Who's Who of AI

Amy Hoy

61 trust @amyhoy.bsky.social · 22,600 followers
Why they matter

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

AI signals
13
past 30d
Sources
12
distinct domains
Discussions
196
past 30d
Latest signal
18h ago
View every signal from Amy Hoy →
- OG web - design - hci - reading - writing - saas biz - edu biz - photos - cats - desert - she "a powerfully repulsive mixture of dense, stubborn and arrogant" to people who don't like facts.
What they're sharing

Articles & links

the anthropic dweebs have "discovered" that claude* uses a sort of swap file/memory. and that if you tell it "white elephant" this swap "space" includes the tokens "white elephant." & that means it's basically conscious *the entire pkg, bc without the harness, the llm is useless

A global workspace in language models \ Anthropic anthropic.com
AI Weekly's analysis →
  • Anthropic says Claude has a 'J-space' of dozens of concepts, under a tenth of neural activity, that mediates multi-step reasoning.
  • Swapping 'spider' for 'ant' inside the J-space changed Claude's leg-count answer from 8 to 6, demonstrating a causal role.
  • A 'J-lens' tool surfaced silent words like 'fake', 'fictional' and 'manipulation' during deception tests, pointing at safety uses.
Read full analysis →
View on Bluesky · ♥ 14 ↻ 1 ↩ 1 · 7 from the directory shared this · 42d ago
↻ Amy Hoy reposted
@thomasfuchs.at

Good

Landmark German ruling declares Google's AI Overviews are Google's own words and makes it liable for false answers the-decoder.com
AI Weekly's analysis →
  • Munich's Regional Court declared Google's AI Overviews its own speech, rejecting traditional search-engine hosting liability protections.
  • Google's AI Overviews answered correctly around 91% of queries, but 56% of those correct answers couldn't be verified through the linked sources.
  • Google confirmed it will appeal the ruling, which could set precedent for AI answer engines including ChatGPT and Perplexity across Europe.
Read full analysis →
 View on Bluesky →
↻ Amy Hoy reposted
Ed Zitron @edzitron.com

No industry of value needs to mislead you or make you feel bad for not adopting their tech. Nobody trustworthy will ever see the need to humiliate or attack somebody for not being excited about a product. These are the actions of a loser. www.wheresyoured.at/the-revenge-of-the…

Revenge of The Business Idiot wheresyoured.at View on Bluesky →
↻ Amy Hoy reposted
Casey Fiesler @cfiesler.bsky.social

I'm starting to see a pretty coherent body of research pointing to this answer. For example, this paper "Social Dynamics of AI Adoption" is about *parents*: "Anxiety about falling behind can drive people to embrace emerging technologies with uncertain consequences." www.nber.o…

nber.org View on Bluesky →
↻ Amy Hoy reposted
Karl Bode @karlbode.com

this article is ostensibly about AI investment driving a merger boom but the author doesn't think it's relevant to mention historic corruption has hollowed out our antitrust regulators (despite Republicans campaigning on being big fans of antitrust) or that AI doesn't make any…

nytimes.com View on Bluesky →
↻ Amy Hoy reposted
Hypervisible @hypervisible.blacksky.app

“What makes ELIZA particularly fascinating is not only its historical significance but also what it reveals about Weizenbaum’s views on both computing and human interaction.”

Inventing Eliza: The First Chatbot’s Hidden Code - IEEE Spectrum spectrum.ieee.org
AI Weekly's analysis →
  • The original ELIZA source code was unearthed from MIT's archives, showing Joseph Weizenbaum's 1966 write-up omitted key technical details.
  • Rather than a simple pattern-matcher, ELIZA was a scripting platform with contextual memory and personas for math, poetry, and relativity.
  • Edwin F. Taylor at MIT's Education Research Center built teaching scripts using conditional keyword matching that branched on prior turns.
Read full analysis →
 View on Bluesky →
↻ Amy Hoy reposted
@carlbergstrom.com

1. This is a very interesting story. But this is not a matter of researchers learning how to understand what the LLMs are doing. It is researchers learning how to get around the companies’ attempts to hide what the LLMs are doing to prevent other companies from training off of…

A New Trick Reveals AI Models’ Inner Thoughts wired.com View on Bluesky →
↻ Amy Hoy reposted
@erikahall.bsky.social

This has been the stupidest technology cycle in my adult life. Somehow stupider than NFTs because so many people I thought were smart got roped into the hype. www.wsj.com/business/chi...

wsj.com View on Bluesky →
Their own posts

Recent commentary

if you didn’t write it… you didn’t write it no, using an llm to “draft” for you is not writing it writers write: that’s one of the four noble truths for writers if you don’t write, you’re not a writer it’s really that simple no wriggling, hairsplitting, whining argument will change that

View on Bluesky · ♥ 88 ↻ 19 ↩ 4 · 2d ago

here's what i think is weird and gross about llm boosters: so, you use it. it works for you. okay? and? i would no more try to talk someone else into using an llm than i would talk someone else into using photoshop i don't understand the impulse, and find it inherently suspect

View on Bluesky · ♥ 48 ↻ 3 ↩ 6 · 12d ago

thomas: i can’t believe people are complaining that the ai watermarking will affect “meaning.” ai text has no meaning me: right?? if you care about word choice, you’re not using an llm. ipso facto, bitches and that’s why i love leisurely mornings over coffee with my husband

View on Bluesky · ♥ 46 ↻ 5 ↩ 1 · 22h ago

i want to read something on why llms focus so much on weird ~ceremonial~ uses of nouns and verbs, freighting them with meaning they don't have, and using them like thought-terminating clichés. it's not just awkward. and i'd say there's very little real-world sample data to cause this.

View on Bluesky · ♥ 30 ↻ 2 ↩ 3 · 6d ago

stuff "designed" with ai has a very special stink that i find deeply, viscerally off-putting

View on Bluesky · ♥ 25 ↻ 3 ↩ 0 · 57d ago
Their network

In Amy Hoy's orbit

Center = Amy Hoy. 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.