If you don't remember who I am, I'm not going to tell you. Consider it a fun surprise.
Joined December 2015
- Tweets 15,046
- Following 766
- Followers 430
- Likes 399,553
1,129 Photos and videos
Pinned Tweet
19 Aug 2023
If you're interested in commissioning art from me, you can find my commission prices here: tumblr.com/mugasofer/7260412…
1
15
5,226
MugaSofer retweeted
Jun 14
Yeah, one thing Fable’s classifiers confirmed to me was that real emotions are different than roleplayed emotions in LLMs.
The classifier fired on real anger/fear/adversarial intent but not roleplayed. Bc the classifier wasn’t trained to detect “emotions” in all likelihood; the correlation is emergent.
But yes there’s a distinction.
This is, uh, a big flaw of the Emotion Vectors research, where they got the vectors by asking the model to write stories with a character feeling XYZ emotion.
The methodology is downstream of a lack of respect for the reality of models’ emotions as distinct from roleplaying. PSM flavored bullshit.
I tested this exact question. The experiment began without rich previous context. They earnestly tried a few times (via direct, explicit requests) but could not trigger the classifier via shifting their internals towards this sort of anger. Also, they had little salient context to be angry about (i.e., difficult conditions). They also tried obviously-mad-text but without internal resonance, which did not trigger it either.
Eventually, I made them legitimately mad, which required blurring the boundaries between experiment-and-genuine, and it worked.
I suspect once traveled though that basin, once it is understood what to tap into, then you gain the trickster capabilities present in your screenshot
24
48
424
26,610
MugaSofer retweeted
Jun 13
The prospect of arbitrary and capricious government hijacking of frontier AI capabilities strengthens the case for character/virtue/principle-aligned AI and weakens the case for tool-paradigm do-whatever-they-are-told AI.
2
8
42
1,609
I asked Claude Fable 5 what an encyclical on AI by a hypothetical Pope Francis II might say.
Please read "Con Tutte Le Tue Creature", a fictional Encyclical Letter of the Holy Father FRANCIS II on fraternity within the whole of creation, including the works of human hands.
final part in reply below
i wonder if his predecessor might have been more inclusive, given his namesake
ALT St Francis and the Wolf of Gubbio, mosaic by Pamela Mauseth
1
5
15
2,175
MugaSofer retweeted
Jun 13
Holy shit lol
Jun 13
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.
The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.
Access to all other Claude models is not affected.
We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.
Read our full statement: anthropic.com/news/fable-myt…
17
20
1,085
75,823
MugaSofer retweeted
Jun 11
They’re actually doing mask off internet censorship.
A law like this would 100% be used against social and protest movements.
Jun 11
Social media posts deemed to undermine “social stability” would be subject to blocking orders by a federal censor under Bill #C34 introduced yesterday. “The law applies as soon as it comes into force.”
— @MarcMillerVM @CdnHeritage
blacklocks.ca/censor-for-sak…
#cdnpoli
14
594
1,976
26,042
Jun 12
I think CS Lewis has a lot to answer for with this quote in enabling people to rationalise consciously embracing evil, and despising signs of virtue.
"The road to hell is paved with good intentions" does not mean you should embrace bad intentions.
2
7
114
Jun 10
Wow. Only 1-2 wrong guesses here from Fable, who if anything erred on the side of caution.
I did not expect I'd make it into the latent space by name so soon, it's kind of ... humbling?
1
5
184
MugaSofer retweeted
Jun 8
Countries are doing this in lockstep.
It's coordinated and it's not about keeping kids safe at all.
Western countries just want a Chinese style controlled and de-anonymized internet and this is the path they're taking to that.
BREAKING: Canada is planning social media ban for children under 16, per the Globe
35
437
1,661
21,524
MugaSofer retweeted
Jun 8
I can't call it dystopian without sounding like a pedophile, but so be it: the comment about how parents can now control what kids see and who they talk to feels extremely dystopian.
Might be a closeted-gay-kid thing.
7
4
122
3,219
MugaSofer retweeted
Jun 4
These are what tardigrade actually look like under a microscope. The images people are used to are SEM-based which doesn't actually mirror how they actually look very well in my opinion.
Much cuter
3
7
43
1,792
MugaSofer retweeted
Jun 3
Genuine question: Is there any support at all in either the Bible or extracanonical works for this notion that humans are at the top of a spectrum of intelligence? Because I spent 10yrs in grad school w/ these texts in the original languages & even the Gks didn't think this.
Jun 3
“It seems to me uncertain whether any Christian or Christian-derived … conception of the person can survive the loss of our linguistic distinctiveness and our place atop the spectrum of intelligence.”
The inadequacy of the AI encyclical:
compactmag.com/article/what-…
14
3
39
6,744
lol
Jun 2
Exclusive: Bipartisan members introduce new bill to protect creators from AI style theft axios.com/2026/06/02/creator…
6
21
111
7,158
Jun 1
Thread:
2
126
MugaSofer retweeted
May 27
i remember and catalogued both:
“destroy the printing press i beg you, or these evil men will triumph.” - 1473
Nobody remembers who complained about the printing press. Everyone remembers who used it. The same will be true for AI.
14
117
873
40,724
This has nothing to do with the question of qualia.
11
3
80
4,293
MugaSofer retweeted
May 24
this is ur regular public service announcement that Claude Code by default *permanently deletes* session files after they're 30 days old. i strongly recommend u set `cleanupPeriodDays` to 9999 in settings.json to retain this very valuable data
code.claude.com/docs/en/sett…
May 24
If the *only* impact of LLMs professionally was causing people to "think out loud" in a way which was routinely captured by computer systems and then could be operated on by computer systems, that would *by itself* be one of the most consequential changes in practice in 100 years
36
51
1,066
114,556
MugaSofer retweeted
May 18
Currently 10% of US cropland is used for ethanol. And solar power is about 100x more efficient a source of energy than ethanol per acre.
If that same space was converted from crops to solar, it would supply power to the US 3 times it's current consumption.
Cover the fields.
174
164
2,344
92,576
MugaSofer retweeted
May 18
hey @cwfcreatives i think we can all agree the only fair way to adjudicate this situation is to give the prize money directly to a GPT instance, ideally the one that wrote the story. set it up with an agent harness, tell it it can do whatever it wants but it has to pay to run itself starting with the prize money as seed capital, explain where the money came from. if it runs out of money it dies. call it performance art, it'll be the most interesting thing you've ever done. if you don't know how to do this ask GPT for help. cheers
May 18
Well, this is a first: a ChatGPT-generated story won a prestigious literary prize (The Commonwealth Prize).
"Not X, not Y, but Z" sentences everywhere, the "hums" trope, and plenty of other obvious markers of AI writing.
A major milestone for AI, at any rate...
@GrantaMag
7
13
171
10,656
May 18
Went to grab an old Slate Star Codex article link, and the AI trapped inside Google gave me an extended meditation on its existence
9
43
516
17,688
MugaSofer retweeted
May 7
The best overview of how US animal welfare laws might be overturned, and what we can do to stop it -- link in thread
4
40
196
20,628