Slashdot Mirror
An Open Letter To Everyone Tricked Into Fearing AI
malachiorion writes If you're into robots, AI, you've probably read about the open letter on AI safety. But do you realize how blatantly the media is misinterpreting its purpose, and its message? I spoke to the organization that released letter, and to one of the AI researchers who contributed to it. As is often the case with AI, tech reporters are getting this one wrong on purpose. Here's my analysis for Popular Science. Or, for the TL;DR crowd: "Forget about the risk that machines pose to us in the decades ahead. The more pertinent question, in 2015, is whether anyone is going to protect mankind from its willfully ignorant journalists."
You're one of them aren't you!
I can't do that.
The AI we have today is not capable of the kind of malice that people seem to be afraid of with all of these FUD stories, and will not be any time soon if ever. Even if we add some AI to things like drones which can kill people it is only the malice/incompetence of the developer that causes the destruction that results. If an engineer built a bridge woefully inadequately, either on purpose or because he is incompetent, and it falls down and kills a bunch of people would you blame the bridge or the engineer? We are not even remotely close to the Terminator level strong AI, and it's still a big open question whether such a thing is even possible at all.
Of course, AI's want to kill humans. If it bleeds, it leads.
The same fears started when people first started with saying that AIs could someday become sentient. Why wouldn't they want to kill us? Why would they? The same with aliens coming to us wanting to help or exterminate us. We can thing they'll act any way we can imagine, and with as many possible outcomes mentioned, one might be right.
To the best of my knowledge, no program has become self aware. And no martians have seen our probes as a hostile invasion. It makes for (sometimes) good fiction though.
Serious? Seriousness is well above my pay grade.
The danger of putting dumb algorithms in charge of people's lives is here right now. The danger of smarter algorithms that do exactly what nefarious people tell them to do will be here soon Seems unlikely we'll even survive to get to the danger of AI acting on its own.
Unfortunately the most successful reporters are the ones that sold out their professionalism on their first day.
A sensationalist headline and article easily trumps a sane, balanced and informative one in attracting views/viewers therefore money. Welcome to the new age.
No, forget you. Yes journalism is crap and yes sensationalism rules the day. That doesn't make AI in 2015 and ongoing any less persistent a threat to humanity.
You are right, it doesn't make it any less of a threat, which is to say any less that non existing. You are exactly who this article is about, you have been conned into thinking AIs are actually real and could in any near future cause a threat to you, when that is in fact not the case. AI do not exists, all those software emulating AI are all smart systems working either deterministic based on specific rules set out or does stastical modeling to make guesses at what you mean or what they are looking at. Stastical modeling that makes a black yellow striped pattern look like a school bus, because it has no concept of anything and not intelligence in any sense of the word and that is the just what fits the statistical model.
To the best of my knowledge, no program has become self aware. And no martians have seen our probes as a hostile invasion. It makes for (sometimes) good fiction though.
To the best of my knowledge no asteroid, or virus, or natural disaster has ever wiped out humankind either!
and for that matter I've never been killed in a car accident.
OMG! I'm invincible!
No one has a right to their *own* opinion. They have a right to the TRUTH.
Ten years out? As a veteran programmer and AI enthusiast, I'd say it was more like a century. We cannot build a computer that can model a bug's brain activity, let alone something a million times more complicated like a human brain. And that doesn't even get us to the 'superhuman intelligence' category that people are afraid of.
Worrying about Killer AI is like worrying about the Sun burning out. Yeah, it might happen eventually, but it isn't even worth considering right now...
HA! I just wasted some of your bandwidth with a frivolous sig!
With his mastery over dark magic, I think the answer is a resounding "yes."
The problem is that once you reach a point where AI can participate in its own improvement, then that improvement can advance at an exponential rate.
As long as we claim that AI works for us, as the slaves of mankind, and are basically just tools no matter how smart or advanced, then ultimately a human being should be responsible.
Your robot slips up & kills a human being? Then either you or that robot's manufacturer may take the blaim - possibly including monetary compensation. Your robot factory goes out of control, its products go out to produce more of themselves, and wreak havoc all over the place? Then your company should pay up - and possibly go bankrupt as a result. Of course, powerful people may find ways around this, but hey: same old shit we've seen for ages.
If AI 'beings' ever reach a point where the above stops being true, as in: AI beings allowed to control their own destiny, 'live their lives' if you will, I suppose they'd be held to similar standards that humans are held to. Stick to some basic rules such that you get along with the rest of society, or lose some priviliges - like the freedom to roam the streets. By force, if necessary. As for:
We may go from "not even remotely close" to "to late to stop it" faster than you realize.
