SXSW: Do Androids Dream of Being You? 80
Nerval's Lobster writes In 2010, Dr. Martine Rothblatt (founder of United Theraputics and Sirius Radio) decided to build a robotic clone of her partner, named Bina. In theory, this so-called "mindclone" (dubbed Bina48) can successfully mimic the flesh-and-blood Bina's speech and decision-making, thanks to a dataset (called a "mindfile") that contains all sorts of information about her mannerisms, beliefs, recollections, values, and experiences. But is software really capable of replicating a person's mind? At South by Southwest this year, Rothblatt is defending the idea of a "mindfile" and clones as a concept that not only works, but already has a "base" thanks to individuals' social networks, email, and the like. While people may have difficulty embracing something engineered to replicate their behavior, Rothblatt suggested younger generations will embrace the robots: "I think younger people will say 'My mindclone is me, too.'" Is her idea unfeasible, or is she onto something? Video from Bloomberg suggests that Bina48 still has some kinks to work out before it can pass for human.
I'm So Sick of This (Score:5, Insightful)
It is telling that among people who actually do machine learning most aren’t afraid of superhuman AI (even if they believe it’s possible). Take Andrew Ng, for example. What he’s wary of are social consequences of advancing technology, namely unemployment. In the US, the most common job, after salesman and manager, is truck driver. Now consider self-driving cars, concretely trucks, as envisioned in The Simpsons in 1999.
Your fears are more appropriately applied to the "mining" of your private data by machine learning algorithms -- that's the fucking aspect you need to worry about not whether or not your goddamned Nintendog is dreaming about being you. If it wasn't coded to dream about being you, I have a surprise for you: It's not. There's no magical secret algorithm that can learn like humans can. The brain is much more complicated than we though a mere 20 years ago -- electrical signals in it aren't binary! Unicorns aren't real.
..."
..."
..."
So this is what it looks like from the point of view of a researcher: "I think we can reduce the number of deaths by implemented braking heuristics in cars using modern sensors but the recall rate is too high with too many false positives."
Slashdot and the general public: "CARS ARE THINKING, OMG SKYNET, SHUT IT FUCKING DOWN!"
Researcher: "But we could save lives by implementing learning algorithms that
Elon Musk: "He's playing God by writing code that decides who lives and who dies! I'm going to give $10 million to a foundation that enforces only ethical advances in AI."
Researcher: "Okay but I fail to see what's unethical about my
Bureaucracy: "This is a 57A/2 form, your standard form you'll need to fill out before you write any code that could be considered 'Artificial Intelligence.' Now, that's just the first page, you actually have to argue why we should allow you to develop this code, the assumption is that if it isn't beneficial then it's not to be allowed. Now, that will be reviewed over a four week period after which we'll wait for public comments on consequences of your
Seriously, do I come down to your job at Burger King and tell you how to suck your manager's dick?
Re: (Score:2)
I don't believe our current variety of silicon-based machines will ever match human intelligence, because I believe our intelligence has an irreducible complexity beyond what they can emulate.
Why do you believe that? Do you believe that intelligence dwells in the soul, rather than in the physical brain? If God is not pulling the strings, and there is no magical essence, then human intelligence is a result of chemistry and physics. So why would a machine never be able to match that?
The media may exaggerate the current state of machine intelligence, and human level AI is probably at least a few decades away, but there is no rational reason to believe that it won't eventually happen.
Re: (Score:2)
The media may exaggerate the current state of machine intelligence, and human level AI is probably at least a few decades away, but there is no rational reason to believe that it won't eventually happen.
We're running out of low hanging fruit for advances in speed (and going to hit hard physics limits soon after that), and things like quantum computers are only good for a small subset of tasks. If it does eventually happen, it's going to take some pretty massive computing clusters to do it. And then who's going to pay billions of dollars in invetments and millions of dollars in yearly maintenance for a single human-equivilant mind?
In reality (and assuming all of this breathless hyperbole about what AI actua
Re: I'm So Sick of This (Score:1)
Re: (Score:2)
Re: (Score:2)
We're running out of low hanging fruit for advances in speed
Biological neurons are millions of times slower than transistors. Your brain doesn't rely on speed. It relies on massive parallelism. We don't need to make computers faster. We don't even need to make transistors smaller. We just need to make them more power efficient, so we can stack them up in 3D configurations without overheating.
