fuck milo....i want a 40 year old foul mouthed biker to drink with and say 'where the fuck have you been all day dogscock' when i open it, i also want a pet wookie.
Beefsandwich on
0
Options
ZephosClimbin in yo ski lifts, snatchin your people up.MichiganRegistered Userregular
edited June 2009
Its an interesting concept to say the least.... I'm curious what the people in the "behind closed doors" demos of milo will say about him, so, well see.
Oh, you wanna see what the bottom of the avatar's shoe looks like, huh? Well...BAM!
I know there seems to be a lot of optimism in this thread but if history has told us anything it's that this stuff is all going to be a joke. Just like the Power Glove and everything else like it.
If microsoft does this properly I think the Wii 2 will have to at least have something similar.
Also if microsoft wants to capture any of the Wii market with this device they are going to have to market the shit out of it. Commercials every five seconds.
Thirdly if Milo is anywhere near as intelligent as they are making him out to be I will be amazed.
Yea, Molyneux's team solved all those issues AI researchers have been trying to make even a bit of progress in for the past half-century.
Or, you know, it was scripted.
Whichever.
Yeah, I don't think the Milo thing is worth pinning much hope on. Frankly I don't really care. Actual AI is not something I'm yearning for. A severely limited approximation of AI is fine by me, as long as it's capable of dealing with the entire scope of whatever application I'm using it in.
But Project Natal itself looks promising. I'm definitely looking forward to what comes of it.
Yea, Molyneux's team solved all those issues AI researchers have been trying to make even a bit of progress in for the past half-century.
Or, you know, it was scripted.
Whichever.
Yeah, I don't think the Milo thing is worth pinning much hope on. Frankly I don't really care. Actual AI is not something I'm yearning for. A severely limited approximation of AI is fine by me, as long as it's capable of dealing with the entire scope of whatever application I'm using it in.
But Project Natal itself looks promising. I'm definitely looking forward to what comes of it.
People don't seem to realize that NLP requires a certain knowledge of the world to work though. It can't work in a vacuum.
The sentences "We gave the monkeys the bananas because they were hungry" and "We gave the monkeys the bananas because they were over-ripe" have the same surface grammatical structure. However, the pronoun they refers to monkeys in one sentence and bananas in the other, and it is impossible to tell which without a knowledge of the properties of monkeys and bananas.
Kipman also showed me a version of Burnout that had been set up to work with Project Natal. Burnout is a serious game, not just a tech demo — it's a polished, fast-paced racing game with high-end graphics, and I happen to have played a lot of it. With Project Natal, instead of using a joystick, you steer by holding your hands up in the air like you're gripping a steering wheel. To hit the gas, you move your foot forward along the floor. To brake, you move it back. To trigger the turbo boost, you do a gear-shifting, fist-pumping movement with your right arm. Awesome.
It takes a few minutes to get the hang of it. You tend to oversteer, since you can't quite believe this thing is going to pick up your movements, so you exaggerate them. But soon you start to trust it, because it does actually work. I couldn't detect any significant latency. And there's definitely an extra edge to playing a game with nothing between you and the screen but your clenched, white-knuckled fists. I'm a hard-core gamer, so I'm not the person Project Natal is targeting. I love my controller as it is. But the appeal of Project Natal is real. You could compare it to the difference between regular movies and 3-D movies: it puts you in the action in a way that nothing else could.
So I actually got to see this demo-ed a few times over the past few months (one of the privileges of working at Xbox LIVE). The Burnout demo that we saw was the thing that impressed me the most. He did this awesome brake-turn-spin maneuver which would be difficult to do on a controller, but easy to do with a wheel or, in this case, a fake wheel. They were just trying to show that this tech can be added to existing games. Whether or not developers go back and do this, though, is another question.
Not sure why they didn't show that demo in the press conference, when they're obviously showing it to reporters. I guess they just didn't have enough time or something.
Yea, Molyneux's team solved all those issues AI researchers have been trying to make even a bit of progress in for the past half-century.
Or, you know, it was scripted.
Whichever.
