OK, counterpoint alert: I've just spent a profitable 20 minutes chatting with GPT-4.5 in Catalan, where it has corrected my errors, explained things and gave me a quiz. It's certainly not "PhD-level" but it was *great*. It's the kind of use case these things are really for.

Dario Amodei has an important op-ed in the NYT today about the proposed 10-year moratorium on states regulating AI. The moratorium sounds reasonable at first - avoiding a patchwork of state laws - but it's actually quite dangerous given how fast AI is moving. www.nytimes.com/2025/06/05/o... 1/6

Just got this from my mom. My work here is done.

For those looking to use LLMs to summarize scientific (in this case medical) papers: "LLM summaries were twice as likely to contain generalized conclusions compared to the original abstracts, indicating an algorithmic overgeneralization tendency." pmc.ncbi.nlm.nih.gov/articles/PMC...

In 1968, U.S. soldiers massacred hundreds of civilians in My Lai, Vietnam. The killing was interrupted by helicopter pilot Hugh Thompson Jr. If pilots are to be replaced by AI, then we should ask, could AI on a drone do the same in a future war?

Wrote a thing on value-based choices when deploying AI. It's based on a military situation (a sci-fi version of the My Lai massacre) but I can't help but think this is a problem lots of leaders will face in more subtle, and less stressful, ways.

As military AI evolves, ethical dilemmas grow. Should drones merely inform soldiers, coach decisions, or enforce rules of war directly? Read the new piece by Minji Jang & CSET's @emmyprobasco.bsky.social, out today in @warontherocks.bsky.social 👇

Bradford G. Smith’s case highlights key issues about the prospect that brain implants and AI will one day merge. www.technologyreview.com/2025/05/07/1...

Great article by the one and only @stephbatalis.bsky.social www.defenseone.com/ideas/2025/0...

Erwin Chemerinsky and @tribelaw.bsky.social: We Should All Be Very, Very Afraid www.nytimes.com/2025/04/09/o...

I'm super excited to see our #CSET report on **AI-enabled military decision support systems** being released today! Great work by @emmyprobasco.bsky.social, @hlntnr.bsky.social, and Matthew Burtell!

⭐️New Report⭐️ Using AI to make military decisions? CSET’s @emmyprobasco.bsky.social, @hlntnr.bsky.social, Matthew Burtell, and @timrudner.bsky.social analyze the advantages and risks of AI for military decisionmaking. cset.georgetown.edu/publication/...

There are plenty of technical & financial hurdles to defense against “ballistic, hypersonic, and cruise missiles, and other advanced aerial attacks” #GoldenDome. But one thing is very different from Reagan’s days: the commercial space industry. A story in @csetgeorgetown.bsky.social charts: 🧵

In this month’s newsletter, CSET's @alexfriedland.bsky.social wrapped up all the latest AI & policy news: 1️⃣ GPT-4.5 & Claude 3.7 Sonnet 2️⃣ Recapping new misalignment research 3️⃣ DOD wants to speed up software acquisition 4️⃣ AI Action Plan recs And much more! cset.georgetown.edu/newsletter/m...

The future of AI leadership requires thoughtful policy. @CSETGeorgetown just submitted our response to @NSF's RFI on the Development of an Artificial Intelligence Action Plan. Here's what we recommend: 🧵[1/]

👏 grad.georgetown.edu/opportunitie...

Had fun chatting with Gen. Tim Haugh (US CYBERCOM/NSA about military applications of AI in Austin this Friday. Video ICYMI. www.youtube.com/live/yIMRxOg...

Check out the latest from @sambresnick.bsky.social and @colemcfaul.bsky.social on how “good enough” Chinese tech could come to dominate exquisite US tech in global adoption of AI www.barrons.com/articles/dee...

Getting into the nitty gritty of energy and compute constraints for drones using AI in conflict arxiv.org/abs/2502.15737

Couldn’t help but go through the list of intelligence failures in this article and ask: would AI have exacerbated or addressed this risk? www.timesofisrael.com/the-intel-on...

Re-upping CSET's report on the space launch industry in light of the @washingtonpost.com article on government contracts for Tesla and SpaceX. Bottom line: USG funding is critical to the industry, & the market is concentrated (5 of every 6 launches are by SpaceX) cset.georgetown.edu/publication/...

Timely webinar today: cset.georgetown.edu/event/how-th...

