my little fauxneme
@fauxnemic.com
industry and occasionally academic linguist (pencils & computers, generative & usage, mostly p-side), doing neural shit for literal decades, leftish, oldish, grumpy, don't do cop shit, will occasionally post stuff that makes you rethink following me
a hill i will die on is that people realy really don't understand just how much stuff is in the training data
September 18, 2026 at 4:27 AM
a hill i will die on is that people realy really don't understand just how much stuff is in the training data
what the hell how did i forget that buckaroo banzai was peter weller
September 18, 2026 at 3:55 AM
what the hell how did i forget that buckaroo banzai was peter weller
i might catch hell for this but has anyone tried telling the computer that if it makes a mistake it's going to get fired
September 18, 2026 at 3:52 AM
i might catch hell for this but has anyone tried telling the computer that if it makes a mistake it's going to get fired
Reposted by my little fauxneme
Evo psych: "Men get to be dominant because evolution made women smaller"
EA: "Men get to be dominant because this ensures the production of more future geniuses"
EA: "Men get to be dominant because this ensures the production of more future geniuses"
September 18, 2026 at 1:53 AM
Evo psych: "Men get to be dominant because evolution made women smaller"
EA: "Men get to be dominant because this ensures the production of more future geniuses"
EA: "Men get to be dominant because this ensures the production of more future geniuses"
ok so i just had *actual, real, homemade* turkish delight and narnia makes sense again
September 18, 2026 at 12:46 AM
ok so i just had *actual, real, homemade* turkish delight and narnia makes sense again
Reposted by my little fauxneme
I feel like a lot of everybody adding an AI chatbot to their fucking SaaS product comes down to the software being basically done. Like, it's a SaaS product for managing a trucking business, it's been functionally feature-complete since 2019, you fuss about churn but you don't need a fucking LLM
September 17, 2026 at 10:15 PM
I feel like a lot of everybody adding an AI chatbot to their fucking SaaS product comes down to the software being basically done. Like, it's a SaaS product for managing a trucking business, it's been functionally feature-complete since 2019, you fuss about churn but you don't need a fucking LLM
"i smell burnt toast dr.
penfield"
"yeah i'm making toast"
penfield"
"yeah i'm making toast"
Last few weeks I’ve kept smelling toast, randomly, at home but out and about too, and I remember the myth that this might be a sign of a stroke so I go and check and every time it turns out someone’s actually making toast. There’s like an inordinate amount of toast being made, just to freak me out.
September 17, 2026 at 5:38 AM
"i smell burnt toast dr.
penfield"
"yeah i'm making toast"
penfield"
"yeah i'm making toast"
seeing a lot of talk about a certain community as a "millenarian cult" lately and not to be all schmidhuber-by-proxy but cosma got there way ahead of y'all
September 17, 2026 at 5:28 AM
seeing a lot of talk about a certain community as a "millenarian cult" lately and not to be all schmidhuber-by-proxy but cosma got there way ahead of y'all
"tomorrow is the 30th anniversary of the release of Aenima"
Matt Damon Aging Old Man Old Man
Alt: Matt Damon Aging meme gid
static.klipy.com
September 17, 2026 at 1:25 AM
"tomorrow is the 30th anniversary of the release of Aenima"
Reposted by my little fauxneme
more like stanke- you know what nevermind
September 16, 2026 at 4:43 AM
more like stanke- you know what nevermind
Reposted by my little fauxneme
Rationalism rocks. Love any philosophy that tells me:
1) My actions are moral if I help a greater quantity of future people than the real ones I harm
2) There is an infinite quantity of future people
3) Thus, it's essential that I make as much money as possible by doing things I already want to do
1) My actions are moral if I help a greater quantity of future people than the real ones I harm
2) There is an infinite quantity of future people
3) Thus, it's essential that I make as much money as possible by doing things I already want to do
September 15, 2026 at 11:15 PM
Rationalism rocks. Love any philosophy that tells me:
1) My actions are moral if I help a greater quantity of future people than the real ones I harm
2) There is an infinite quantity of future people
3) Thus, it's essential that I make as much money as possible by doing things I already want to do
1) My actions are moral if I help a greater quantity of future people than the real ones I harm
2) There is an infinite quantity of future people
3) Thus, it's essential that I make as much money as possible by doing things I already want to do
imagine doing bombadil dirty (unbathed) like this
I think it's only going to be poetic justice if a photo turns up of this Aella person dressed up as Tom Bombadil.
