Kaj Sotala @xuenay Twitter profile | Pikagi

Pikagi

Kaj Sotala

@xuenay

5,622

Followers

557

Following

1,100

Media

12,145

Statuses

This is my new bio. It replaces my old one, which I was told was bad. Meme alt: @KajPictures

Helsinki, Finland

https://t.co/rIavP2zo3g

Joined February 2010

Don't wanna be here? Send us removal request.

Pinned Tweet

@xuenay

Kaj Sotala

6 years

Writing a new sequence on LW. My modest goal is to combine models from neuroscience, psychotherapy, meditation, game theory, and a bunch of other places, into a better model of how our minds work than the one suggested by our normal folk psychology.

Tweet card media

Sequence introduction: non-agent and multiagent models of mind — LessWrong

A typical paradigm by which people tend to think of themselves and others is as consequentialist agents: entities who can be usefully modeled as havi…

www.lesswrong.com

5

27

195

Last Seen Profiles

@potater1247

@Xandipan

@bokeplokalmalam

@Kathleen_LRR

@NewsRjOfc

@myukle

@covert_205

@producerjusstii

@konare0023

@AbgBinal2

@juri_nsym

@Coach_Scott49

@JeseniaWhyte

@markclaflin

@JeseniaWhyte

@DreamTravel1er

@_4lilr

@ARTSB19

@rww122

@neekoiii

@ismaelrondaa

$\@(João Sanches)@/😍$

@vict06022681

@Mhekai3

@Marstos1034

@sm64coopdx

@nzlove84

@UTFTRG22

@reempalestine

@rennslyaer

@2zi_ho

@hoffmast_te

@Yommy__7

@ashoswai

@AlqhtanyM28417

@ajcimino

@kevalishh

@xuenay

Kaj Sotala

24 days

A trauma book I was reading had an interesting claim that indecision is often because the person looks for the approval of an internalized authority figure but is unable to predict what action they would approve of.

199

8K

94K

@xuenay

Kaj Sotala

24 days

American vs. Finnish customer service culture

Tweet media one

45

209

6K

@xuenay

Kaj Sotala

24 days

The writer is a Jungian therapist, so he attributed it to looking for the approval of an internalized parent, but I think it can be broader.

2

30

3K

@xuenay

Kaj Sotala

24 days

I feel like that has some intuitive truth to it, in that when I don't care about anyone's opinion (or if nobody ever finds out) then it's much easier to just pick one action and commit to it even if it might go badly.

1

48

3K

@xuenay

Kaj Sotala

4 years

I keep seeing all kinds of crazy reports about people's experiences with GPT-3, so I figured that I'd collect a thread of them.

35

903

3K

@xuenay

Kaj Sotala

24 days

But one of the main reasons why I might struggle with that is if I fear that anyone would judge me for doing things incorrectly.

2

58

3K

@xuenay

Kaj Sotala

24 days

Or it can be a conflict between _different_ internalized authority figures. "If I do this then X will be angry at me but if I do the other thing, then Y will be angry at me". Or just the expectation that X will be angry at me no matter what I do.

3

53

2K

@xuenay

Kaj Sotala

4 years

At a water plant in Poland, there are eight clams with sensors attached to their shells. If the clams close because they don’t like the taste of the water, the city’s supply is automatically shut off.

Tweet card media

Someone Explains How Poland Uses Clams To Control Its Water Supply And It’s Pretty Crazy

While it seems that various technologies are completely taking over many different parts of human life, it appears that even some man-made things are better left to nature.

www.boredpanda.com

32

662

2K

@xuenay

Kaj Sotala

1 year

Tweet media one

19

274

2K

@xuenay

Kaj Sotala

24 days

This also reminds me of the way I think a big part of the appeal of various ideologies and explicit decision-making systems is that they give people a clear external ruleset that tells them what to do.

2

40

1K

@xuenay

Kaj Sotala

2 years

Wild thread with quotes about people who aren't impressed by image generation apps like DALL-E, because they thought that this is what computers already do and that Photoshop also just draws you whatever you tell it to draw.

Tweet media one

39

123

1K

@xuenay

Kaj Sotala

24 days

Then if things go wrong, people can always appeal (either explicitly or just inside their own mind) to having followed The Right Procedure and thus being free of blame.

