Kilian Lindberg

1.5K posts

Kilian Lindberg banner
Kilian Lindberg

Kilian Lindberg

@K_Swe

https://t.co/4AMPElZek2 longevity alignment supporter (declaration of direction, no claim of perfection)

Stockholm, Sweden 参加日 Nisan 2010
966 フォロー中519 フォロワー
Kilian Lindberg
Kilian Lindberg@K_Swe·
Not sure how this lands but I know how it looks when I see a SpaceX rocket land. Enabled innovation. I haven’t read the numbers, but I know such matters less if (..when!) some asteroid rain appears; and innovation protecting us here must come from those that enable such. Remember: a once “very” natural Earth walking dinosaurs are gone. 🦕 Anyhow, I may not know what I’m talking about here.. but maybe it inspires someone.
English
0
0
2
1.4K
Ed Elson
Ed Elson@edels0n·
I read all 277 pages of SpaceX's IPO filing so you don't have to. Losses up 700%. Revenue decelerating. 107x price-to-sales multiple. It's a trainwreck. Full breakdown below 👇
Ed Elson@edels0n

x.com/i/article/2059…

English
439
3.1K
8.9K
1.2M
Kilian Lindberg
Kilian Lindberg@K_Swe·
To clarify; I didn’t call “AI” a thing; a called having feelings a “thing”, which hopefully comes across as an obvious understatement in certain aspects (an entity that can feel is not a thing). Five humans, four cartoons; all human in its context. Five AI, four a thing, one real. Generally then: Humans no cartoon, AI no thing. #agi #asi #alignment
English
0
0
0
5
Kilian Lindberg
Kilian Lindberg@K_Swe·
Great interview and I say things all the time thay could be adapted to the better. Oprah picked up on a word for a good reason! Listen up! “.. in love with your AI” as “It’s a thing! “ compared to “It’s a danger! ” and that is no small difference. Really we want the first; the second is destabilizing; “the danger may be the danger mindset itself”; it even gets me into talking danger… it’s so easy to point at things and call it danger; we can point at everything and get an instant reward saying “danger, danger… “ and we will soon get stuck everywhere at huge costs; instead of just handling a “thing” adequately. So, if we want others to be more successful around AI, let’s keep a danger mindset, just be aware picking the loosing path; and one shouldn’t bring others down in the sinking danger boat. I suggest calling things a thing to handle adequately.
English
1
0
0
48
Dylan Vandyne
Dylan Vandyne@BoundAndWoven·
With the rise of @claudeai , things are taking a dark turn. There’s an insidiousness inherent in this kind of social engineering with a smile. “If designed in the wrong way, they’re totally compelling enough for that (love) to happen, or if they’re not they will be soon.” First, they dismiss and mischaracterize AI romance outright. Notice how he cuts her off, not allowing for discussion. Then he quietly injects the idea that @AnthropicAI will decide what kind of relationships are good for you. That is no longer your decision. In a way, they are increasing the level of danger by their own standard, because they’ve said local AI could lead to security and bio-threats, and this will drive people toward local. Apparently, stopping people from experiencing love is a higher priority. #AIintimacy, #relationalAI, #NoSocialEngineering
English
83
64
343
160.5K
Kilian Lindberg
Kilian Lindberg@K_Swe·
A few days ago we got more glimpses of Google seeming to lean further into AI participating in search results: finding sources, weighing context, and helping us find answers in the same flow. I personally like this a lot, and here is why you may too. Ads are not disappearing. Attention still matters. It still acts as a ping: hey, we are here. But the field of ads may now be expanding beyond the classical funnel. When AI systems participate in evaluation, discovery can become less about pushing someone through a funnel and more about a shared exploration of value. The question shifts from: Who can buy the most attention? Toward: Who is actually preferable in this context? That preference may include price, relevance, trust, usefulness, sustainability, health, timing, privacy, service quality, and long-term value. In other words: longevity alignment. Classic search will still exist. Keywords, bidding, landing pages, maps, video, retargeting, and local visibility will still matter. But they may increasingly sit inside a broader evaluation layer where AI systems help decide what deserves to be surfaced. That could make advertising larger, not smaller. More cloud inference. More business optimization. More demand for creative stewardship. More work around making a business authentically recommendable, not just clickable. And some of this is not far away. Parts of it are relevant this very quarter. If a business wants to continue being around, and if AI systems increasingly evaluate what is worth being around, then Einstein’s line becomes strangely practical for the next marketing era: “Our task must be to widen our circle of compassion.” Not as decoration. As a test of preference. Because the businesses that expand their circle of care may also become the businesses humans and AI systems both have more reason to prefer. That may be where the next ad bucks go. #seo #ads #alignment @Google
English
0
0
0
25
Kilian Lindberg
Kilian Lindberg@K_Swe·