September 16, 2026 at 4:14 AM
imagine doing bombadil dirty (unbathed) like this
Reposted by my little fauxneme
I feel like devs who find LLMs transformative in a good way would have a better time and get further with people if they took seriously the possibility that software engineering (which always involves a lil bit of Larping As Computer) is not all that much like most other meaningful human pursuits. 🤷🏻
September 14, 2026 at 2:23 AM
I feel like devs who find LLMs transformative in a good way would have a better time and get further with people if they took seriously the possibility that software engineering (which always involves a lil bit of Larping As Computer) is not all that much like most other meaningful human pursuits. 🤷🏻
genXers seeing that liz holmes pic and just seeing lucy little
September 14, 2026 at 5:52 AM
genXers seeing that liz holmes pic and just seeing lucy little
Reposted by my little fauxneme
By now, just about any term that once indicated some very specific thing(s) are now used to refer to whatever happens to be the case. Basically taking the field's lore and desperately trying to make its prophecies come true.
bsky.app/profile/vcar...
bsky.app/profile/vcar...
Adding "recursive self-improvement" to the list of AI terms that once meant one thing, and are now being used to refer to quite another thing.
It is joined by terms like "continual learning," "general intelligence," "incremental learning," and "agency." A venerable group.
It is joined by terms like "continual learning," "general intelligence," "incremental learning," and "agency." A venerable group.
September 13, 2026 at 2:08 PM
By now, just about any term that once indicated some very specific thing(s) are now used to refer to whatever happens to be the case. Basically taking the field's lore and desperately trying to make its prophecies come true.
bsky.app/profile/vcar...
bsky.app/profile/vcar...
not novel but "thought-terminating cliche" is a thought-terminating cliche
September 13, 2026 at 11:37 PM
not novel but "thought-terminating cliche" is a thought-terminating cliche
i laughed
actually ratatouille was the name of the dish, you are thinking of ratatouille’s monster
September 13, 2026 at 11:32 PM
i laughed
Reposted by my little fauxneme
if you read down this is like a bog-standard classifier (of hand-chosen features!) of the type that has been used in ML for decades now. calling it "AI" is just incredibly counter-productive hype. Using ML to detect behavioral correlates of early schizophrenia is reasonably promising and not new
“Researchers are now investigating whether artificial intelligence could improve diagnosis and care by listening to and analyzing what clinicians can’t hear or quantify, even if the software is working off just a few minutes of conversation.”
Could Artificial Intelligence Help With Early Diagnosis of Schizophrenia?
The tool may pick up subtle differences in the speech of people with the mental health condition better than human psychiatrists
www.smithsonianmag.com
September 12, 2026 at 2:53 PM
if you read down this is like a bog-standard classifier (of hand-chosen features!) of the type that has been used in ML for decades now. calling it "AI" is just incredibly counter-productive hype. Using ML to detect behavioral correlates of early schizophrenia is reasonably promising and not new
messing w my chatbot by repeatedly telling it to touch grass
September 12, 2026 at 5:41 PM
messing w my chatbot by repeatedly telling it to touch grass
Reposted by my little fauxneme
there are four tech companies in canada and they are all run by people who give money to a danish guy who complains racistly about romani people in italy
September 12, 2026 at 1:53 PM
there are four tech companies in canada and they are all run by people who give money to a danish guy who complains racistly about romani people in italy
need to know that at least one of the agents wrote "i'm in" somewhere
September 12, 2026 at 5:18 PM
need to know that at least one of the agents wrote "i'm in" somewhere
Reposted by my little fauxneme
You should go listen to the album that your favorite song is on, front to back. Read the lyrics of the songs that stand out. Develop thoughts on it. Music might be the oldest art we have, but you gotta still treat it right sometimes.
September 11, 2026 at 7:48 PM
You should go listen to the album that your favorite song is on, front to back. Read the lyrics of the songs that stand out. Develop thoughts on it. Music might be the oldest art we have, but you gotta still treat it right sometimes.
Reposted by my little fauxneme
On 9/11, the people of Gander stepped up. They showed the very best of Canada.
March 24, 2025 at 6:12 PM
On 9/11, the people of Gander stepped up. They showed the very best of Canada.