nihilfit.bsky.social
Cognitive scientist @ A*STAR IHPC; erstwhile philosopher; infrequent poster, but damn, look at that signal/noise ratio!
93 posts
93 followers
77 following
Regular Contributor
Active Commenter
comment in response to
post
But with more singularity of purpose! Nobody in Akihabara was there to gawk. You were there to find that one damn replacement battery cover or obsolete PDA cable adapter, and somewhere in one of those buildings there was one man waiting to grumpily sell it to you.
comment in response to
post
Oh man, you’re making me miss the 90s version of Akihabara (when it was for IT otaku rather than anime/maid otaku). Multiple buildings with floors full of stalls, each stall run by a dude who lives for that one component and has like 400 variants sorted in little bins.
comment in response to
post
But incidentally, gamerqnt is owned by the same ghoul who just bought and destroyed Polygon, so best to avoid anything from that domain.
comment in response to
post
UHS was amazing! It’s a tragedy that that approach seemed to die out.
Ironically, LLMs would probably be good for that sort of thing: “Read this GameFAQs document and tell me whether I’m supposed to be able to get through the metal door in Act 1, with no spoilers about how to do so”
comment in response to
post
“Separate social spheres” sounds like something more than just “separate interpretive categories” — is this about discouraging chatbot makers from tuning them to present as humanlike, or …? I’m curious!
comment in response to
post
Hmm. I was nodding along for your first few points, which I took to be about the user’s interpretive stance (and how that stance can feed back into the dialogue with an LLM). But this last point seems to be gesturing at something else. Could you say a bit more?
comment in response to
post
The closest thing in that neighborhood that’s well-supported is setting contextual expectations — eg in this room / during these hours I can use the phone, and in this room the phone doesn’t come out. Still painful to self-enforce, but once set they do make it easier to focus.
comment in response to
post
Sadly it didn’t have any real basis in empirical studies or neuroscience — though like any lifestyle disruption, it might help to kick you at least temporarily out of your current habits and give you a chance to forge new ones.
comment in response to
post
It also features a cover of the original TMNT theme, by … Mike Patton 🤯
comment in response to
post
Whoops I was talking about Shredder’s Revenge! Didn’t realize there’s also a TMNT roguelike.
comment in response to
post
TMNT is a great brawler, executed by people who know and love the genre. If the trailer appeals to you at all, then it’s a safe buy.
comment in response to
post
"... although I cannot discuss this further at the moment."
Mostly I prefer the modern/American communication style, but there are certain forms of beauty that can only exist outside of it.
comment in response to
post
Tasting Middle C?
comment in response to
post
Anyway I'm firmly on team "worry about the owners, not the AIs", so I don't have much of a stake in these hypothetical-superintelligence-safety debates. But even just purely taken on its own terms, this one just stopped me dead, so I had to share.
comment in response to
post
And it's Neal Stephenson! Not a dim guy!
I'm left wondering if he's secretly on Team Basilisk? Unlikely, I know! But why *else* would he try to persuade us that the way to mitigate the risk of dangerous super-AIs is to *teach them to fight, then evolve them under conditions of brutal competition*?
comment in response to
post
The funny thing is, your first sentence is @qntm.org’s short story Lena, which I see someone linked below … and your second sentence is the sequel-story “Driver”, from the same collection. Both highly recommended!
comment in response to
post
Really wanted “the failure” rather than “a failure” above, but had to sacrifice it to the character limit; apologies in advance for any semantic whiplash 🙇
comment in response to
post
1h a day is a lot! It’s enough, if you keep going. Just be sure to look for opportunities to speak; remembering words and patterns is much much harder if you only hear or read them.
comment in response to
post
Did not expect a Kobo Abe reference to pop up here! Does Kojima say whether the use of boxes for stealth is meant to be part of the metaphor?
comment in response to
post
Wonderful colors! I would never have called the bright green radiator against that wall color, but it totally works, bringing the brightness of the window into the room. Cool.
comment in response to
post
“… shared Miike excitedly.” 😄
comment in response to
post
(Fellow video-haters, be reassured: it’s an information-dense ten minutes, and all the core ideas are covered within the first four.)
comment in response to
post
I try not to overhype this stuff, but this might be the biggest government opsec failure in history.
comment in response to
post
For sure. And the thing is, training in just *one* of those disciplines (or math, or physics) tends to leave people more vulnerable to it, not less.
comment in response to
post
Thoughtful people with at least a bit of reading in any two out of <bio / phil / history / anthro / socio > will reliably overcome it. But that will only ever be a tiny minority. Making a competing mass ideal out of progress and change is one proven counter-move; are there any others?
comment in response to
post
And yeah, the interactions with Qualtrics’ sales & billing folks did have that monopolist feel: more concerned with policing potential infractions than with retaining customers.
comment in response to
post
A few years back my department had a mass exodus from Qualtrics (mainly due to lack of support for team-based work), and now everyone I know of is on Alchemer (nee SurveyGizmo). Cheaper, simpler and more than configurable enough for our purposes.
comment in response to
post
Now I’m imagining Peter Parker with PTSD, spider-sense pinging constantly …
comment in response to
post
(As a former logic and stats teacher, this drives me up the wall. But the appeal is undeniable. It’s epistemic fantasy!)
comment in response to
post
it would be extremely funny if the maximalist version of the "if you train a model to write insecure code it becomes evil" paper were true and it's just that truth, beauty, and goodness are entangled to such an extent that an evil, competent LLM is a logical impossibility.
comment in response to
post
Inaccessible things that can still stir imagination/fantasy even when you understand that actually experiencing them would just be unpleasant?