Nick Franklin

Brown University Rejects White House Deal for Special Treatment

So proud of my alma mater! Awesome.

www.nytimes.com/2025/10/15/u...

Don’t let anyone tell you this technology doesn’t work

It is faster to sit around and wait for the AI to write my code, but man, I’m just starring at my phone all day now

We’re hiring!

@sucholutsky.bsky.social and I are seeking a postdoc and RA for a project on trust in AI systems with folks at NYU, Princeton, BU, and Cornell

Positions open until filled. Apply soon! Please share 🔁

postdoc: apply.interfolio.com/175495

RA: apply.interfolio.com/175497

I'm recruiting grad students!! 🎓

The CoDec Lab @ NYU (codec-lab.github.io) is looking for PhD students (Fall 2026) interested in computational approaches to social cognition & problem solving 🧠

Applications through Psych (tinyurl.com/nyucp) are due Dec 1. Reach out with Qs & please repost! 🙏

Pretty proud of our Andrew Liu's work on Flash-IPA, a hardware efficient geometric attention mechanism, which was accepted to this years #NeurIPS openreview.net/forum?id=gKs...

ATTN🚨: I will be looking for PhD students through NYU's Center for Data Science PhD program this year. Applicants should have an interest in either NeuroAI (specifically biological attention or AI interpretability) or ML for Remote Sensing. Visit my lab website for more info: lindsay-lab.github.io

Kempner Research Fellowship - Kempner Institute The Kempner brings leading, early-stage postdoctoral scientists to Harvard to work on projects that advance the fundamental understanding of intelligence.

If you work on artificial or natural intelligence and are finishing your PhD, consider applying for a Kempner research fellowship at Harvard:
kempnerinstitute.harvard.edu/kempner-inst...

I don’t like that I had to look this up

Or worse, they know both and are just really cynical

You are also assuming they are motivated to keep it a secret, which, maybe

In fairness, the steel-man argument that LLMs are exactly equivalent to human cognition is worth making! It’s a good, (easily?) falsifiable hypothesis

Just 1 week left to apply! 🙏 Pls share with anyone you think might be interested. If there are any Q's, get in touch!
👉 3x PhD position: hmc-lab.com/ERC_PhDs.html
👉 2yr Postdoc position: hmc-lab.com/ERC_Postdoc....

Project Vend: Can Claude run a small shop? (And why does that matter?) We let Claude run a small shop in the Anthropic office. Here's what happened.

Anthropic let Claude run a vending machine business at their HQ.

People tricked it into buying a bunch of tungsten steel cubes that it sold at a loss.

www.anthropic.com/research/pro...

Not a chance in their current form, and there is research to support “hard” question are beyond their reasoning abilities

I still can't get past the idea that people in America are now supposed to accept that this is a valid form of law enforcement and just assume it's fine when random dudes in groups with face-coverings, no ID, no warrant, & no uniform grab people off the street and force them into unmarked vehicles

Language Models in Plato's Cave Why language models succeeded where video models failed, and what that teaches us about AI

This is an excellent and very clear piece from Sergey Levine about the strengths and limitations of Large Language models.
sergeylevine.substack.com/p/language-m...

You see, it’s a linear hyperplane in a Hilbert space, but we actually don’t need to construct the… Blasphemy! All of it

Yes, this is true. SVMs are an affront against god

We Made Top AI Models Compete in a Game of Diplomacy. Here’s Who Won. The models that did the best learned to lie, deceive, and betray their fellow players

When Claude, Gemini, o3 battle each other for world domination in Diplomacy:

DeepSeek turned warmongering tyrant. Claude wouldn't lie—everyone exploited it ruthlessly. o3 orchestrated a secret coalition, backstabbed every ally, and won.

tx textArena.ai and friends

every.to/p/diplomacy
🧠🤖
#AI #LLM

Based on current capabilities and known issues with LLMs and humanoid robots, it turns out that in retrospect the most accurate depiction of actual robots in fiction was C-3PO. Lucas does it again, the absolute madman.

I’m down for this as long as it is append only

User: BlueskiAi, fact this claim about Trump
Ai: Although the anthropomorphic nature of my response makes me appear sentient, that is an illusion and the scientific consensus on my intelligence is ambivalent

No, I want to be able to alter other people’s AI prompt

Congrats!

Oh, I misunderstood the context. Zenodo doesn’t make sense for that. Seems like anything you can agree to (s3, Dropbox) should be fine?

Check out zenodo

FINALLY.

Some gumption from the President of Harvard today:

I’m going to be mad if this joke is still funny 90 years from now