@cutterferry.bsky.social
Interested in mathematics, computer science, history, political thought, and the (lack of) "ethics" of AI. Anonymous because questioning AI is career limiting behaviour.
Makes one regret waxing lyrical about how interesting it is!
January 19, 2026 at 3:28 PM
Makes one regret waxing lyrical about how interesting it is!
If only we could spot a shift in the structure of the media landscape that makes challenging fascism harder and the world more incessantly overwhelming, like the news was prodding you for attention every few seconds…
Anyways I’m glad we have social media since before there was appaz no free speech
Anyways I’m glad we have social media since before there was appaz no free speech
January 19, 2026 at 9:20 AM
If only we could spot a shift in the structure of the media landscape that makes challenging fascism harder and the world more incessantly overwhelming, like the news was prodding you for attention every few seconds…
Anyways I’m glad we have social media since before there was appaz no free speech
Anyways I’m glad we have social media since before there was appaz no free speech
Being millennial means remembering when Apple’s pitch was
“Buy our computer it’s pricy but bas a rainbow!”
Not
“Buy our phone it’s pricy but it’s made by ~slave labour in China so we’ve at least minimised *our* cost base. PS we shipped all the mfr IP there so can’t do it without them now”
“Buy our computer it’s pricy but bas a rainbow!”
Not
“Buy our phone it’s pricy but it’s made by ~slave labour in China so we’ve at least minimised *our* cost base. PS we shipped all the mfr IP there so can’t do it without them now”
January 19, 2026 at 6:57 AM
Being millennial means remembering when Apple’s pitch was
“Buy our computer it’s pricy but bas a rainbow!”
Not
“Buy our phone it’s pricy but it’s made by ~slave labour in China so we’ve at least minimised *our* cost base. PS we shipped all the mfr IP there so can’t do it without them now”
“Buy our computer it’s pricy but bas a rainbow!”
Not
“Buy our phone it’s pricy but it’s made by ~slave labour in China so we’ve at least minimised *our* cost base. PS we shipped all the mfr IP there so can’t do it without them now”
Being millennial means remembering when Google’s marketing pitch was
“Hey we just sell ads we’re pretty innocent would you like cool free stuff”
Not
“we are a key monopolist in an incipient fascist superpower and respond to any attempt at regulation with geopolitical blackmail OPPOSE US NOT”
“Hey we just sell ads we’re pretty innocent would you like cool free stuff”
Not
“we are a key monopolist in an incipient fascist superpower and respond to any attempt at regulation with geopolitical blackmail OPPOSE US NOT”
January 19, 2026 at 6:48 AM
Being millennial means remembering when Google’s marketing pitch was
“Hey we just sell ads we’re pretty innocent would you like cool free stuff”
Not
“we are a key monopolist in an incipient fascist superpower and respond to any attempt at regulation with geopolitical blackmail OPPOSE US NOT”
“Hey we just sell ads we’re pretty innocent would you like cool free stuff”
Not
“we are a key monopolist in an incipient fascist superpower and respond to any attempt at regulation with geopolitical blackmail OPPOSE US NOT”
Being millennial means remembering when Microsoft’s Marketing pitch was
“We are unstoppable and ruthless and will use computers to suck the joy out of any human activity that can yield a profit. Oppose us and die.”
Still is they’ve always been like that
“We are unstoppable and ruthless and will use computers to suck the joy out of any human activity that can yield a profit. Oppose us and die.”
Still is they’ve always been like that
January 19, 2026 at 6:43 AM
Being millennial means remembering when Microsoft’s Marketing pitch was
“We are unstoppable and ruthless and will use computers to suck the joy out of any human activity that can yield a profit. Oppose us and die.”
Still is they’ve always been like that
“We are unstoppable and ruthless and will use computers to suck the joy out of any human activity that can yield a profit. Oppose us and die.”
Still is they’ve always been like that
Being millennial means remembering when Google’s marketing pitch was
“Hey this is better than the other stuff”
Not
“We have stolen thought and wish to replace you and starve you. Do you dream about replacing your smelly meatbag friends with our AI too? It can talk to them FOR you“
“Hey this is better than the other stuff”
Not
“We have stolen thought and wish to replace you and starve you. Do you dream about replacing your smelly meatbag friends with our AI too? It can talk to them FOR you“
January 19, 2026 at 6:41 AM
Being millennial means remembering when Google’s marketing pitch was
“Hey this is better than the other stuff”
Not
“We have stolen thought and wish to replace you and starve you. Do you dream about replacing your smelly meatbag friends with our AI too? It can talk to them FOR you“
“Hey this is better than the other stuff”
Not
“We have stolen thought and wish to replace you and starve you. Do you dream about replacing your smelly meatbag friends with our AI too? It can talk to them FOR you“
Being millennial demeans remembering when Google’s marketing pitch was
“Hey this new stuff is cool”
Not
“Resistance is futile”
“Hey this new stuff is cool”
Not
“Resistance is futile”
January 19, 2026 at 6:38 AM
Being millennial demeans remembering when Google’s marketing pitch was
“Hey this new stuff is cool”
Not
“Resistance is futile”
“Hey this new stuff is cool”
Not
“Resistance is futile”
Probably. Stats/sql are pretty cursed in terms of the library and tool support imho so gAI is simultaneously more tempting and likely better (due to vast numbers of tutorials and like “liberated“ into training sets.)
