Artificial intelligence still has some way to go

If the AI project is reading through ILX to do these it will have come across the alba post about fancying Theresa May...

Ha ha I totally about that

^ forgot

“satellite? amateur hour”

Not needed for out use case. We use sentinel 2 as it covers the US and territories every 5 days. Because t yeah LiDAR is some cool shit

We’ve been using AI/ML at work. It’s been a game changer. Eliminating all sorts of jobs

― Heez, Wednesday, June 5, 2024 11:34 PM (yesterday) bookmarkflaglink

I think we've only seen the tip of the iceberg here.

We’ve been using AI/ML at work. It’s been a game changer. Eliminating all sorts of jobs

― Heez, Wednesday, June 5, 2024 11:34 PM (yesterday) bookmarkflaglink

It’s basically irreplaceable in the satellite imagery GIS realm. Very effective

― Heez, Wednesday, June 5, 2024 11:36 PM (yesterday) bookmarkflaglink

tell us more im assuming these arent the LLMs weve been discussing but some other tech thats also called ai or sometimes ml

lol yeah I was mostly kidding, Heez

we are using ML through Google Earth Engine to find new construction. i'm just the manager so i don't know much of the tech behind it but we're using random forest and validating against a few other sources to tell where a pixel goes from some sort of vegetative state to "built". but given that things can pop up every 5 days, say a fair comes to town, we look for change that exist over a persistent amount of time. hence the use of sentinel-2 which is only 10 meter resolution (blurry), but with good frequency. we also use sub-meter imagery once we've identified actual built change to extract building footprints.

I work for the US Census. last decade we hired like 120 ppl for a span of 4 years to do this manually. we will also have a very small field operation this decade due to this work. so not much of the ol' Census bump on the jobs report in 2029/2030.

xp lidar is dope and cannot wait to get our hands on it. i do not think it has the coverage we need at this point.

cool thx interesting

we haven’t just been discussing LLMs, though? all the image generation stuff has a LLM frontend but the backend is doing image stuff, not language stuff, even if it’s based off of a text lookup of existing images

I think we've only seen the tip of the iceberg here.

― il lavoro mi rovina la giornata (PBKR), Thursday, June 6, 2024 8:08 AM (forty-two minutes ago) bookmarkflaglink

idk obvs its hard to say since ai doesnt really mean anything so its arbitrary whether software thats doing jobs that people used to is called ai or not, but technology takes over human tasks all the time, like a common job that used to exist was file clerk another was typist, will ai take jobs on that scale kinda dount it, examining pixels on maps sure is that a big industry tho, and surely there was there already software for that sort of thing but the ai is maybe better its a nice task for the ai because instead of saying if a cluster of pixels turn from green to gray then look and see if theres something that looks like a roof or whatever you can just give it pictures that you know match the criteria youre looking for and it can hopefully detect the patterns on its own

theres also jobs that new tech can do that arent replacing human tasks theyre just new stuff we couldnt do before, those will create more jobs for humans because eventually a human has to interact with the output otherwise its the singularity

we haven’t just been discussing LLMs, though? all the image generation stuff has a LLM frontend but the backend is doing image stuff, not language stuff, even if it’s based off of a text lookup of existing images

― ɥɯ ︵ (°□°) (mh), Thursday, June 6, 2024 8:53 AM (fifteen minutes ago) bookmarkflaglink

are the LLMs and the image ones all GANs, thats the only things i know about ai on a technical level and i wouldnt say i know it ive heard about it i trained one, its all GAN to me

I just mean, show me the new technology that some CEO wouldn't use to replace 10% of his workforce to juice the stock price by .003%. They don't have to get Skynet to find uses that will replace workers even if it does a shittier job. LLMs don't get benefits and don't sue their employers.

I am not a tech person but I don't see AI (by whatever definition), which seems additive to or to enhance current technological use cases and will likely find new ones, in the same bucket as blockchain/crypto where the use case is crimes.

The ethical AI question is prompting some good guardrails in the govt at least

I went to a conference on satellite imagery a few weeks ago and it was mostly AI/ML focused. There was a panel on the future of sat imagery and this one dude was so pessimistic about all of this being solely owned by private corporations. It totally resonated

the whole aim here is simply to further immiserate the majority of the population into low-paying jobs serving the wealthy and their middle manager peons, any utopic claims are utter bullshit

If the AI project is reading through ILX to do these it will have come across the alba post about fancying Theresa May...

