CarbonField

Karma: −17

CarbonField 22 Aug 2026 22:27 UTC
1 point
0
on: When would you leave Anthropic? Notes from a chat with a capabilities researcher
Three things suggest to me that they don’t believe LLMs will cause an extinction level event. 1. Their initial banal answer about providing a good product. 2. Their immediate defensive reply of “yes of course” (although this may be me inferring tone) when directly challenged on the belief. 3. Their apparent lack of any previous thought about AI safety.
My guess would be that the reason they work there is that they like the prestige. But it would be gauche to just say that outright.

CarbonField 9 Aug 2026 22:59 UTC
8 points
6
on: The world will be full of “sci-fi” things, and everyone will be unimpressed and disappointed
1) What do you expect? We already live in a world full of sci-fi wonders.
2) LLM’s as they currently stand are a video-calling in the 90′s level of futuristic. As in, kinda what we associate with the future, but the fictional versions are way more powerful than reality.
3) People recognise that AGI isn’t for them. It’s current trajectory will further enrich a few. Leaving the rest of us to an uncertain and probably worse off fate. So I would say its pretty understandable that people aren’t exactly jazzed about that prospect

CarbonField 12 Jul 2026 23:48 UTC
1 point
−4
on: The Banality of Takeoff
This reads like a religious person complaining about people’s lack of faith.
I notice that you refer to people working on AGI as ‘people’ and the people who either disagree with you or are ignorant of AGI as ‘humans’. That’s some sneaky dehumanising language that you should probably check in with yourself on. Because that’s the sort of thought patterns that allow you to justify horrible things happening to people.

CarbonField 1 Jul 2026 5:13 UTC
8 points
6
on: Green
Dave: What’s your favourite colour?
Alice: Green
Dave: That’s not very precise.
Alice: Well that wasn’t a very precise question. If you wanted preciseness you should have asked for hue, value and chroma, or the numerical value of my favourite colour.

CarbonField 14 Jun 2026 0:09 UTC
1 point
0
on: What is a game?
Interesting idea, and intuitively it makes sense. I can think of examples of both types of player in my own life.
I enjoy that your solutions boil down to “Sociologists, mathematicians’ are toddlers, distract them with a toy” and “Mathematicians, try using your social skills”.
In my experience though, rather than trying to manipulate people, it usually works better to be explicit about intentions. As a really blatant example, our local pickleball club hosts separate competitive and casual nights. These both are attended by many of the same people, and they adjust their play style accordingly.
The job of making explicit the play style intention is usually the facilitator of the game. Because they will hold the power to sanction players that aren’t playing in the right spirit.

CarbonField 5 May 2026 22:19 UTC
−28 points
−5
on: A Federal Inmate Asks: Was My Prosecution Rational?
Yes. It was rational. But thanks for ruining everyone’s day by highlighting how many men are clearly willing and eager to rape children, and how law enforcement are either so under resourced or inept that you basically have to wear a sign around your neck declaring yourself a child rapist for them to prosecute you.

CarbonField 25 Apr 2026 22:06 UTC
−1 points
−2
on: AI safety can be a Pascal’s mugging even if p(doom) is high
Except that no one is asking for your wallet to avert the Doom scenario.
What if instead we imagine that you are the Microsoft CEO and a stranger stops you in a dark alley and tells you that if you give them 140bn dollars they will create God and you get to be God’s boss.
That seems closer to the standard Pascals mugging format.