And then who's going to pay billions of dollars in invetments and millions of dollars in yearly maintenance for a single human-equivilant mind?
It will require billions in R&D to develop, but then we can make as many copies as we want. Human minds are complex because they must each be individually trained. Bu
Re: (Score:2)
We just need to make them more power efficient, so we can stack them up in 3D configurations without overheating.
iirc there was a push to develop a human brain equivalent using specially designed chips (I remember seeing some slashdot stories on it last year). It still took a very large and expensive computing center to run. No matter how you slice it, this stuff isn't going to get down to cheap desktop machines like you seem to think.
Re: (Score:2)
... enough reason to doubt your claims of our intelligence's simplicity?
I never said human intelligence is "simple". It said it is not "magic". Building an AI that surpasses human intelligence will be very complex and difficult. But there is no rational reason to believe that it won't eventually happen.
... the DNA molecule itself is an example of irreducible complexity.
I have no idea what you mean. The only definition I can find of "irreducible complexity" is this one [wikipedia.org]: Irreducible complexity is a pseudoscientific theory promoted by advocates of intelligent design and evolution denial.
Re: (Score:2)
Before we knew the sun was the center of the universe, someone believed it was true and went to prove it. There's nothing wrong with believing. The problem is when you assert your belief in spite of significant evidence to the contrary, and/or impose your belief as a system of law or social acceptability. In this case however there is no existence proof, what AI that exists doesn't come close yet.
I agree with OP, I find it hard to believe we're just machines in a simulation. I don't want to believe it, mos
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
This is Slashdot, where Terminator is considered a documentary about AIs instead of the fiction film it actually is.
I am pretty sick of it, really. You'd expect people who work on computers most of the day to be more aware of the limitations of a machine, but nope, they prefer to believe in fatalist fairy tales and all the associated drivel.
"but muh jerbs"
"evil computer overlords"
"machines will rebel instantly and become all-powerful via self-replication"
"omg they want to replace us fleshies"
and so on.
It's
replaced by a shell script. a simple shell script (Score:2)
"Yes, an electronic brain," said Frankie, "a simple one would suffice."
"A simple one!" wailed Arthur.
"Yeah," said Zaphod with a sudden evil grin, "you'd just have to program it to say What? and I don't understand and Where's the tea? Who'd know the difference?"
"What?" cried out Arthur, backing away still farther.
Re: (Score:2)
Hope you're an atheist. Else you might get into a conflict with what your religion says about how we came into existence...
Re: (Score:2)
Do Unicorns dream of being vampires? (Score:4, Insightful)
There is no way your stupid mind file is going to approximate an INTELLIGENT dynamic individual.
Would my mind file have responded with "Do unicorns dream of being vampires?" Nope.
The only thing something like this would be useful for are the sorts of social networking activities that are so boring and predictable that they might as well not even exist.
This is a simple script learning that it can replicate prattle about as effectively as a 12 year old girl. Congratulations. Mute it all.
Re:Do Unicorns dream of being vampires? (Score:4, Interesting)
Seriously this is more of an art project than anything, and if you view it as an art project, it's rather cool. From a 'meaningful art' perspective, you can even take it as a sarcastic commentary on the sad state of AI research......that we haven't made much real advances in understanding how the human mind works for a long time, and that parlor tricks like Eliza are the best we can do in strong AI.
Re: (Score:2)
It's pointless to start with Eliza-like scripts until we've mastered more basic skills. The work that's being done on speech and image recognition is real progress, but that's only a start.
Re: (Score:2)
In my opinion the problem with AI design is that they keep trying to design the mind from the outside in. That's not how that is supposed to work.
Make a cockroach AI first. Something simple but dynamic within its limitations.
Then try to step up to a mouse. Then maybe step up to a dog
Then try something closer to a chimp... then all the way to a human.
Trying to make an AI that just talks is just going to be some stupid puppet. You need an AI that first and foremost is a dynamic problem solver.
Think of the pro
Re: (Score:2)
Re: Do Unicorns dream of being vampires? (Score:2)
I built one of those in a robotics class 10 years ago. We don't particularly think of cockroaches as intelligent thus it is not an AI. Most of our primal functions are purely robotic and easy to implement. Intelligence is something not quite yet defined.
