Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Tack]

I mean, there’s that game where factories are made of paperclips but uh.. _{probably not ^{really feasible}}

[Egan_BW]

turn the whole universe into paperclips using only paperclips for construction

[Loud Whispers]

Just something really cool, that had been bothering me for a while: Prehistoric dentistry.
I always wondered at how humans coped with the sheer rate of attrition caused by dental diseases, as poor dentistry is surprisingly debilitating and lethal to humans today. Looking into it though, the tales of our ancestors' attempts to sort out their teeth is fucking awesome. For starters, millions of years ago, before homo sapiens, there's evidence to suggest that our hominid ancestors were trying to figure our ways to scrape off infected teeth tissue. And while dental issues were debilitating and lethal back then, they were surprisingly rare - it's only until homo sapiens begins developing agricultural societies about 10,000 years ago that suddenly everyone's developing poor dental issues, with yet more surprising finds as women seem to have contracted more than men, and all of this cannot be entirely explained by a regular diet of carbohydrate rich foods. Even cooler, over 9,000 years ago the Indus river civilization set humanity on the path to sick dentistry by inventing the first dental drill. The rest is pre-history

[redwallzyl]

Just something really cool, that had been bothering me for a while: Prehistoric dentistry.
I always wondered at how humans coped with the sheer rate of attrition caused by dental diseases, as poor dentistry is surprisingly debilitating and lethal to humans today. Looking into it though, the tales of our ancestors' attempts to sort out their teeth is fucking awesome. For starters, millions of years ago, before homo sapiens, there's evidence to suggest that our hominid ancestors were trying to figure our ways to scrape off infected teeth tissue. And while dental issues were debilitating and lethal back then, they were surprisingly rare - it's only until homo sapiens begins developing agricultural societies about 10,000 years ago that suddenly everyone's developing poor dental issues, with yet more surprising finds as women seem to have contracted more than men, and all of this cannot be entirely explained by a regular diet of carbohydrate rich foods. Even cooler, over 9,000 years ago the Indus river civilization set humanity on the path to sick dentistry by inventing the first dental drill. The rest is pre-history

You should see Neanderthal teeth. You want to see dental horror just look at them.

Spoiler (click to show/hide)

Exposed dental pulp was the norm.

[Loud Whispers]

That's pretty Dorfy dentistry

[Bumber]

Unless the metal is paperclips.

No metal needed, just the odd human...

Humans that could otherwise be turned into paper clips? I think not.

[Maximum Spin]

Humans that could otherwise be turned into paper clips? I think not.

I mean, you can always turn them into paperclips later.

That's like saying "corporations want to make the most money, so they can't harm anyone to make more money because they'd have to spend money to do it".

[wierd]

So, you are saying the AI would produce a product lifecycle management policy for human-based FTL devices, such that humans are strapped in (using only non-metallic materials), connected directly to the AI using fiber optic cables (Fun fact, neural tissue can respond to light stimulus!) so that the AI can abuse them until they die horribly from neglect, then their bodies can be processed in a nuclear fusion reactor (made of ceramics, and materials too dense to serve as paperclips) to be converted into iron, which will then be used to make paperclips?

That's a lot of process there. Doubtful that the simple maximizer would come up with that using a bland old mutation based algo.

[Maximum Spin]

There's no reason why it can't just use metal, you realise. As long as the expected gain in paperclips outweighs the short-term cost of paperclips you aren't making now, the AI will still choose to invest the metal in paperclip-seeking projects. And yes, as long as it has enough memory, the paperclip maximiser is mathematically guaranteed eventually to come up with that plan, and every other finitely describable plan. An evolutionary algorithm eventually exhausts every finite subset of the possibility space.

BTW, wireheading won't save you either because a rational agent will still devote some resources to finding ways to increase its ability to wirehead even faster by obtaining more processing cycles to tick up the paperclip counter with, so that just devolves to the basic "destroy the universe to make itself smarter" case.