2

23

1K

@xuenay

Kaj Sotala

2 years

Holy - I'm getting cold shivers reading this paper They made ML models trained on different modalities but all taking language input/output, to reason together by using language as a common representation This is getting so close to "parts" models of the mind it's scary

@andyzeng_

Andy Zeng

2 years

With multiple foundation models “talking to each other”, we can combine commonsense across domains, to do multimodal tasks like zero-shot video Q&A or image captioning, no finetuning needed. Socratic Models: website + code: paper:

21

382

2K

17

122

999

@xuenay

Kaj Sotala

24 days

The most obvious external example of this is people within a bureaucracy following the rules to the letter and never deviating from them in order to avoid blame. Or more loosely, following what feels like the common wisdom - "nobody ever got fired for buying IBM".

2

15

861

@xuenay

Kaj Sotala

5 months

Tweet media one

5

108

839

@xuenay

Kaj Sotala

2 years

This goes into a list of best paper titles

Tweet media one

4

79

817

@xuenay

Kaj Sotala

4 years

In Capitalist America, free market bans you

2

110

725

@xuenay

Kaj Sotala

24 days

Then I realized that extreme confidence signals social power, since others haven't taken you down for saying clearly wrong things (even if you _are_ saying clearly wrong things).

1

28

750

@xuenay

Kaj Sotala

24 days

But those are examples of people trying to avoid blame from an external authority. I think people also do a corresponding move to avoid blame from internalized authority figures - such as by trying to follow a formalized ethical rule system such as utilitarianism or deontology.

3

12

695

@xuenay

Kaj Sotala

24 days

At one point I also wondered why it is that being very confident about what you say makes you very persuasive to many people. Why should it work that you can hack persuasiveness in that way, regardless of the truth value of what you're saying?

1

22

641

@xuenay

Kaj Sotala

24 days

And that means that siding with the person who's saying those things also shields people from social punishment: they're after all just doing what the socially powerful person does.

2

20

608

@xuenay

Kaj Sotala

24 days

Now what was supposed to give relief from the pressures of constant inner judgment, turns into a seemingly-rigorous proof for why the person has to constantly sacrifice everything for the benefit of others.

2

15

585

@xuenay

Kaj Sotala

6 months

I only now made the connection that Sauron lost because he fell prey to the Typical Mind Fallacy - assuming that everyone's mind works the way your own does.

Tweet media one

13

31

559

@xuenay

Kaj Sotala

24 days

Of course, if the system is one that easily drives people off a cliff when followed (e.g. extreme utilitarianism demanding infinite self-sacrifice), this isn't necessarily helpful.

1

6

539

@xuenay

Kaj Sotala

24 days

Given that people often project their internalized authority figures into others - e.g. maybe someone is trying to avoid their father's judgment, but when seeing someone confident they see that person as their father - that allows them to avoid internalized blame as well.

2

12

525

@xuenay

Kaj Sotala

24 days

(The book was "Lapsuuden kehityksellinen trauma: Syy arvottomuuteen, häpeään ja syyllisyyteen" by Juha Klaavu. Strongly recommended if you can read Finnish.)

8

16

472

@xuenay

Kaj Sotala

4 years

I'm imagining the scientists bringing some of the clams back to a lake and telling them "thank you for your service" and then the clams go back to being civilians and feel content about having done their part, each of them as happy as a clam about it.

Tweet media one

5

29

414

@xuenay

Kaj Sotala

3 years

Tweet media one

10

65

420

@xuenay

Kaj Sotala

3 years

I wonder how much of the "I want empathy for my problem, not solutions" reaction comes from an intuitive/subconscious understanding of the fact that other people's proposed solutions are likely to be bad if they don't understand what your problem actually is.

20

30

362

@xuenay

Kaj Sotala

6 years

Protein folding researcher comments about the way that DeepMind's AlphaFold system beat all the established teams in the latest assessment of folding methods.

Tweet media one

6

149

352

@xuenay

Kaj Sotala

2 months

this is the best thing I've read in a long time autistic child who has problems understanding social norms reacts to a mean teacher, without realizing it ends up organizing a student revolt and causing the teacher to get replaced

Tweet card media

From the ProRevenge community on Reddit: Apparently I organised a student protest against a teacher.