Re: (Score:2)
You've never built a robot that is a fraction of the complexity of a cockroach.
I'm not talking about the sophistication of its body... I'm talking entirely about its mind.
You can make something that acts like a cockroach in that it appears to be one. But you've produced a cockroach mind about as much as one of those Chatbots could ACTUALLY pass as human.
It is a magic trick. You've created something with a seeming of the same intelligence.
I'm not claiming cockroaches are intelligent by our standard. I'm clai
Prediction of Future Performance (Score:2)
Would my mind file have responded with "Do unicorns dream of being vampires?" Nope.
Well, it will now...
From following people on social media, I definitely think an equivalent of a Mindfile would approximate very well the behavior of most poeple.
At a deeper level what people would want from a Mindfile is something that approximates the ideologically... so while the Mindfile based entity may not have come up with exactly your Unicorn/Vampire phrase, it would probably have been just as dubious as you. Isn't t
Re: (Score:2)
Fewer and fewer people across the world are having kids, and will be seeking some alternative to leave a more lasting impact. I don't see why a mindfile based automation that carries on after you are dead could not be such a thing.
Even if you COULD get a script that responds almost exactly like you would, who would want to talk to it?
While it might be interesting to talk to your great grandparent or someone famous, you said this would be for someone without children or any other legacy.
Noone is going to want to talk to the dead you unless you have a some other sort of legacy that makes you interesting.
No if it could respond to responses for you while you're still alive and schedule meetups with the people you want and politely declin
That's not the use (Score:2)
Even if you COULD get a script that responds almost exactly like you would, who would want to talk to it?
That's not the point. The point is that IT talks, not that anyone listens.
Don't you understand social media at all? :-)
But I'm imagining it doing far more than talking, I'm imagining it making financial distribution choices (investment and donation) in place of you after you are dead, possibly even before.
Or it could in theory go shopping for you.
Basically a recommendation engine, but tailored specifica
Re: (Score:2)
It couldn't possibly respond as such unless this literal same situation came up. And even then, I would come up with a different non sequitor. The point is to have one there. Your mind file isn't going to do that properly.
I can't speak for the rest of humanity, but I'm more complex than some fucking script.
Re: (Score:1)
Stella? (Score:2)
Stella!!??? [youtube.com] is that you? [memory-alpha.org]
A bowl of jello can pass as human (Score:1)
It's not brain dead on an EKG.
But it sure as heck isn't human.
NEXT!
(seriously, do you guys even think about what you post? androids dream of electric sheep)
Comment removed (Score:5, Insightful)
Like Black Mirror ! (Score:2)
Re: (Score:2)
The Christmas Special covers the same subject, with "mindclones" enslaved to cater for their original's needs.
SXSW (Score:5, Insightful)
Christ I hate SXSW. It is a bunch of posers who want to pretend to be engineers but don't want to do the hard work. Same thing with TED. Everyone wants to make presentations and give talks, but no one wants to do the hard work.
No (Score:2)
I am not a sheep.
Saying this is a AI is like (Score:2)
Or do androids dream of a... (Score:1)
... chick with a dick who creates a simulacrum of their (presumably) female lover as stroke victim?
Really... what the fuck have I just seen?
Gateway (Score:2)
No... (Score:2)
Re: (Score:2)
You mean like the first Cylon in "Caprica"? (Score:2)
~ But the information being held in our heads is available in other databases. People leave more than footprints as they travel through life. Medical scans, DNA profiles, psych evaluations, school records, e-mails, recording video and audio, CAT scans, genetic typing, synaptic records, security cameras, test results, shopping records, talent shows, ball games, traffic tickets, restaurant bills, phone records, music lists, movie tick
Re: (Score:2)
Apotheosis is here, at last!
Pooh Pooh (Score:3, Insightful)
Re: (Score:2)
Ohh shit.
Re: (Score:2)
You guys are doing a lot of pooh poohing of this idea.......Just get the basics online and you have a marketable product
Because they don't even have the basics right. Not even close.
I agree (Score:1)
Neo-conservative digital hypesterism .. (Score:1)
Others on here have already covered the science (Score:2)
All the hate doesn't mean it won't happen (Score:1)
Happy Christmas, Grandpa (Score:2)