[wierd]

That assumes that the AI has the solution space defined that broadly, and that the AI does not encounter a local optima.

[Maximum Spin]

That assumes that the AI has the solution space defined that broadly, and that the AI does not encounter a local optima.

Evolutionary algorithms don't get stuck in local optima unless you did it wrong - once you hit a local optimum, you put just as much effort into searching the probability space as always, and eventually you find an even better optimum and flip to that. And once you have exhausted your current definition of the solution space, you expand your definition. All things AI are currently perfectly capable of doing.

[Reelya]

I think this specific point in the article (which is about gaming addiction) has the cause and effect backwards:
http://www.news.com.au/technology/home-entertainment/gaming/gaming-disorder-to-be-an-official-mental-health-condition/news-story/cdfeff5ae01fda47603ae6f22ad152e2

The idea that video games can have impairments on other significant parts of a person’s life, such as work, was explored earlier this year when an American Time Use Survey linked working less hours to video games.

Between 2004 and 2007, men between 21 and 30 years old played two hours of video games per week, but that has now risen to 3.4 hours per week according to the report.

Men aged between 21 and 30 years old saw their working hours decline by 12 per cent annually from 2000 to 2015, compared with an 8 per cent decline for older men.

Wow, this is some shonky research. Correlation isn't causation for a start, and you can't say which one is the cause or effect here even if there is a link. Also the numbers don't add up when you work them out:

a 10% decrease per year in men's hours means men are losing hours off their working week every year. Meanwhile, a 90 minute increase in gaming over a decade means that each year they add 9 minutes of extra gaming per week. Hell, if you're skipping work so you can cram more gaming in, you'd expect to get more than an additional 9 minutes of gaming for each hour of worked skipped: you can fit 12 hours of gaming into the time need to get ready, commute, work 8 hours, then come home. If you only managed to fit 8*9 = 72 minutes of gaming into that day off, then gaming wasn't the reason you didn't go to work.

 Clearly, gaming cannot actually account for why men are losing hours so quickly. Extra gaming is therefore a symptom and the vast majority of the lost working time is spent on things unrelated to gaming, which have been conveniently excluded from the discussion.

So it's 100% a beat-up issue. Men have more time off, they do more of everything you do when you have time off, including a little gaming. Gaming is a symptom not the cause. But it's the typical media thing where a difficult issue like the collapse of people's livelihoods is written-off with a victim blaming narrative where they're just all lazy gamers now.

[Reelya]

I just wanted to mention, on a useful tech note, that firefox's Pocket bookmarking system is absolutely excellent. I stuffed a whole bunch of links in there from different cloud services that I had (feedly, trello, youtube watch later videos) which simplified things immensely. Now if i want to find something again at some unspecified future point i just right click on the link and send it to pocket, where you can later add tags, read watch and delete stuff.

But the really cool feature of Pocket turns out to be that it's not just a decent cloud bookmarking system. It reads the actual page you linked and boils them down to just the actual content and not all the site crap (navigation, recommended stuff, comments section etc), converting everything to a nice uniform legible font, too. Basically it's good if you want to keep stuff but also if you want to look at stuff without downloading a pile of junk that's not the actual article or youtube video, it's a great way to get just the article itself in a nice printable / tablet readable format.

[MrRoboto75]

Maybe those men were playing sonic 06, so the other 51 minutes or so went to loading screens.

[syvarris]

I always thought that the paperclip thing was relevant because it's describing an easy-to-understand programming bug that could wipe out the human race, not because it's actually likely.  It just illustrates that a small error could result in everyone dying, which is pretty worrying even if it only has a tiny chance of happening.  Worse, while the specific example has plenty of flaws in it, that doesn't mean there aren't other possible AI bugs which could lead to human extinction.

Programmers tend to screw up in all sorts of unexpected ways, it just isn't scary because in the worst case scenario they brick a computer.  The idea that a perfectly mundane mistake could brick the planet, however, is very scary.