News for hackers. Stuff that matters.
Re:The Darwinian argument for worrying about AI by digbybk on Friday June 30, 2023 @11:22AM (#36536568)
About “pandering to authority”, I keep seeing this move in debates:

- people with safety concerns just don’t understand the tech. - “here’s a list of people with safety concerns who demonstrably do understand the tech very well” - you’re pandering to authority.

A related move:

- greater intelligence leads to greater concern for universal values. - “here’s a list of very intelligent people who disagree” - there are a lot of very intelligent people in cults too, you know.

I don’t think I’m a doomer but they do seem more coherent than the people saying there’s no cause for concern.

Re:The Darwinian argument for worrying about AI by ethanbond on Friday June 30, 2023 @11:35AM (#36536880)
Yeah the other fun move is: AI will be powerful and unbridled from human control enough to achieve great outcomes that humans cannot conceive of. But in the negative case: AI will not be operating outside of human control, it will not be handed immense resources to pursue goals (verifiably false given the capital and compute investment already backing AI).
Re:The Darwinian argument for worrying about AI by robertlagrant on Friday June 30, 2023 @11:50AM (#36537152)
It's not backing AI. It's backing humans to try and make an AI to do something.
Re:The Darwinian argument for worrying about AI by ethanbond on Friday June 30, 2023 @11:52AM (#36537199)
What bearing do you think this has on the question at hand?
Re:The Darwinian argument for worrying about AI by robertlagrant on Monday July 03, 2023 @04:25AM (#36570259)
It's not verifiably false that people want AI to be autonomous and it's not true that it has been handed immense resources. People have been handed some money to make better AIs to do things.
Re:The Darwinian argument for worrying about AI by pydry on Friday June 30, 2023 @12:05PM (#36537466)
I have safety concerns about stepladders but I'm not a stepladder doomerist.

Which of these experts are doomers?

Re:The Darwinian argument for worrying about AI by digbybk on Friday June 30, 2023 @12:19PM (#36537711)
One of the guys who invented back propagation, Geoffrey Hinton, for one. Of the three people who won the Turing Award for AI research, two are sounding the alarm. Hinton said he’s most concerned about the existential risk.
Re:The Darwinian argument for worrying about AI by logicchains on Friday June 30, 2023 @01:24PM (#36538799)
>Of the three people who won the Turing Award for AI research, two are sounding the alarm.

Two of them who won it for stealing ideas Schmidhuber came up with over a decade ago and not giving credit.

Re:The Darwinian argument for worrying about AI by digbybk on Friday June 30, 2023 @02:20PM (#36539732)
I'm not entirely sure what we're supposed to take from this. Are you saying that they are not actually AI experts, and we're returning to the claim that only people who do not understand the tech are concerned about these risks? That's a stretch.

Or is it a new move:

- people with safety concerns just don’t understand the tech.

- “here’s a list of people with safety concerns who demonstrably do understand the tech very well”

- Those particular experts are bad people so we shouldn't listen to them.

I'm really not committed to a side of this debate, and if I check my own bias, it's probably on the Steven Pinker "progress is good" side of things. I just keep hearing bad arguments from the AI-optimist side, and your comment is just another example.

Re:The Darwinian argument for worrying about AI by tinideiznaimnou on Friday June 30, 2023 @03:18PM (#36540685)
>I'm not entirely sure what we're supposed to take from this. Are you saying that they are not actually AI experts, and we're returning to the claim that only people who do not understand the tech are concerned about these risks? That's a stretch.

The point could be that they are legitimate AI experts, but narrow-minded: their thinking utilitarian, their reasoning motivated. Unlike the original inventor, whose thinking was abstract enough to originate a novel method, but at the same time not pragmatic enough to even get the credit that he is owed. It's not an entirely unfamiliar scenario in tech, is it?

An analogy for those people would be an engineer who designs machine guns, or fighter planes. They may be really great at figuring out how to make the gun shoot farther, or the plane fly faster. At the same time, they may not have a very good grip on the idea that, in the end, someone's gonna get killed by their invention.

Such an engineer will probably have a pretty decent rationalization that would enable them to keep getting paid for doing what they love: inventing these fun machines for throwing bits of lead across the air and/or flying around. Whoosh!

Re:The Darwinian argument for worrying about AI by nradov on Friday June 30, 2023 @08:00PM (#36544386)
Fighter plane engineers are fully aware of how their inventions will be used and generally agree with the mission.

"Because I make killing machines. And they were both damn fine killing machines."

https://www.youtube.com/live/_MUK241uZHM?feature=share

Re:The Darwinian argument for worrying about AI by tinideiznaimnou on Monday July 03, 2023 @07:02AM (#36571257)
Of course they're gonna interview the guy who's gonna say that - it makes for better viewing! Others might hold more conflicted outlooks, we just never hear about them on popular media. (A quote attributed to Mihail Kalashnikov does come to mind.)

But that's kinda beside the point. I'm more concerned that you missed the primary logical error that I committed in my analogy: the AI experts in question aren't doing the rationalizations but the scaremongering.

Re:The Darwinian argument for worrying about AI by AnimalMuppet on Friday June 30, 2023 @12:11PM (#36537577)
It's not necessarily a "move" - not if it comes from different people.

This is a problem in a public forum. You're not debating one-on-one. You're debating N-on-M. When you think you have a point nailed down - people with safety concerns understand the tech, say - then someone else comes in and argues against you from their viewpoint.

(If it's the same person, then yes, it's either a "move", or it's just that they're arguing just to argue, and not listening at all. If so, just move on - one cannot reason with a brick wall.)

Re:The Darwinian argument for worrying about AI by digbybk on Friday June 30, 2023 @12:22PM (#36537785)
I was actually thinking of a podcast interview I just listened to that Sam Harris did with Marc Andreessen, so just one person over the course of a long conversation.
Re:The Darwinian argument for worrying about AI by ChatGTP on Friday June 30, 2023 @06:41PM (#36543495)
I listened to that episode too and I saw him on Lex Fridman .

I’m my opinion if feels Marc is intentionally misleading people to protect his interests. He can’t be that silly.

I’ve already seen someone on here now parroting the “thermodynamic” argument Marc was using.

It’s as if he’s delivered a stupid arguments playbook to many trolls.