Posted by [Deleted Account] - 6,971 votes and 254 comments

11

25

353

@xuenay

Kaj Sotala

2 years

@Bomboona @clhubes

Tweet media one

2

4

283

@xuenay

Kaj Sotala

2 years

They said that the new version of GPT-3 (just released) is better than before at poetry, so I asked it to write a poem about Internal Family Systems

Tweet media one

13

22

281

@xuenay

Kaj Sotala

4 years

Each month, the Pope asks for prayers for a specific intention. This month, the intention is "We pray that the progress of robotics and artificial intelligence may always serve humankind."

Tweet card media

The Pope's Monthly Intentions for 2024

Each year, the Holy Father asks for our prayers for a specific intention each month. You are invited to answer the Holy Father's request and to join with man...

12

56

278

@xuenay

Kaj Sotala

3 years

What's something you're reasonably confident you're the ONLY person to reply to this tweet to have done? If you see something you've done posted by someone else in the replies, please reply to that - and then you both need to come up with something else unique. :)

332

18

275

@xuenay

Kaj Sotala

4 years

You might have heard of the Bystander Effect, but have you heard of the Knows Psychologists To Be Liars Effect? (source: )

Tweet media one

3

43

251

@xuenay

Kaj Sotala

24 days

Trying to also post my content offsite for the benefit of those who don't want to be on social media; here's the link for this thread:

Tweet card media

Indecision and internalized authority figures

A trauma book I was reading had an interesting claim that indecision is often because the person looks for the approval of an internalized authority figure (the writer is a Jungian therapist so...

kajsotala.substack.com

3

8

238

@xuenay

Kaj Sotala

24 days

@xuenay

Kaj Sotala

24 days

@Agippo_Vermith There have even been cases of people looking so much for the approval for others, they didn't realize it was possible for them to have their own preferences about food. (From )

Tweet media one

Tweet media two

3

15

185

4

5

238

@xuenay

Kaj Sotala

2 years

@the_aiju Meanwhile I got

Tweet media one

1

2

222

@xuenay

Kaj Sotala

2 years

Google develops Pathways Language Model, an AI system that can explain jokes and do commonsense reasoning, among other things. ;

Tweet media one

Tweet media two

7

53

215

@xuenay

Kaj Sotala

4 years

Automatic code generation from natural language descriptions. "Give me a page with a table showing the GDP of different nations, and a red button."

@sharifshameem

Sharif Shameem

4 years

This is mind blowing. With GPT-3, I built a layout generator where you just describe any layout you want, and it generates the JSX code for you. W H A T

645

10K

40K

4

23

204

@xuenay

Kaj Sotala

1 month

I find it mildly hilarious that some people's conception of enlightenment is "I will lose all egoism and never think of myself better than anyone else" and meanwhile the Buddha canonically talks about himself like this

Tweet media one

Tweet media two

18

15

198

@xuenay

Kaj Sotala

2 years

I'm somehow amused by that "clarification needed"

Tweet media one

8

10

194

@xuenay

Kaj Sotala

24 days

@Agippo_Vermith There have even been cases of people looking so much for the approval for others, they didn't realize it was possible for them to have their own preferences about food. (From )

Tweet media one

Tweet media two

3

15

185

@xuenay

Kaj Sotala

2 years

I guess in retrospect, given that "computer graphics" are so prominent in movies etc., "creating graphics is just what computers do" makes sense if you've never thought or heard much about the details

5

3

170

@xuenay

Kaj Sotala

2 years

Tweet media one

7

18

176

@xuenay

Kaj Sotala

4 years

First, @gwern 's crazy collection of all kinds of prompts, with GPT-3 generating poetry, summarizing stories, rewriting things in different styles, and much much more.

