like many other issues, I am becoming emotionally aligned with a more obviously worse proposition ("AI is good actually") because the criticisms by the right side are lazy and facile.

@sun

> by the right side

I'm pretty convinced that the secret to happiness is adopting the Asian philosophy that nothing is good or bad, it just is -- acceptance.

@eriner I am giving just a tiny bit of deference to them and their feeling that something isn't quite right about AI stuff, but personally have adopted a different attitude about it and a few other pieces of software where I admit I am spreading around a smaller percentage of displeasure to get a bigger percentage of productivity. on average I come out ahead using AI so I get over the "smell displeasure" of AI.

@sun It has its faults, but it's objectively useful for many use-cases even if it has a failure rate of ~10%+, as long as the user accepts that and accounts for it.

Most of the non-remediable issues are simply side-effects of the increased tool sophistication, particularly for those with little to no technical skills (i.e, spam, people using tools to create malicious content)

@eriner I watched a really long video about AI risk that postulates about an attacker inserting malicious code into your codebase via LLM service. It's a perfectly valid thing to think about but literally every comment is a dumbass leveraging it to argue that LLMs are useless. I would feel unsatisfied if every comment on my videos heaped me with praise but misunderstood what I said

@sun Yeah, the spam problem is not going to get better. I mean, look at some of the Twitter bots. Some of them are pretty sophisticated (and existed before LLMs) in generating enough spam to drown out reasonable discourse.

It's demoralizing, but only highlights the importance of socializing primarily AFK and making (and maintaining) connections with real people.

The golden era of early Internet is ending, if it wasn't over years ago.

@eriner over on the farcaster network there is super sophisticated llm spam (usually gets caugght but not always) and it exists despite the traditional idea that having to pay 6 bux for an accoubt will stop spam

@sun

NAS has ~1.4k active accounts.

At $6 per, it would take a mere $8,400 to have sophisticated bots in equal proportion to legitimate users.

$33,600, $55 less than a new model Hybrid Honda Accord, buys you control of 80% of the network.

In practice, it probably only takes a percent or two to poison the well.

AI has significantly lowered the bar for creating linguistically passable bots.

It will continue to be abused.

There is no technical solution.

Acceptance is the only solution.

@sun Yeah, a social "web of trust"-like system (which I guess FoaF is -- I'd never heard of it) is probably the only way forward.

Though, I could envision a future where it's allowed to become bad enough (Problem, Reaction, ...) that the "solution" people clamor for is government-backed centralized identity online.

@eriner the future is centralized heavily on overt government control and broken up by firewalls between hostile government alliance lines. very bleak but thats whats coming because the good future is too hard under the best conditions
@eriner yeah the open internet wasn't planned it was a road bump
@eriner we can't even replicate a free overlay network on whats coming

@sun lol, I guess you're more black-pilled about it than I am. What do you mean by this?

@eriner I am extremely black pilled. personal encryption is going to be outlawed and there will be a secure execution chip on every device that you'll have to authenticate yourself to that tags every packet you send with your identity.
@eriner it will happen in my lifetime and I am in my 40s
@sun @eriner it won't happen
no western government is competent enough to come up with such a scheme, nor has the political will to stand up to overwhelming opposition from every business that finds out how much it'll cost and how many years it'll take to upgrade their infrastructure.

would you like a shitty, half-assed national digital ID system that costs hundreds of billions more than planned, and takes 10 years longer than planned to build out so it's worthlessly outdated by the time it's ready to go instead? online banking will require a java plugin that's so fucking poorly written and unreliable that it's easier to just drive to the bank and wait in line for the teller.
@skylar @eriner a firmware update to the intel trusted execution engine will re-enable the trusted computing screen to keyboard control that was attempted in 2005 and shot down

plans for national id are already written they already have vendors picked out, they are waiting for the right emergency to roll it out
@sun @eriner and then almost immediately after such a firmware update is released, it will be rescinded in a panic after they realize it's breaking compatibility with many of the countless combinations of devices attached to PCs

this isn't a rapidly implement in the aftermath of an emergency scale project, it's a decades and trillions of dollars in infrastructure project. major breaking changes on the protocol level will largely require new hardware, and no vendor or group of vendors has the production capacity to crank out new hardware at a rate that matches demand from even a slow, sector by sector rollout affecting only the banking industry, then the healthcare industry, etc. software vendors won't have enough competent programmers to implement the changes in the accumulated decades worth of business software.
with the way literally every business and local government in the west spends the absolute bare minimum on IT infrastructure and holds onto everything old until it literally breaks, it ain't happening.
i am still finding hubs used in production, sunman. the fax machine will outlive us all.
@skylar @eriner did you know they already badly do it in many countries and people just put up with it
@skylar @eriner its not really overnight, a lot of it already exists or can be done through phone apps.

none of this can actually be done until there's a working single id system but it's being built block by block since about 2006
@eriner @skylar texas tried to resist REAL ID Act and the federal government went so far as to say they would shut down all international airports in Texas if they didn't comply, and texas rolled over immediately
@sun @skylar @eriner yeah good idea, call out the program that is billions over budget and half a decade behind schedule to beat Skylar’s allegations of programs billions over budget and half a decade behind schedule
@sapphire @skylar @eriner I said it's inevitable and would happen in my lifetime and that central id is the key component with most of the other components already being in place. you can squirm around all you want, several states have already implemented it completely
@sapphire @eriner @skylar several european countries straight up have national smart card ids, we are gonna have that here eventually. all that banking software you guys are claiming doesn't exist is already being used

@sun @skylar @eriner @sapphire fwiw, we have smard card enabled ids and they are used by almost no one. might be because germans are usually pretty pro-privacy.

another thing i noticed: you probably want to have migrants which form subcultures. i think the millions of turks we have in germany are the best insurance against the state going full throttle authoritarian.

@bonifartius @skylar @eriner @sapphire how much immigration we should have is a different discussion from should there be a giant human trafficking pipeline across the border to satisfy america's hunger for the cheapest possible labor
Follow

@sun absolutely. the mass immigration we had in germany in the last decade wasn't good as well.
@skylar @eriner @sapphire

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.