What is the indirect rate for research grants from the private sector? I saw a few pharma companies are around 25%. I also saw a few big tech companies advertising between 0 and 5%. I get that the infrastructure needs may be different depending on the research but 5% and below seems cheap...

Good explanation of how the ways we discuss AI (i.e. "AGI" and vague goals) impede progress towards creating AI that benefits humanity. Reminded me of what I call AI's Inigo Montoya problem: "You keep using that word. I do not think it means what you think it means" Bravo @borhane.bsky.social

‼️”there are only so many times that I can say that these systems are amazing, but get things wrong all the time in ways that matter, and so the best uses cases so far are things where the error rate doesn’t matter or where it’s easy to see.”‼️ www.ben-evans.com/benedictevan...

Seeing news like this makes me so grateful for @csetgeorgetown.bsky.social and our @usspaceforce.bsky.social fellows, who have been publishing on space topics for the past several year. 1/3 www.airandspaceforces.com/ussf-victus-...

Book fact: After a revolt at the silver mines in Joachimsthal in 1525, the mine investors agreed to end the truck system, allowed extra shifts, and pay wages in good, full-silver pennies, minted on site. In return, the miners had to stop getting drunk and exposing themselves in town.

Fun, quick reminder of AI’s current limitations chiraaggohel.com/posts/llms-e... h/t @igormt-cset.bsky.social

UK published a government efficiency review and a plan to leverage AI for digital services, including digital drivers licenses. Would be nice to see digital passports too! www.gov.uk/government/n...

Trying to keep up with news about AI that’s relevant to national security? Me too. This crew is a great help. go.bsky.app/5AWz4Ht

Bravo @katecrawford.bsky.social on this important essay. As we experiment with AI apps that improve our lives, we cannot overlook the potential for subtle manipulation that plays with the emotions and cognitive biases that make us human.

Good read on AI applications from @breakingdefense.com breakingdefense.com/2024/12/kill...

“Basically, a frontier model like OpenAI's O1 is like a Ferrari SF-23. …In contrast, a BERT model is like a Honda Civic. It’s also an engineering triumph, but …engineered to be affordable, fuel-efficient, reliable, and extremely useful.” 😍 huggingface.co/blog/modernb...

Generally, I hear too much hype about the application of LLMs for the military. But sometimes I hear too much skepticism. Here's a nice piece to help calibrate: LLMs for classification markings. Maybe not as sexy as course-of-action development, but truly helpful. blog.palantir.com/ais-role-in-...

How do Chinese defense experts think about the military use of AI? CSET's @sambresnick.bsky.social joined @gmfus.bsky.social's @bglaser.bsky.social to discuss his recent report. Listen now 🎧 podcasts.apple.com/us/podcast/c...

Anyone know where to get the beanie Coach Newberry is wearing? #gonavy #usna

Come for the neat proof of concept by CSET's @emmyprobasco.bsky.social and Christian Schoeberl - stay for the creative use of ETO's Map of Science 🗺️🧪

Shared some thoughts about the OpenAI/Anduril/Palantir announcements with @technologyreview.com www.technologyreview.com/2024/12/10/1...

The Economist published "Will the bubble burst for AI in 2025, or will it start to deliver?" Key quote: "To get the most out of the technology, bosses need to create an environment that incentivises openness and experimentation, rather than secrecy and suspicion." Good advice for DOD & academia

For all my favorite Excel mavens... esportsinsider.com/2024/11/exce... h/t @alicrawford.bsky.social ...almost embarrassed to admit I'd enjoy watching this. Almost.

SXSW London: great place for a discussion on AI for defence (or defense). Please follow the link to vote for my panel with Aki Jain, Christine Fox, and Joe O’Callaghan: sxswlondon.awardsplatform.com/entry/vote/l...

I read today that “52% of companies that use AI give their employees training on how to use it, leaving 37% of employees feeling underconfident with their AI skills” Here’s my paper about why this is a safety & liability issue tangled up with user “confidence.” cset.georgetown.edu/publication/...

In case AI comes up at your Thanksgiving table (and to avoid other awkward conversations) here is a fun article for discussion. www.wsj.com/articles/tha... h/t Jacob Feldgoise

Automation isn't just a technical challenge. Organizational policies and human factors play a big role in managing automation bias risks. Our new report dives in. Read more ⬇️ cset.georgetown.edu/publication/...