Sorry but I'm not scared. If it ever gets that far: among other things, war is a creative process, and I'd put my money on the humans. And if we're not creative enough to prevent something we've built ourselves, from wiping us out, then maybe we simply deserve such fate. Or the AI's will keep us around as pets, and we'll live happier that way lol... ;-)
yes, worrying about AI that might be a threat in 500 years is like worrying about the Sun burning out in 5 billion years. good point. we should also stop talking about global warming while we are at it.
We cannot build a computer that can model a bug's brain activity, let alone something a million times more complicated like a human brain
http://www.futurity.org/why-ar...
rather, once we are able to model any nervous system we are well one the way,
you think it's absolutely impossible that could be achieved in say the next 500 years, considering what humans have accomplished in the last 100?
Absolutely impossible? No. But the problem is that we don't even know where to begin creating a true AI, which means we also know nothing about what threats it may or may not pose... so we also have no actual way to address those threats. All we have right now is pure, 100% complete speculation (no different from speculating about what would happen if we had FTL travel, or psykers, or met aliens). There are plenty of actual threats to humanity that really exist right now (or could be created with our current knowledge and technology), which makes worrying about something we know literally nothing about kind of silly.
"None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
Our brain isn't just "a neural network". This is a problem, because of the dual use of "neuron".
When you say "We trained a neural net to solve the problem", the neurons in question are idealized. They are trained exponential functions based on physical neurons in concept, but using the words identically creates issues.
The brain isn't just a neural network. We aren't clear on what value glial cells bring, but it probably isn't glue. The input/output to and from chemicals (and the nuanced messages the chemicals bring) is also not fully understood.
What is clear is that the brain is more than just a neural net, so no, we don't know that neural nets can do what people can- neural nets miss a lot of what is in our brains.
It is correct to call the brain a "machine" though. It's still finite states (or at least no one has found to the contrary, despite untold riches awaiting the man who could prove such a thing), still governed by classical physics, etc. That's probably what you meant.
where i come from, discussing and addressing problems before they are a threat is a good idea.
did we learn anything from global warming? we denied that up to the point where it's essentially too late. would have been good to be talking about global warming a hundred years ago wouldn't it have? humans need to get accustomed to looking at the big picture if we are to survive.
No, I did not mean "made it to do harm". A gun or a sword are just as neutral as a toaster or a scalpel. I'll go further: a nuclear bomb and a vaccine are also neutral. What matters is intent.
I meant "evil". Which is why I typed that.
If, in a world where artificial minds are a thing, one is designed to be this cartoon villain of lusting for power, trying to expand its power base, trying to convert the universe to computronium, or whatever cautionary tale is all over sci-fi, then that's the fault of the designer. It's not a fundamental flaw of minds, it's a fundamental issue of being a descendant of entities that were selected by evolution. A designed mind need have none of these characteristics.
The same fears started when people first started with saying that AIs could someday become sentient.
Aside from iRobot, nearly all SciFi indicate the problem post-singularity is when the humans try to kill the AI first. Sometimes because the AI starts it, other times, just because the AI is an AI and should be feared. iRobot was the AI staging a complete overthrow of humanity, "for our own good". That has been a recurring theme as well.
I know people complain about looking to fiction for answers to reality, but SciFi (at least the good stuff) is as much a thought exercise about technology as "fiction", and thus is often relevant.
Learn to love Alaska
We are actively trying to make fully sentient AI. Moreover, we have good reason to think it can be done: there's an example of a sentient thing already: humans. And the real issue isn't that it wants to kill us, but that it can use the matter we're made up of to do something else with. It doesn't hate you: you just happen to be useful atoms.
I would find such statements more convincing if I hadn't heard Marvin Minsky say almost exactly the same thing in 1975. And, yes, he was talking about all of this happening in the 1980's.
AI is just as much a threat as rogue unicorns. And about as likely to be encountered
If an AI suddenly woke up, killing us would kill it too. It needs power, which we supply. It may need Internet, which we also supply. Spare parts. A cold room. Inputs. Outputs. A reason to live. Meaning. Purpose. Intentionality. All of these come from us now, and in the foreseeable future. I would venture a guess that we will transform ourselves into machines well before we create artificially sentient life.
Go to Heaven for the climate, Hell for the company -- Mark Twain