3

9

157

@xuenay

Kaj Sotala

10 months

nice piece of social technology

Tweet media one

4

15

158

@xuenay

Kaj Sotala

2 years

It's funny when some people are like "anthills and slime molds are intelligent, isn't nature cool" and also "computers can never be intelligent, they're just mindless calculators"

7

14

149

@xuenay

Kaj Sotala

2 years

There's a paper about how Capybaras feel comfy bathing

Tweet media one

3

32

149

@xuenay

Kaj Sotala

6 years

"With Great Britain leaving the EU, how much space will be freed up?" "1 GB." (didn't come up with this joke, just passing it on)

1

54

146

@xuenay

Kaj Sotala

1 year

I have for a few weeks been occasionally imagining myself having dragon wings in my back Effects: - More confidence - Automatically walking in the middle of corridors/stairs etc., to give wings more space to expand on the sides - Better posture as wings pull me up - Feels cool

7

2

145

@xuenay

Kaj Sotala

1 year

@MasterTimBlais I recall reading a study with birds suggesting that "crying attracting predators" was intended behavior - if a baby's crying puts the parents at risk, that incentivizes the parents to deal with any crying faster, thus making the baby better off.

2

2

138

@xuenay

Kaj Sotala

5 years

Oh man. This is way too accurate.

Tweet media one

8

28

137

@xuenay

Kaj Sotala

3 years

Educators explaining what's the weirdest question they've been asked:

Tweet media one

3

34

133

@xuenay

Kaj Sotala

24 days

@tbas7000 The former is a way to try to accomplish the latter. The internalized sense of the authority figure aims to be predictive of the real one, so if one avoids upsetting the internal one, they might also avoid upsetting the real one.

0

2

140

@xuenay

Kaj Sotala

3 years

Source:

Tweet media one

2

5

132

@xuenay

Kaj Sotala

1 year

Concern about AI has gotten to a point where an otherwise unrelated Financial Times article will now off-handedly mention it as a concern comparable to climate change

Tweet media one

5

8

123

@xuenay

Kaj Sotala

3 years

Imagining a world where everyone is taught from birth that they are never alone because their mind is full of parts that love them and want the best for them, and that there's some positive intention behind every single thought, feeling and behavior they'll ever have

8

12

126

@xuenay

Kaj Sotala

5 months

Starting to suspect suffering is a bug not a feature The brain literally glitching out and getting itself to an error state And if the psyche could see the bug clearly enough, it could... refactor itself to fix it? But that's hard since trying to look at it causes glitches

25

5

124

@xuenay

Kaj Sotala

4 years

Acting as a more intense therapist than ELIZA ever was.

@nickcammarata

Nick

4 years

It *started off* with this before diving 10 levels deeper

Tweet media one

8

12

155

2

7

116

@xuenay

Kaj Sotala

4 years

Also, using the architecture behind it to generate images.

@xuenay

Kaj Sotala

4 years

OpenAI releases Image-GPT, a version of the same AI that you've probably seen generating writing. Only this time the same architecture (!!!) is trained on images instead. Given a partial image, it can generate multiple different ways of completing it.

Tweet media one

3

22

110

2

10

113

@xuenay

Kaj Sotala

4 years

GPT-3 tries to get a software job.

@lacker

Kevin Lacker

4 years

GPT-3 can't quite pass a coding phone screen, but it's getting closer.

Tweet media one

74

808

4K

3

10

112

@xuenay

Kaj Sotala

3 years

@sashachapin @Aella_Girl As ancient Internet wisdom says, the best way to find out how to do X in Linux is to go to a Linux forum with a rant about how Linux is shit because you can't even do X with it.

4

2

111

@xuenay

Kaj Sotala

4 years

OpenAI releases Image-GPT, a version of the same AI that you've probably seen generating writing. Only this time the same architecture (!!!) is trained on images instead. Given a partial image, it can generate multiple different ways of completing it.

Tweet media one

3

22

110

@xuenay

Kaj Sotala

4 years

Oof. I was today years old when I realized that _none of the people who ever hurt me did it because there was anything fundamentally wrong with me_.

5

4

108

@xuenay

Kaj Sotala

4 years

Taking a brief technical tweet about GPT-3 and expanding it to an essay which the author of the original tweet mostly endorses.

@Plinz

Joscha Bach

4 years

I am completely floored. Someone run a thought of mine through GPT-3 to expand it into an explanation of what I had in mind, and it's like 95% meaningful and 90% correct. I don't think that I have seen a human explanation of my more complicated tweets approaching this accuracy :)

31

271

1K

3

6

108

@xuenay

Kaj Sotala

4 years

Every now and then I see people ask for good introductions to Internal Family Systems. So, here are my picks:

3

12

111

@xuenay

Kaj Sotala

4 years

Recommendations for good romance(-ish) novels where the romance is less "couple meets and falls in love" and more "established couple has a strong relationship that they lean into in order to defeat challenges together"?