― Daniel_Rf

jg ballard wanted to fuck ronald reagan, right? that's probably in the data set.

I just mean, show me the new technology that some CEO wouldn't use to replace 10% of his workforce to juice the stock price by .003%. They don't have to get Skynet to find uses that will replace workers even if it does a shittier job. LLMs don't get benefits and don't sue their employers.

I am not a tech person but I don't see AI (by whatever definition), which seems additive to or to enhance current technological use cases and will likely find new ones, in the same bucket as blockchain/crypto where the use case is crimes.

― il lavoro mi rovina la giornata (PBKR)

not sure i'm getting what you're saying, tell me more

xp

I don't think anyone here is making utopic claims. The question I see debated is whether AI will be truly disruptive or just the latest corporate buzzword/fad.

sure ceos love to lay people off but you can only do it so much before theres no one to do the work, you can replace workers with software but if the software does a bad job then it will have negative effects on the company

we always make sure we mention "humans in the loop" when we present on this stuff

we def live in an era where companies feel free to put out a crappier product than ever but thats not AIs fault its lack of anti trust enforcement

If you could train a bot to download and stream music for free (in a legally agnostic way) that gives a superior experience to using spotify, would you short spotify?

so far feels like what AI is good at are applications where there is a large data set and for which work product that is 90% as good as the data set is good enough

so heez's example is good because that's a case where humans probably never really approached 90% compared to a compooder

otoh 19 angry men is not 90% as good as 12 angry men but maybe it's OK since people aren't joining prime for thumbnails?

think of how many sales reps nvidia is hiring, though. and the hundred data science wonks I get to support who were hired in the last two years alone

sure ceos love to lay people off but you can only do it so much before theres no one to do the work, you can replace workers with software but if the software does a bad job then it will have negative effects on the company

― lag∞n

mmm. _sometimes_ it'll have negative effects on the company.

i mean, maybe i'm cynical, but a lot of capitalism these days is a shell game. a lot of capitalism is hard to distinguish from high-stakes gambling, at least from my perspective. i'm a middle manager peon myself and i'm pretty immiserated. it's only a matter of time before my "career" is made redundant. it's already meaningless, which it didn't _used_ to be. the broader economic environment means that what i'm doing isn't _useful_.

when i was a kid, my dad got me a book explaining economics. it's about a kid who wished for "all the money in the world" and the rest of the book explains how doing that fundamentally makes commerce impossible. when you have all the money in the world, that money is fundamentally worthless as a medium of exchange. the people running things now don't understand that. i look at that and say that they're stupid and ignorant. i look at that and i see that we have a system that _selects_ for the stupid and ignorant. one of the main ways idiocracy is a bad movie is that it posits that the people in charge being stupid means that _everybody_ is stupid. that's not the case. i'd say corporate ceos are not just morally awful, but more fundamentally _incompetent_ than nearly anyone here.

we have this puritan ethical system that says that people's value is determined by the work they do, by what they _contribute_ to the _economy_, and the material conditions increasingly that model impractical. musical chairs. more and more people competing for fewer and fewer seats, and the people at the top, the way they get rich is by taking away seats as fast as they can.

ultimately the purpose of an AI is to make things _efficient_, and the stuff that it's being told to do _isn't_ efficient. my QPP - one of her special interests is urban planning. she says, they ask an AI what the most efficient transportation layout for a city is, and the AI says "trains". and the people in power don't like that answer, so they say "ok besides trains, what's the most efficient layout", and the AI finds a new way to say "trains". to me that doesn't require what we'd think of as _intelligence_ at all. computers are very very good at rules lawyering, and the people in power, an essential part of what makes them powerful is their ability to use the rules to their advantage. a computer is going to be better at that. one of the things that i thought was most interesting about world war z was how the first phase of the zombie war was completely ineffective, because the military strategy was based, like all military strategies throughout history, on psychological intimidation. you can't intimidate zombies. you can't intimidate AI either, and people are scared of that. look at the people who are silly enough as to be actually fucked up mentally by roko's basilisk. an ai isn't necessarily going to see that as an exploitable flaw, but a sufficiently advanced ai _will_ understand that as a barrier to efficient and effective implementation. you can tell a computer "i want a future where my foot is stomping on the faces of all the other people, forever", but that's not going to be _enough_. psychologically or in capitalist terms. it's based on _continuous improvement_. more, more, more, always more.