Anecdotally they’re pretty good at eg. sql, partially because sql is repetitiou.
Anecdotally they’re pretty good at eg. sql, partially because sql is repetitiou.
January 18, 2026 at 8:54 PM
Probably. Stats/sql are pretty cursed in terms of the library and tool support imho so gAI is simultaneously more tempting and likely better (due to vast numbers of tutorials and like “liberated“ into training sets.)
Anecdotally they’re pretty good at eg. sql, partially because sql is repetitiou.
Anecdotally they’re pretty good at eg. sql, partially because sql is repetitiou.
Hah I wish… hopefully someone else here does though.
Tbh I find myself wondering if there’s a lot of independent research? The frontier labs certainly tried to suck in as many ethicists as possible, one can’t help but feel to tamp down criticism, maybe they succeeded…
Tbh I find myself wondering if there’s a lot of independent research? The frontier labs certainly tried to suck in as many ethicists as possible, one can’t help but feel to tamp down criticism, maybe they succeeded…
January 18, 2026 at 8:51 PM
Hah I wish… hopefully someone else here does though.
Tbh I find myself wondering if there’s a lot of independent research? The frontier labs certainly tried to suck in as many ethicists as possible, one can’t help but feel to tamp down criticism, maybe they succeeded…
Tbh I find myself wondering if there’s a lot of independent research? The frontier labs certainly tried to suck in as many ethicists as possible, one can’t help but feel to tamp down criticism, maybe they succeeded…
Put another way: the code is used to do some work (probably statistical) to supply justifying data to some argument in the paper. So the argument is only valid if the code works as expected. This is always tricky! But particularly for gAI because of weak correctness assurance.
January 18, 2026 at 8:44 PM
Put another way: the code is used to do some work (probably statistical) to supply justifying data to some argument in the paper. So the argument is only valid if the code works as expected. This is always tricky! But particularly for gAI because of weak correctness assurance.
Ethical externalities around IP theft, energy remain.
Narrowly intellectually: explicability and trustworthiness of assertions about code risk compromise. (reading code historically not a strong guarantee of bug-freeness, so can assertions about work done by code in paper be as reliable?)
Narrowly intellectually: explicability and trustworthiness of assertions about code risk compromise. (reading code historically not a strong guarantee of bug-freeness, so can assertions about work done by code in paper be as reliable?)
January 18, 2026 at 8:41 PM
Ethical externalities around IP theft, energy remain.
Narrowly intellectually: explicability and trustworthiness of assertions about code risk compromise. (reading code historically not a strong guarantee of bug-freeness, so can assertions about work done by code in paper be as reliable?)
Narrowly intellectually: explicability and trustworthiness of assertions about code risk compromise. (reading code historically not a strong guarantee of bug-freeness, so can assertions about work done by code in paper be as reliable?)
I mean to be less flippant, no I don’t, but I also think that the ethical dimension of this has been vandalised by a highly effective political campaign.
I for my part have become radically critical.
If you feel like you can’t compete without feeding this beast, don’t know what to advise.
I for my part have become radically critical.
If you feel like you can’t compete without feeding this beast, don’t know what to advise.
January 18, 2026 at 8:37 PM
I mean to be less flippant, no I don’t, but I also think that the ethical dimension of this has been vandalised by a highly effective political campaign.
I for my part have become radically critical.
If you feel like you can’t compete without feeding this beast, don’t know what to advise.
I for my part have become radically critical.
If you feel like you can’t compete without feeding this beast, don’t know what to advise.
“Don’t”?
Note in passing that any serious use of AI in an academic work would likely involve passing parts of 3rd party work in as part of the prompt/context; and depending on vendor, that data might be reused for their own purposes. Less than clear anyone has a legal or moral right to do this.
Note in passing that any serious use of AI in an academic work would likely involve passing parts of 3rd party work in as part of the prompt/context; and depending on vendor, that data might be reused for their own purposes. Less than clear anyone has a legal or moral right to do this.