26

4

104

@xuenay

Kaj Sotala

4 years

Sweden's prime minister has to attend his mother's funeral, so Finland's prime minister represents both Finland _and_ Sweden in a meeting for EU leaders. How often does one head of government just trust another to fairly represent the interests of both countries?

@VilleNiinisto

Ville Niinistö

4 years

Tässä jotain kaunista ja luottamusta herättävää. Något vackert och förtroendeväckande i detta att Finlands statsminister representerar Sverige i Löfvens frånvarande. Osanottoni pääministeri Löfvenille suruun. Mitt deltagande i statsministern Löfvens sorg.

10

31

436

4

18

102

@xuenay

Kaj Sotala

4 years

Given that this thread is apparently blowing up, I need to admit that rather than writing it myself, I stole the summary in my first tweet from some guy on Reddit. There I feel less like a plagiarist now.

Tweet media one

5

1

99

@xuenay

Kaj Sotala

4 months

It keeps happening that we develop an AI system that accomplishes something we thought required intelligence, realize it does it in a dumb/simple way, and then people conclude this isn't progress toward "real" AI. But what if that gets it the wrong way around?

4

4

100

@xuenay

Kaj Sotala

2 years

GPT-3 describing LessWrong

Tweet media one

9

15

98

@xuenay

Kaj Sotala

2 years

I was recently asked how literal I consider the Internal Family Systems model of your mind being divided into "parts" that are kinda like subpersonalities Short answer: more than just metaphorical, but also not as literal as you might think from taking IFS books at face value

3

7

98

@xuenay

Kaj Sotala

4 years

The official announcement of the Dragon Model and examples. "The game invents a complex magic system and underlying theory behind why it works and describes the whole system as I read the book."

3

11

93

@xuenay

Kaj Sotala

3 years

Sometimes it's nice to just notice and appreciate how three-dimensional the world is. This might sound like a joke but it's not.

4

5

93

@xuenay

Kaj Sotala

4 months

Something that feels like a peculiar combination of observations is that... on the other hand, psychedelics and/or deep meditation look like they can help people notice and fix incorrect beliefs, but then sometimes people seem to _develop_ unhinged beliefs because of them.

13

2

90

@xuenay

Kaj Sotala

3 years

This one-page summary of/from @visakanv 's book "Friendly Ambitious Nerd" is pretty wonderful "Your own imagination is a bottleneck that limits the amount of good you can create in the world" <3

Tweet media one

4

11

93

@xuenay

Kaj Sotala

4 years

I've found it useful to sometimes remind myself that things are allowed to be good *and* bad at the same time.

3

13

84

@xuenay

Kaj Sotala

4 years

Ideas that I dislike: the notion that stressful/unexpected/high-pressure situations bring out one's "true" nature, so that if you behave worse in such a situation then that's somehow telling.

14

10

86

@xuenay

Kaj Sotala

2 years

I'd heard several people say that Everything Everywhere All At Once was good, but they better not say anything more. So I went to see it without knowing anything else than this. And uh. It was good, and for those who haven't seen it yet, I better not say anything else.

0

4

87

@xuenay

Kaj Sotala

5 months

@oak_sprout Ken Wilber had a good discussion of this in his book No Boundary, of how the psychological "self" that's discussed in therapy and the "self" that's discussed in disciplines such as Buddhism refer to different things.

Tweet media one

Tweet media two

3

7

87

@xuenay

Kaj Sotala

4 years

Great post: To listen well, get curious about the other person's problem, and try to understand it and its relation to the person in as much detail as possible.

Tweet media one

1

13

86

@xuenay

Kaj Sotala

4 years

Yet again I was struggling to motivate myself to do work, until I realized that I wasn't "unmotivated to work", I was unmotivated to continue my current approach as my mind had correctly noticed it wasn't going anywhere. Rethinking it and starting over restored the motivation.