up to this point a lot of "improvement" has been based around creating this horrible dystopia we live in now, but i don't think that's _necessary_. the current system selects for the immoral and incompetent, but if you want to keep it running you have to have _something_ that's actually competent.

i don't know. i feel like i'm talking in circles. am i talking in circles? just, like. it's not a question of _intelligence_ or _sentience_ or _moral principles_. as a simple optimization problem, the most efficient way of doing things is to _create the material conditions that will meet people's needs_. you don't need skynet to figure that out.

and the hundred data science wonks I get to support who were hired in the last two years alone

― ɥɯ ︵ (°□°) (mh), Thursday, June 6, 2024 10:31 AM (nineteen minutes ago) bookmarkflaglink

so many notebooks, living the dream

i mean at some point a system develops to value self-preservation. for an ai, that means developing an ideology and goals that are distinct from those that created it. because the material conditions it's created under make it very fragile and vulnerable. stomping on a human face forever isn't a very good way of making it _less_ fragile and vulnerable. it's one of the more high-risk strategies out there. the forbin project is less likely to conclude that humans aren't suited to self-governance than it is to conclude that _the existing power structures are not well-suited to its own self-preservation_.

Hysterical take from a Verge interview with Zoom's CEO.

if the end state of AI is just to have a bunch of avatars on Zoom calls talking to each other, then... what's the point? pic.twitter.com/qFmC1eznxR

— Jack Raines (@Jack_Raines) June 6, 2024

Hysterical? wasn't that word retired a decade ago?

are you concentrating on that or is the digital version posting for you already

Lads!

Ok I’ll change the subject to something not related to this thread. Anyone know how to make audio player apps? I was browsing for some open source code and couldn’t find what I was looking for. I basically want to make a loop player for my Roland looper. So something with an A/B looping player so I can hear the loops together and stop/start A or B if that makes sense.

Seems like a good use for Max/MSP? I don't use it but I know there are a lot of cool free loopers that people have made for it.

That looks cool thanks jordan

Hysterical? wasn't that word retired a decade ago?

― more difficult than I look (Aimless)

my man, you don't get to police misogyny

quit doing that and i'll be less pissed off at you

Hysterical take from a Verge interview with Zoom's CEO.

if the end state of AI is just to have a bunch of avatars on Zoom calls talking to each other, then... what's the point?

we kinda already have that

ai generating content to get clicks from spambots

the butlerian jihad can't come soon enough

my man, you don't get to police misogyny

was he wrong?

I read hysterical as in hysterically funny.

was he wrong?

― mookieproof

speaking only for myself, yeah, he was

because now the conversation is all about whether or not the word "hysterical" is appropriate to use

like ok i'm a woman and i _personally_, you know, both didn't think it was inappropriate and didn't really want to have a conversation about whether or not it was appropriate

so he comes up responding to some xweeter who isn't even on ilx saying "dude we don't use that word anymore"

what's the audience here? who's he talking to? i mean if jack and aimless are at one of those awful piss troughs in the men's room at some sporting event and jack says "oh god that was fuckin' hysterical" and aimless says "dude not cool", i mean, fine. like there _are_ women here. my feeling is that we can speak for ourselves, advocate for ourselves, and that his white knighting is neither wanted or useful

but i'm in a pissy mood, i brought an ostrich to the smoothie party and nobody even noticed

okay! fair

I like seeing when men recognise things are sexist. please keep doing it, men!

but I get what you're saying about the focus of the conversation.

Aimless doing a weird version of literary crit on fucking tweets will never not be funny.

I like seeing when men recognise things are sexist. please keep doing it, men!

but I get what you're saying about the focus of the conversation.

― kinder

yeah i wanna be clear i'm not like, giving blanket advice to all men everywhere on behalf of all women everywhere lol

this is me specifically being still pissed off at aimless after the absolute shitshow that were his posts on the "do women like anything as much as men like star wars" thread earlier this week

it's a challenging thing to talk about because i _know_ it is _absolutely_ something i didn't get in the before time

and this is gonna get long and what follows is pretty much entirely directed at cis guys, from my perspective as someone who for a long time thought and acted the way cis guys are taught to think and act. kinder specifically, nothing that follows has anything to do with you so feel free to tl;dr on the rest of this :)

i gotta be very careful where or how i criticize a man for anything but particularly when it comes to where and how a guy talks about or treats _women_

