what's the dumbest shit in AI annoying everyone today, I need to write one up

[glyph@mastodon.social]

@davidgerard anyhow: sorry I didn't answer the question, I guess I will watch your video now

[davidgerard@circumstances.run]

@glyph if this worked on these people they wouldn't be into AI

[madargon@is-a.cat]

@peter_mcmahan @davidgerard this definitely should be on top... I'm sick of hearing techbros talking about feelings of machines and software, at the same time mercilessly exploiting their workers, living and feeling human beings.

[peter_mcmahan@mas.to]

@madargon @davidgerard
I've been seeing an upsurge in this kind of discourse recently. Generally more discussion around "please dont be mean to the token generators" like this from Troy Hunt a couple of weeks ago https://www.troyhunt.com/weekly-update-501/

[davidgerard@circumstances.run]

@peter_mcmahan @madargon i hope you saw that and resolved to be meaner to linear algebra. it had it coming

[peter_mcmahan@mas.to]

@davidgerard @madargon
I bully matrix algebra every chance I get

[europlus@social.europlus.zone]

@davidgerard “hallucinations” being seen as anything but normal genAI output.

“Guardrails” (including stupid prompts) being seen as the solution to the above.

Everything above being accepted anyway and risk-based problems, as well as all the ethical/moral, societal and environmental ones, just ignored for “left-behindism).

That therefore LLMs fail on a simple measure against all of the UN’s Sustainable Development Goals (SDGs).

Dog, I need a drink now.

[europlus@social.europlus.zone]

@benjaoming @davidgerard OMD! It’s Douglas Adams’ Electric Monk…which killed someone because of what was effectively an hallucination!

[pikesley@mastodon.me.uk]

@peter_mcmahan @madargon @davidgerard

sorry did he just use the word Robophobia?

[davidgerard@circumstances.run]

@pikesley @peter_mcmahan @madargon so many people wanting to fuck linear algebra with a face drawn on

[fayedrake@furry.engineer]

@peter_mcmahan @davidgerard that’s… they’ve just described exactly the opposite of what they’re actively trying to do to humans

If you consider “jailbreaking” to be equivalent to the continuous psychological warfare going on

The hypocrisy is so thick you could cut it with a knife.

[europlus@social.europlus.zone]

@jwz @davidgerard staring agree.

[rycochet@furs.social]

@davidgerard @pikesley @peter_mcmahan @madargon They hate that actual human beings have the ability to tell them they're wrong and say no to their propositions, while with AI, the answer is always yes and in their eyes, true personhood is only ever granted to people who agree with them.

They're so narcissistic that outside perspectives are never sought, nevermind considered, so it only stands to reason the thing they really want to fuck is a sycophantic virtual mirror, as free of morals as they are, only in it's case it's a quirk of design rather than an intentional choice.

[celeduc@mastodon.social]

@madargon @peter_mcmahan @davidgerard Limitation of liability is the driver here. The operators seek to deny their human reviewers grounds for negligence by putting "don't be mean" policies in front of the users.

[barubary@infosec.exchange]

@peter_mcmahan @madargon @davidgerard Oh hey, isn't that Troy "Elon Musk is right" Hunt?

[bredroll@mas.to]

@davidgerard That people named "Al" are innocently caught up in the whole thing

[somebody@tech.lgbt]

@peter_mcmahan @davidgerard this website feels 100% serious in the same way alex jones is, which is to say whoever wrote this is so high on bullshit there's functionally no difference between lying and being completely earnest

[glasspusher@beige.party]

@peter_mcmahan @davidgerard @madargon

The little bit of linear algebra/matrix math I used for my telescope controller software is getting smeared by association

[davidgerard@circumstances.run]

@punissuer @robbo @njoseph Ia! ia! Fhtagn!

Ki! Ki! Ki!