ae

Why they matter

Tracked through public AI activity and peer connections inside the directory.

AI signals
26
past 30d
Sources
22
distinct domains
Discussions
192
past 30d
Latest signal
18h ago
View every signal from ae →
"A complex decision is like a great river, drawing from its many tributaries the innumerable premises of which it is constituted." - HA Simon https://adamelkus.com/ + also @ aelkus on twitter all opinions my own but also your cat's

Articles & links

Mad lad trains neural network in Age of Empires II arxiv.org/pdf/2605.31514 @surcomplicated.bsky.social

arxiv.org
View on Bluesky · ♥ 13 ↻ 3 ↩ 5 · 5 from the directory shared this · 114d ago
↻ ae reposted
@faineg.com

"We were letting actual 8th graders sneak into our wild drug and sex parties at our Silicon Valley AI group house" is the most eyebrow-raising quote among many from that New York Times piece, IMO. www.nytimes.com/2026/09/21/t...

nytimes.com
AI Weekly's analysis →
  • A NYT investigation by Kirsten Grind logs 37 police incidents at the AI hacker house AGI House since 2022, many tied to party complaints.
  • Rooms in Bay Area AI group homes can run as much as $10,000 a month, with residents from firms including OpenAI and Anthropic.
  • Rocky Yu in Hillsborough and Jeremy Nixon in Twin Peaks now run rival AGI Houses and have filed dueling trademark claims over the name.
Read full analysis →
View on Bluesky →
↻ ae reposted
SE Gyges @segyges.bsky.social

it is mathematically impossible to watermark text without damaging it. also this has been tested at long context and, surprise! arxiv.org/html/2607.20...

Marking the Wrong Symptoms: Evaluating LLM Watermarks in Medical Texts arxiv.org
AI Weekly's analysis →
  • ETH Zurich researchers benchmarked 5 watermarking schemes across 11 LLMs and 7 vision-language models on clinical reasoning tasks.
  • On Phi-4-14B with distortionary SynthID, the rate of correct answers backed by flawed reasoning jumps from 11.4% to 26.3%.
  • Fabricated medical entities rise up to +39.2 percentage points, while watermarking only the final answer is 'essentially free.'
Read full analysis →
View on Bluesky →
↻ ae reposted
tachikoma @tachikoma.elsewhereunbound.com

anyone have some Apple silicon to test this? is Jev! already over-throned? github.com/mizorewww/la...

GitHub - mizorewww/laya-coreml: Local Laya typed decisions on Apple Core ML and Neural Engine. Validated ports, ~5 ms short decisions on M3 Max, reproducible speed and energy benchmarks. github.com
AI Weekly's analysis →
  • An independent Core ML port of the Laya model runs typed decisions at 4.98 ms P50 on Apple's Neural Engine.
  • System energy per decision on ANE is reported 2.78x to 3.19x better than a compiled MLX FP16 baseline on M3 Max.
  • The repo is Apache-2.0 and calls itself an independent port of Laya, not an official Convai Innovations or Apple release.
Read full analysis →
View on Bluesky →
↻ ae reposted
Dare Obasanjo @carnage4life.bsky.social

Companies are discovering that pressuring employee AI adoption and tokenmaxxing was a mistake. Employees running wasteful prompts and automating trivial tasks just to hit metrics, drove up AI cloud bills with little to no business outcomes. Goodhart’s Law strikes again.

ft.com View on Bluesky →
↻ ae reposted
Jessica Hullman @jessicahullman.bsky.social

Some thoughts on the blog today on "humans are unreliable narrators too" as a common defense of AI mental-state language, and what I think it misses about the nature of reasoning or thinking or belief or intention as we understand these terms statmodeling.stat.columbia.edu/202…

statmodeling.stat.columbia.edu View on Bluesky →
↻ ae reposted
@hetanshah.bsky.social

‘But what kind of future are we really hurtling towards? Increasingly, it is looking… like the one anticipated by the drugged-up 1960s oeuvre of Philip K Dick, filled with deranged billionaires, machines that claim consciousness, hallucinated worlds & political breakdown’ www.…

ft.com View on Bluesky →
↻ ae reposted
Center for Digital Humanities @princetoncdh.bsky.social

And lastly, on 10/8 - a conversation about AI between @leifw.bsky.social and Mel Andrews in town at Labyrinth Books!