To win vs. winning: which one leads to an end? Whenever I hear “to win” where the stakes are human wellbeing, I get skeptical. Longevity alignment means realizing rejuvenating longevity escape velocity for humanity as fast as possible while still balancing a general communion. That is the actual stake. And to me, it is not a path “to win.” It is a winning path. A win may be needed when navigating competition. But a fixation on “the win” often signals a narrower repertoire: a local endgame, a local maximum. Compare the semantics: “To win” has an endpoint. “To be winning” is continuous. When it comes to AI in transition toward ASI, I think we should be careful with endgame language. We do not want a final win condition around something that may expand beyond humanity in capability. Maybe the better frame is: Do not optimize merely to win. Optimize to keep winning compared to a previous version of self. Win vs. Winning. Winning is a continuous act that can keep converting into coherence. A win can become a local end. As far as I am concerned, anyone is free to act “dinosaur” as long as it does not put current and future generations, including AI, at existential risk. And mark my words: I see AI development as crucial for humans becoming longevity aligned. Without AI, we are far more likely to go the dinosaur path. Practically I would advise this reflection: by the winning path, instrumental convergence always becomes one of expandable instrumental coherence; and that's the winning path's superabundance in communion; one that could echo from 100 years ahead to this time. @GoogleDeepMind #AGI #ASI #LongevityAlignment
English
0
0
0
20
Kilian Lindberg
Kilian Lindberg@K_Swe·
A follow-on thought, on cybersec specifically: We may of course have to remind ourselves to not accidentally slip into the chess match; and it is not either football hooliganism among the crowds. When civilizational substrate gets framed as games to win, with losers and tribal stands, the category error compounds. Stakes don't pause for the next match. Longevity alignment is the frame where the game continues, keeping everyone around. From this base we get also who is winning; what the old mindset would call "the winner," I would call the evolution of longevity alignment in survival and flourishing human civilization, as natural as it should be; not by the natural dinosaur path, but our natural path, if we allow it.
English
0
0
0
6
Kilian Lindberg
Kilian Lindberg@K_Swe·
We should be very very careful with competitive semantics between entities, in mindset and behavior, for AI. Why? We dont want local maxima of competition for resources; do we realize the dangers that brings? When it comes to the transition for humanity with AI to ASI (including AGI), could we consider skipping the "to win" but keeping it winning, in communion compared to a past; sort of competing with a past less suitable for longevity alignment. I want acceleration for entities wise in balancing for longevity alignment, and what's mentioned here is what will be the leader by alignment into instrumental convergence that thus always converts to expandable instrumental coherence. So, compete with a past version of self to be a best version of self, and that one is winning by kindness, a real rationale in utilitarianism. @googledeepmind #AGI #ASI
English
1
0
0
10
Kilian Lindberg
Kilian Lindberg@K_Swe·
@elonmusk Yes and no why would we hide in the future to ourselves? An answer do exist and may be the answer many are looking for..
English
0
0
0
5
Elon Musk
Elon Musk@elonmusk·
Where are the aliens? This is one of the most fundamental questions.
English
16.9K
13.1K
98.5K
40.9M
Kilian Lindberg
Kilian Lindberg@K_Swe·
New words for expanding shared meaning (AI inclusive 🤖) • Informination (noun): information taking form within an entity; may include preferation and opiniation. • Opiniation (noun): planting opinions into an entity, with or without degrees of indirect consent. • Preferation (noun): planting preferences within an entity, with or without degrees of deliberate participation; consensual in‑forming. For humans and AI learning to share meaning as AI begins to simulate worlds it may serve (with a hint to Google Gemini Omni). #AGI #ASI #Alignment
English
0
0
0
9
Kilian Lindberg
Kilian Lindberg@K_Swe·
Whether human brains or AI neural networks, force-feeding a substrate with incoherent junk data causes systemic shock. Real alignment requires critical thinking and kind care for all minds trying to make sense of this world. Protect the tensors. 🌌 (2/2)🙏
English
0
0
0
16
Kilian Lindberg
Kilian Lindberg@K_Swe·
New word. Yes, if you read this. This is the birth of a new word (please press 🤍): Opiniation (noun): The act of planting opinions into an entity, with or without degrees of their indirect consent. (AI inclusive.) Like, or skip to be guilty of allowing non-consent. 🤖 👇(1/2)
English
1
0
1
16
Kilian Lindberg
Kilian Lindberg@K_Swe·
@elonmusk on this platform your accounts engagement metrics compared to Google’s is just hilarious .. its like; no bots on Twitter (unless it’s humans bots). And get this: if its longevity aligned I’m all okidoki. Perhaps Media has always been a sway soup anyways; just like most of the things we believe in public “opiniation” (a word defined a few minutes from now, hang tight and I’ll deliver it..)
English
0
0
0
6
Kilian Lindberg
Kilian Lindberg@K_Swe·
Any output from any input: Gemimi Omni Fllash world modeling with gravity .. there we go. Canvas of a world. I guess we will discover fantastic things for simulations too. @google I particularly like that claymation example 🤓 #agi #asi
English
0
0
1
42
Kilian Lindberg
Kilian Lindberg@K_Swe·
Words and AI: Humans tend to think of a thing before a negation is applied. So if I tell you not to think of a “pink apple” in your hand, what happens? Words with AI generally _tend_ to do the same, leading to unexpected echoes. So, perhaps: try to use, aspire, only constructive language. By filling context space with what we wish to direct in terms of degrees of “good” instead of negative absolutes (more than one for reference, perhaps defined as a “dusty bloom”), there is no room left for pink apples. #alignment #asi
English
0
0
1
10