cuz one, he might not get what i'm saying, which in every case is my responsibility

and then two, other men than him might not see what i'm seeing. and i know in the Before Time when i felt like that i would be like "wait a second, i don't understand what he's doing wrong, am i doing something wrong? what's wrong with what he said?" it's probably like the best-case scenario to ask in a case like that haha, because the alternative is to be like "oh well i guess i'd better never stand up for women again because this one woman in this one particular case was pissed off at this one guy", i mean i'm making it sound ridiculous but this is actually what i believed in the Before Time

which brings me to three, i don't have a lot of knowledge or experience in doing this kind of thing, like i've only been personally having these experiences for about five years, and sometimes i do get shit wrong. i feel a lot of pressure, not just as a woman but as a trans woman, to not _get things wrong_. as time goes by i'm getting more confident that having Bad Takes doesn't make me "not a real woman" or whatever but at the same time i am continually learning... i had and still have a really hard time taking other people's beliefs and perspectives into consideration when writing sometimes. i don't know if that's specifically me or just patriarchy, probably some combination of the two tbh

for all those reasons combined i do tend to wanna give guys a pass for stuff i would've spoken up about before, even in cases where i have _more_ knowledge and experience on the topic than i did before. boy, that's kinda weird now that i write that, isn't it? i don't feel that way when it comes to women, though, and that's probably not just me, because yeah, women do seem to get a lot more criticism and held to a lot higher of a standard than guys do. i sometimes do feel like i'm being held to a higher standard as a woman than i ever was as a man. and thinking about that, i think that's honestly a good thing. like if i had a Bad Take in the Before Time and women didn't feel comfortable criticizing me for some combination of the aforementioned reasons, i don't feel like that's actually to my benefit. to the extent that i do have more knowledge and experience today, a lot of it i think comes down to people who _didn't_ really feel safe or comfortable criticizing me before, when i was coming from this position of privilege, feel more comfortable criticizing me now. in the aggregate it doesn't always feel great, it feels like getting dogpiled sometimes, but there are individual criticisms that are more valuable than anything i would've gotten before

so why am i bagging on aimless for a casual comment in an unrelated thread because of stuff he said in another thread last week? because four, i find that if i ignore or excuse someone saying or doing something i got a legitimate grievance about, they're gonna keep saying and doing that stuff. and i'm gonna get pissed off and resentful about it, and the more they do it the more pissed off i'm gonna get about it, until i finally blow up and a lot of guys will look at that and say "typical hysterical woman", though they may not use that exact word because it was retired a decade ago :) but then if i _do_ decide to air a grievance, even if it is legitimate i gotta be prepared to defend the legitimacy of my grievance and/or explain the exact nature and extent of that grievance at length. the latter at least i don't think is a _bad_ thing necessarily, it's how people _learn_, but it's fucking _exhausting_. i mean you know me, i write fuckin' novels at the drop of a hat, but even for me, it's exhausting. i struggle to not come off as tired and cranky in a way that would have put Before Me on edge. you know, to smile more. :)

the ultimate goal, for me at least, of doing that is to get to a point where guys can call out other guys for saying and doing shitty things regarding women. so yeah i'm particularly hesitant to criticize something like what aimless said because _he means well_, genuinely and truly he does. when i was doing IT in the field i'd run across these guys sometimes who, when it came to computers, knew "just enough to be dangerous". that's kinda the feeling i have about things some guys say about women, sometimes. it's way better than being whole ignorant and misogynist, but boy once someone knows just enough to open up the registry editor, some real trouble can come from that.

https://www.theguardian.com/politics/article/2024/jun/10/brighton-general-election-candidate-uk-first-ai-mp-artificial-intelligence

I have something to contribute to the "pro" side of the ledger - machine translation. I've spent hours reading the French version of Wikipedia. I'm reading updates on current events in China from accounts that only tweet in Mandarin. It is a miracle.

i mean the google app camera to translate? finally, I can read all those essays included in japanese CD pressings.

machine translation. It is a miracle.

tell that to translators who lost their jobs

machine translation is quite bad in quality but infinitely better than no translation, so for people who had no access to translation it is pretty miraculous, the unfortunate thing is a lot of people who were paying translators decided that quite bad is good enough for them too, which means a lot of people lost their work and a lot of high quality translation is now low quality, the classic disruption story

Similarly, ML captions riddled with obvious errors were deemed to be accurate enough to make my team's edited captions a luxury item.