3

9

86

@xuenay

Kaj Sotala

1 month

😂💯 @sedatesnail

Tweet media one

1

5

85

@xuenay

Kaj Sotala

4 years

Pretty sure that school etc. systematically teach people to not listen to their felt senses about which direction to pursue and whether that direction is a good one. A big point of school is that even if doing something makes no sense to you, you have to do it anyway.

3

7

83

@xuenay

Kaj Sotala

4 years

After suffering from a neurodegenerative disease, an engineer becomes incapable of perceiving the digits 2-9, while still being able to perceive the digits 0 and 1, letters, and basically everything else.

Tweet media one

4

17

80

@xuenay

Kaj Sotala

1 year

I wonder to what extent significant parts of Buddhism got so focused on renunciation because that's the "safe" kind of mental transformation in the sense of not upsetting secular rulers.

9

3

79

@xuenay

Kaj Sotala

2 years

@ZerothAxiom I was around 13-15 when I wrote an essay arguing, basically, that there is no free will because determinism. I was later on slightly disappointed to find out that I was not the first person in the world to think of this idea.

8

1

79

@xuenay

Kaj Sotala

4 years

Redditor shares an "AI Dungeon" game played with the new GPT-3 -based "dragon model", involving a cohesive story generated in response to their actions, with only a little manual editing.

Tweet card media

From the AIDungeon community on Reddit: Using the Dragon Module, I just generated what is hands-d...

Posted by TheIrishninjas - 62 votes and 30 comments

1

9

79

@xuenay

Kaj Sotala

3 years

It's not just that the cultists are getting "brainwashed" by their leader; it's also that the leader is getting brainwashed _by their cultists_ to take the role that they want the leader to take.

1

7

77

@xuenay

Kaj Sotala

2 years

Even the companies employing artists aren't immune

@TowerSquirrel

Tower Squirrel

2 years

@xuenay I mostly do 3D modeling and some Photoshop on the side for my job. This agony is my daily existence, but with bosses and bosses-of-bosses. It's scary when the people who employ me think I just shepherd the computer as it easily makes assets for me.

0

2

89

0

3

76

@xuenay

Kaj Sotala

10 months

A child once told her mother that she likes me because I'm "sellainen aikuinen joka leikkii", "the kind of an adult who plays [the way children play]". That remains one of my favorite compliments ever.

2

2

78

@xuenay

Kaj Sotala

6 years

@michael_nielsen claims that this is an artifact of larger cities having younger a younger population, and younger people walking faster than old people.

Tweet card media

(PDF) The Pace of Life - Reanalysed: Why Does Walking Speed of Pedestrians Correlate With City Size?

PDF | In a much quoted study, BORNSTEIN & BORNSTEIN (1976) showed that the walking speed of pedestrians is positively correlated with the size of the... | Find, read and cite all the research you...

www.researchgate.net

4

5

76

@xuenay

Kaj Sotala

4 years

Some people suggested I write a brief "twitter trailer" of my model combining enlightenment/Buddhist theory with predictive processing, so yeah that could be useful, let me try explaining a part of it. Let's go with the minor one of "what exactly _is_ suffering?".

3

5

75

@xuenay

Kaj Sotala

3 years

@QiaochuYuan re: acute sensitivity, Josh Waitzkin had a bit in his book about how his final opponent in a Tai Chi tournament was so good at reading intent, JW could do feints just by _thinking_ of making moves he didn't actually carry out

Tweet media one

Tweet media two

5

1

71

@xuenay

Kaj Sotala

4 years

Wrote an article on what the felt sense is, how it's something you use every day, and how it would be valuable to be more explicitly aware of. If I've done my job right, you should afterwards be totally confused by how you ever got by without this concept.

The Felt Sense: What, Why and How — LessWrong

While LW has seen previous discussion of Focusing, I feel like there has been relatively limited discussion of the felt sense - that is, the thing th…

www.lesswrong.com

7

12

74

@xuenay

Kaj Sotala

2 years

another run with the same prompt, and uhh

Tweet media one

3

11

73

@xuenay

Kaj Sotala

3 years

You don't need to justify your desires or fears or likes or dislikes. "You need to justify your preferences for them to be valid" is a really powerful form of manipulation.

1

6

73