Yeah, I don't think the Milo thing is worth pinning much hope on. Frankly I don't really care. Actual AI is not something I'm yearning for. A severely limited approximation of AI is fine by me, as long as it's capable of dealing with the entire scope of whatever application I'm using it in.
But Project Natal itself looks promising. I'm definitely looking forward to what comes of it.
People don't seem to realize that NLP requires a certain knowledge of the world to work though. It can't work in a vacuum.
The sentences "We gave the monkeys the bananas because they were hungry" and "We gave the monkeys the bananas because they were over-ripe" have the same surface grammatical structure. However, the pronoun they refers to monkeys in one sentence and bananas in the other, and it is impossible to tell which without a knowledge of the properties of monkeys and bananas.
Yeah, I understand the current limitations of AI. I should have been a little clearer. I'm assuming that any application intending to use natural language processing is going to have some pretty stringent, obvious limitations. I'm not talking about having deep and meaningful discussions with your Xbox. I'm talking about telling your Xbox to play a video, ordering your squad to move to point C, etc. Selecting an option from a finite list. Not so much AI as speech recognition, which is basically what the Natal trailer showed.
It's worth bearing in mind that even if we somehow moved hundreds of years ahead in AI research and developed a flawless NLP system, it would be useless unless we had a game to go along with it, capable of adapting to any possible command. So even if we somehow came up with this NLP system sometime in the foreseeable future, that limits it to pretty much an AI 'buddy', that can chat with you. I really don't think it's worth worrying about 'real' AI right now - it's just not feasible.
So there are definitely some elements to that presentation that aren't "fake". It'll be curious to see how Milo interacts with the other "VIP" reporters.
(And FYI, while I knew about Natal months ago, Milo was just as big a surprise to me as it was to everyone else today)
On Monday night, I got the briefest of demos of what may be the most ambitious piece of software known to be in development for Microsoft's new motion-control Project Natal: the virtual-person game known as Milo.
most ambitious piece of software
This coupled with Molyneux, plus the fact that it is a damn kid means this is fucking awful straight from the get go.
Seriously. What part of any of this is appealing to any one? The dude needs to have a lobotomy or a developer license system created so we can take his away.
On the TV, a boy, Milo, sat on a swing hanging from a tree. Molyneux asked me to stand up in front of the TV and to begin looking around. The game was using head-tracking technology, I discovered. Leaning my head to the left or right — or backing away or walking closer to the TV — smoothly adjusted the game's camera angle. The tracking was responsive and intuitive.
That's pretty cool.
Darlan on
0
Options
RaakamToo many years...CanadalandRegistered Userregular
It still looks neat - if you scan your own things into games designed for it, it'll be cool. You can customize the software to be more personal, which is neat. The whole game show thing with the families also looked neat honestly. The skype-type video conferencing was pretty cool too. It's obviously peppered in with some very obvious bullshit (the girl puts the dress on and it fits perfectly on her picture, Milo, flawless voice recognition) but it still looks neat.
Raakam on
My padherder they don't it be like it is but it do
Wow, just saw the videoes and I'm sorry, but I'll believe it when I see it in real action. Look far too good to be true to me.
I'm not really concerned with whether it's really that good or not, it's whether it'll catch on. I see it as similar to an uncanny valley effect: people are weirded out by doing some things. I think this is one of those things.
They have more than one hurdle to overcome but if they succeed then it is interesting technology.
Peter Molyneux: Milo can recognise the emotions on your face and the emotions in your voice. He can recognise certain words you say. You can have conversations with him, you can read stories to him. We're trying to bring all these things together. Some of them are tricks - I'll be absolutely honest with you - to make you believe Milo's real.
He can recognise what you're wearing. If he notices you've got dark bags under your eyes he will say, 'You look tired today.'
Eurogamer: Really?
Peter Molyneux: Absolutely, all of that works. We're combining all that together to make you really believe that he understands what you say.
Let's try an experiment. When a human voice says something funny, there's a different tone in the voice. Even though Milo's not trained to recognise your voice, if you say something funny to him, he should recognise it as something amusing. Try it now.
Eurogamer: Tell him a joke, you mean?
Peter Molyneux: Yeah.
Eurogamer: OK. Milo?
