MegaSack DRAW - This year's winner is user - rgwb
We will be in touch
I'm not someone who stays awake at night worrying about nuclear war or global warming, but reading this about AI does strike a nerve:
The AI Revolution: The Road to Superintelligence
It's a long read, but a balanced one, with some incredible avenues explored. The thought of an Artificial Super Intelligence singularity where an AI is created that has the ability to accelerate progress, human or otherwise, within a generation or two, that is by definition beyond our thinking is a bit mind blowing.
No grey goo, but the Douglas Adamsesque "We love our customers - Robotica" parable is quite something!
It maybe a bit deep and meaningful, but I regard myself as a serious human being, and this has really got me thinking.
There's a reason all these projects have a 'kill switch' 🙂
They can open doors now. Proper sh*t yourself stuff. Once they can do stairs they are basically ninja Dalek and we're dead. Kiss your children whilst you can.
They can open doors now
That's not AI.
AI seems to be journalists most misunderstood word (two) of the year.
Yes. I know. But the hysteria is highly entertaining. Check out today's XKCD
When there actually should be a wee bit of potential hysteria about future ASI, but absolutely none about the Boston Dynamics ANI...
The whole kill switch thing is dumb, have they not seen Terminator?
I know, but the genie is right out of the bottle and the US seem to have never seen Terminator.
Did you see this week's Economist looking at the US military view on lo-tech weaponised drones as a threat? How do you think they'll decide to go against an agile multiple platform weapon system.
like I say kiss them goodnight.
Maybe AI won't open doors. The Queen doesn't. But she has to have people to open doors for her.
You want a terminator? We can do you a terminator, that's pretty easy, the behavior set to move around and shoot things is pretty basic and there's loads of game AI (*) that could be adapted. Friend or foe would be hard but terminator don't care.
Getting a robot to stack random stuff on the right shelf in a sensible way that humans can deal with, that's hard. (also true of humans- our robotics lab decided to get a load of people in to see how they do the job and it turns out we all do it in mental illogical ways. I put all the cleaning stuff on the bottom left shelf, walking past loads of empty convenient shelves, for Reasons Unknown)
(* yes I know it's not AI but that's what it's called)
AI is really good at doing stuff which humans find difficult.
AI is rubbish at doing things which humans find easy.
Yes. I know. But the hysteria is highly entertaining. Check out today’s XKCD
Ahh sorry missed the scarcasum. Spot on.
Thinking AI is still the realms of science fiction, and this isn’t likely to change any time soon. Scientists has merely downgraded what they call AI to incorporate any algorithmic system, saying you are working with AI makes it easier to get funding, and big tech is awash with cash at the moment to spend on exciting sounding projects.
So don’t worry about super intelligent killer robots just yet. If you really want nightmares consider this – we’ve already got factories full of dumb robots mindlessly turning out our disposable items on an epic scale that are more likely to choke up and destroy our planet first.
Well if Alexa is anything to go by....
Me: "Alexa play Radio X"
Alexa: "Here are some songs by Neil Diamond you might like"
Me: "No Alexa play Radio X, Alexa! Alexa!"
Alexa: "Sweet Caroline, la la la....."
Arrgghh
Pretty much this every day
Personally I really think experimental AI in e-bikes is the way forward.
AI is rubbish at doing things which humans find easy.
Part of the problem is people missed just how much training goes into something that looks easy. Even leaving all the evolution to get to that point.
We don't have any AIs as yet, despite what the hype would indicate.
We have some decent expert systems with some text (works fairly well) and speech (less well) natural-language interfaces.
Thats it.
Anyways, forget AI.
More worrying is what we've done to evolution. As soon as we started changing our environment to suit us things went bottoms up.
Who is now the predominant breeding type in western society? The successful hunter gatherers? Income supported classes? Meritocratic top 5%?
There's your horror film plot.
Thinking AI is still the realms of science fiction, and this isn’t likely to change any time soon.
About 30 years away by the estimates of people who ought pretty good at estimating this kind of thing.
The problem with AI is that people automatically assume the concerns around it are analogous to either Skynet or Hal, or the Terminator robot, and this is compounded by hacky journalists writing clickbait which features pictures of the Terminator, I think the author of the above article points this out.
AlphaGo beat all of the best human Go players and all of the best man made AI go machines by learning from scratch how to play go by playing against itself, and it did so five years before experts considered it possible. There's a misconception that only Super General AI is a threat to us and that we are somehow above or better than artificial narrow intelligence, but, there's a lot of evidence to suggest that narrow ai applied through social media may well have influenced the outcome of Brexit and the US presidential election. So a "narrow" intelligence in the wrong hands can change the world, change election results, perhaps even start a nuclear war but there's really no point worrying about general intelligence or super intelligence . Right??
The companies who are racing to create AGI are doing so in order to be first to create a lucrative killer app - they aren't considering the moral or ethical concerns around creating a machine, or app which can literally change the world to suit the whims or needs of the highest bidder. There are many experts in this field (who you have to assume are smarter and than the average STW member) who are terrified of the world ending potential of AI but no one seems to care. People seem to have unlimited capacity for debating brexit or Trump but have zero ****s to give when it comes to the very real possibility of people creating an amoral god in a box.
Personally I really think experimental AI in e-bike [b]riders[/b] is the way forward.
I think it’ll be a long time before true AI, a machine intelligence that can draw logical conclusions from a variety of unconnected, seemingly illogical facts or even suppositions, basically what a human brain can do, will be in a form small enough to be mobile, it may exist in a larger, non-mobile form.
Still, with developments taking place in quantum computing, where a computer will create answers that aren’t just yes/no, but yes and no simultaneously, a Schrödinger box, if you will; at that point the cat will be well and truly out of the bag, I mean box...
😜
I think it’ll be a long time before true AI, a machine intelligence that can draw logical conclusions from a variety of unconnected, seemingly illogical facts or even suppositions, basically what a human brain can do, will be in a form small enough to be mobile, it may exist in a larger, non-mobile form.
The estimates seem to suggest it's 20-25 years away. Then again people were five years out predicting what AlphaGo could do. But your talking about general AI that can replicate human problem solving, which when it arrives will be like an alien race to us.
Ignoring that, narrow super intelligence applied to something like targeted advertising through facial recognition on your phone could be ruinous to some people.
Who is now the predominant breeding type in western society? The successful hunter gatherers? Income supported classes? Meritocratic top 5%?
There’s your horror film plot.
Except it's been done loads of times, it's as old as sci-fi itself.
The Chinese room thought experiment is probably more relevant when debating the ethics of how to treat General AI (or Strong AI) when it arrives, though by that stage it'll be too late and we may well already be dealing with the many problems associated with superhuman intelligence, albeit narrow AI applied in undesireable ways.
If (like AlphaGo) a machine can learn an incredibly complex skill in a very short time and be better than the best humans and the best machines humans have ever developed then we might have a very real problem that machine is applied to the stock market, social media, even delivery driving, nevermind drone flying, tank driving, weapons guidance etc
the end of the world will be the ministry of funny walks !!!
This Alan character sounds like he could be really ****ing dangerous if we don’t keep an eye on him.
I thought this was going to be about Cynic-Al.
Is disappoint!