January 18, 2026 at 8:33 PM
“Don’t”?
Note in passing that any serious use of AI in an academic work would likely involve passing parts of 3rd party work in as part of the prompt/context; and depending on vendor, that data might be reused for their own purposes. Less than clear anyone has a legal or moral right to do this.
Note in passing that any serious use of AI in an academic work would likely involve passing parts of 3rd party work in as part of the prompt/context; and depending on vendor, that data might be reused for their own purposes. Less than clear anyone has a legal or moral right to do this.
“And this was scarcely odd at all: there were no more Tories to jump”?
(With apologies to Carrol)
(With apologies to Carrol)
January 18, 2026 at 8:28 PM
“And this was scarcely odd at all: there were no more Tories to jump”?
(With apologies to Carrol)
(With apologies to Carrol)
Which sortof points to the other big intellectual problem … it’s all very well writing off the US, but if you make yourself as dependent on eg FR and DE and then (say) Le Pen and AfD win, you’re back where you started.
There’s no magic bullet. But less dependence on a single hegemon might be good.
There’s no magic bullet. But less dependence on a single hegemon might be good.
January 18, 2026 at 8:22 PM
Which sortof points to the other big intellectual problem … it’s all very well writing off the US, but if you make yourself as dependent on eg FR and DE and then (say) Le Pen and AfD win, you’re back where you started.
There’s no magic bullet. But less dependence on a single hegemon might be good.
There’s no magic bullet. But less dependence on a single hegemon might be good.
Mm. Though I think what is likely more important than the length of the next Dem admin is what it arranges to lose to. (Really the mark of success of any true democrat is who they lose to: to really govern well is to shift the window so that your opposition is less repugnant at end than start.)
January 18, 2026 at 8:15 PM
Mm. Though I think what is likely more important than the length of the next Dem admin is what it arranges to lose to. (Really the mark of success of any true democrat is who they lose to: to really govern well is to shift the window so that your opposition is less repugnant at end than start.)
Personally I’m stopping at anger. Smile sweetly and refine plutonium.
January 18, 2026 at 7:44 PM
Personally I’m stopping at anger. Smile sweetly and refine plutonium.
So while I really liked your column (sorry should’ve led with that, even tho my opinion matters little!) it’s hard to see there being much choice in the long run, at least afaics…
January 18, 2026 at 7:22 PM
So while I really liked your column (sorry should’ve led with that, even tho my opinion matters little!) it’s hard to see there being much choice in the long run, at least afaics…
It feels hard to see the Republicans recovering rationality at the moment: and if the current or next president being a hostile fascist is remains a live possibility, it’s hard to have a trusting alliance, even if eg. someone sane wins in ‘28 and the house is Democratic.
January 18, 2026 at 7:21 PM
It feels hard to see the Republicans recovering rationality at the moment: and if the current or next president being a hostile fascist is remains a live possibility, it’s hard to have a trusting alliance, even if eg. someone sane wins in ‘28 and the house is Democratic.
(So much of AI activity at present feels like broad parody of some of the worst modern trends. Paving the cowpaths I guess.)
January 18, 2026 at 7:17 PM
(So much of AI activity at present feels like broad parody of some of the worst modern trends. Paving the cowpaths I guess.)
🤣 To be fair this also roughly parodies what was the promotion culture at big tech for a long time: I wrote this big promo packet about my messenger app, it’s so long the app must be genius etc etc.
January 18, 2026 at 7:16 PM
🤣 To be fair this also roughly parodies what was the promotion culture at big tech for a long time: I wrote this big promo packet about my messenger app, it’s so long the app must be genius etc etc.
Kinda sad — it feels like they can’t even enjoy a good messiah complex except by proxy: here is my monster, here’s what it tells me I should enjoy saying etc etc.
Too bad we’re all gonna get fired for this. (But then: who doesn’t?)
Time to learn some origami.
Too bad we’re all gonna get fired for this. (But then: who doesn’t?)
Time to learn some origami.
January 18, 2026 at 7:09 PM
Kinda sad — it feels like they can’t even enjoy a good messiah complex except by proxy: here is my monster, here’s what it tells me I should enjoy saying etc etc.
Too bad we’re all gonna get fired for this. (But then: who doesn’t?)
Time to learn some origami.
Too bad we’re all gonna get fired for this. (But then: who doesn’t?)
Time to learn some origami.
Sounds like great performance art to me. The consumer becomes the consumer etc etc. Summa cum laude.
January 18, 2026 at 6:58 PM
Sounds like great performance art to me. The consumer becomes the consumer etc etc. Summa cum laude.