are still not on track to deliver even these. A global tripling of renewables capacity as agreed at COP28 last year means at least a tripling of wind capacity. The vast majority of countries that currently have wind capacity installed also have wind targets for 2030. However, the sum of those targets only aims for just over a doubling of capacity, and falls short of the tripling needed to stay on the 1.5C pathway. Forecasts for global wind capacity suggest that meeting the 2030 national targets is achievable, in aggregate. However, this is almost entirely due to China’s dramatic rise as a global wind leader. Without China, the rest of the world is set to fall far short of national targets and even farther from a tripling. Wind has an important role to play in meeting electricity generation needs associated with the global tripling of capac

Listen: this is obscenely important. There is a growing complacency that over-achieving solar can make up for under-achieving wind, and it is *not true*.

Wind and solar generate in a complementary way. Solar might max out. Wind needs much more support!!

ember-climate.org/insights/res...

a chart showing most models are red - ie, no disclosure - for alphabet, microsoft, openai, anthropic

a chart showing most models are red - ie, no disclosure - for alphabet, microsoft, openai, anthropic

a chart showing most models are red - ie, no disclosure - for alphabet, microsoft, openai, anthropic

a chart showing most models are red - ie, no disclosure - for alphabet, microsoft, openai, anthropic

New @sashamtl.bsky.social et al paper - assessing whether machine learning model developers disclose basic environmental / energy information with their models. They mostly do not!!

Grim results for most of the big companies:

arxiv.org/abs/2506.15572

huggingface.co/spaces/sasha...

Yes. I am right. I am seeing how Chat GPT is ruining students critical thinking and writing skills in real time. It is not the future. It is a tool designed to render the populace helpless, to make people doubt their innate intelligence, and to foster overreliance on technology.

Today on Juneteenth, I am amplifying #BlackInTheIvory anecdotes that were shared by Black scientists 🧪 in summer 2020.

Screenshots from Twitter presented in anonymized format 2025 is the era of the "anti-DEI crusade".

If a human told you things that were correct 80% of the time but claimed, flat out, with absolute confidence, that they were correct 100% of the time, you would dislike them & never trust a word they say. All I'm really suggesting is for people to treat chatbots with that same distrust & antagonism.

But the idea that “the product doesn’t suffer” seems so questionable. And also (per other comments) just doesn’t seem possible

Peer review is slow! This is a pre-print that is ideally in the review process now. Pre-print sharing is common in academic spheres. Though grain of salt required…

No idea if these authors have submitted this for peer review or not.

www.statnews.com/2025/06/16/n...

Super important point here-- privatizing GRFP or really anything funding science makes folks doing the work beholden or at least feel beholden to whatever special interest is funding. This drastically impact messaging, results, and even the questions that are asked.

Museum Specialist - New York, NY 10024 - Indeed.com American Museum of Natural History

The Invertebrate Paleontology department at the AMNH is seeking a full-time Museum Specialist.

www.indeed.com/viewjob?from...

This seems accurate to me! Thanks for chiming in :)

Cutting NSF, NIH, etc funding in this manner prevents knowledge generation, innovation (even though I hate that concept), risks human/public health, and keeps the populous less informed. That, my friends, is fascism.