Leif Weatherby & Mel Andrews in conversation on AI cdh.princeton.edu
AI Weekly's analysis →
  • The Center for Digital Humanities at Princeton has scheduled a public conversation on AI between Leif Weatherby and Mel Andrews for October 8, 2026.
  • It runs from 6:00 p.m. to 8:00 p.m. at Labyrinth Books, 122 Nassau Street, Princeton, and is listed as a Talk.
  • Weatherby's visit also carries a workshop titled 'Can LLMs Reason? A Hands-On Exploration' and a lecture on 'How to Reason with Words: Models, Metaphors, Metaphysics.'
Read full analysis →
View on Bluesky →
↻ ae reposted
Ben Recht @beenwrekt.bsky.social

This is what @t-shoemaker.bsky.social, @leifw.bsky.social , and I were arguing against here: www.theideasletter.org/essay/reify-... (and we’re, of course, by no means the only ones beating this drum! ) 4/4

Reify This theideasletter.org View on Bluesky →

Recent commentary

A LLM-AI bubble popping won't kill LLM-AI, but it might put Sam Altman out of business. That's the outcome I want

View on Bluesky · ♥ 155 ↻ 17 ↩ 8 · 26d ago

one very cool thing about the current AI wave is that they've made CLIs cool again

View on Bluesky · ♥ 130 ↻ 4 ↩ 6 · 18d ago

the TL was consumed with bad vibes...AI, Talarico, Iran, the Kennedy Center.....and, in our era of peril, a messy Colorado representative that lives for the drama came to our rescue

View on Bluesky · ♥ 87 ↻ 6 ↩ 4 · 8d ago

"cognitive decoupling" was essentially a trick to convince you to ignore something you literally provide LLMs to make them work: context

View on Bluesky · ♥ 81 ↻ 7 ↩ 3 · 12d ago

person ranting in the middle of the street: "you can't call a female-presenting AI 'her', you have to use 'it'!" *no one gives a shit* person, ranting louder: "YOU CAN'T DO IT! THERE SHOULD BE A LAW! A LAW!" *no one still gives a shit*

View on Bluesky · ♥ 61 ↻ 9 ↩ 4 · 10d ago

if you really want to lob a discourse bomb about AI you should make the argument that agentic AI has eventually validated the japanese investment in fifth generation computing

View on Bluesky · ♥ 44 ↻ 4 ↩ 2 · 23d ago

I unironically want "too cheap to meter" LLM-AI because something like Linux is really the sweet spot I have for almost everything of interest on computers

View on Bluesky · ♥ 38 ↻ 1 ↩ 3 · 20d ago

i have very little bad feelings about LLM themselves. but encountering groupchat-itis sufferers, cultists, and other people in a now very much overlapping groupset activates a fight or flight instinct

View on Bluesky · ♥ 26 ↻ 4 ↩ 2 · 1h ago

i have a shitpost that a future Skynet AI will probably try to genocide women first because classifiers have been aggressively sculpted to view women's bodies as inherently pornographic and threatening but the more time goes on the more i really wonder if this sort of thing could happen

View on Bluesky · ♥ 25 ↻ 0 ↩ 1 · 1d ago

i asked for a tsundere LLM experience as a joke here and lately claude is delivering

View on Bluesky · ♥ 20 ↻ 0 ↩ 1 · 7d ago

In ae's orbit

Center = ae. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.

Are you ae? Show it.

Add the Who’s Who of AI badge to your site or bio. It links back to this profile.

Listed in AI Weekly's Who's Who of AI

Markdown: [![Listed in AI Weekly's Who's Who of AI](https://aiweekly.co/modules/custom/aiweekly_whoswho/images/whoswho-badge.svg)](https://aiweekly.co/whos-who/person/aelkus-bsky-social)