Milo: [Looks up, smiles and nods]
Eurogamer: Bloody hell. Er, OK. A Times New Roman walks into a bar. The barman says, 'We don't serve your type.'
Milo: [Giggles]
Eurogamer: Bloody hell.
Peter Molyneux: Now, he didn't really understand every word you said, but from the tone of your voice he guessed you were telling a joke.
Peter Molyneux: Milo can recognise the emotions on your face and the emotions in your voice. He can recognise certain words you say. You can have conversations with him, you can read stories to him. We're trying to bring all these things together. Some of them are tricks - I'll be absolutely honest with you - to make you believe Milo's real.
He can recognise what you're wearing. If he notices you've got dark bags under your eyes he will say, 'You look tired today.'
Eurogamer: Really?
Peter Molyneux: Absolutely, all of that works. We're combining all that together to make you really believe that he understands what you say.
Let's try an experiment. When a human voice says something funny, there's a different tone in the voice. Even though Milo's not trained to recognise your voice, if you say something funny to him, he should recognise it as something amusing. Try it now.
Eurogamer: Tell him a joke, you mean?
Peter Molyneux: Yeah.
Eurogamer: OK. Milo?
Milo: [Looks up, smiles and nods]
Eurogamer: Bloody hell. Er, OK. A Times New Roman walks into a bar. The barman says, 'We don't serve your type.'
Milo: [Giggles]
Eurogamer: Bloody hell.
Peter Molyneux: Now, he didn't really understand every word you said, but from the tone of your voice he guessed you were telling a joke.
So it's an illusion then, but a clever illusion?
Also that joke sucked. Bad Milo. Bad.
Of course it is. What did you expect?
As was stated earlier in the thread AI has a long way to go. So you have to rely on tricks like this in order to make up for defecites in real understanding on the AI's part.
I imagine it being similar to trying to communicate with someone who doesn't speak or understand your language. You look for clues, their facial expressions, their tone of voice (I'd like to see if Milo can pickup on deadpan humor.. let Bill Murray at it and see how it does, would be interesting), body language etc. I believe that's the kind of "tricks" Peter is talking about.
LittleBoots on
Tofu wrote: Here be Littleboots, destroyer of threads and master of drunkposting.
Even if the Milo thing does work exactly as they make out it does, I still don't really see what the point is. I play games to... y'know, have fun - not have faked conversations with a computer. If I fancy a chat with someone I'd talk to a real person, not a creepy CGI boy.
Ultimately it seems like an impressive tech-demo (assuming it wasnt all scripted at any rate) but kind of pointless... why would I want to spend money on it?
Cowbomb on
0
Options
RaakamToo many years...CanadalandRegistered Userregular
Even if the Milo thing does work exactly as they make out it does, I still don't really see what the point is. I play games to... y'know, have fun - not have faked conversations with a computer. If I fancy a chat with someone I'd talk to a real person, not a creepy CGI boy.
Ultimately it seems like an impressive tech-demo (assuming it wasnt all scripted at any rate) but kind of pointless... why would I want to spend money on it?
Substitute boy for woman with huge breasts that is naked and saying dirty things and you suddenly realize how much money they could make off this thing.
"O no! You caught virtual herpes! Once a month, your avatar will look like a street hooker with a giant sore!"
Sex sells, and while it's sad, any company that manages to bring something for all the lonely geeks of the world is going to be filthy rich.
Raakam on
My padherder they don't it be like it is but it do
I just wish these gimmicks weren't as hyped by outlets as they are in these, really, pre-existing conditions.
This really doesn't matter in the slightest until you get it to perform as a practical application. Talking to a boy that is half gimmicky, half unfinished, and all nothing doesn't really inspire me to believe. I can see where it would work, but it needs to work before I'm convinced it's anything more than an elaborate expectation script. Give me a practical example where this works BETTER than already existing systems, and then we can talk.
Molyneux had me wrap up the demo by approaching Milo again. He asked me to call Milo over by saying my own name and acting happy. Milo walked up to me and complimented me on my blue shirt (my shirt was indeed blue). Then he stood there, waiting for me to say things. There was a little awkwardness here. The Milo demo was partially being manipulated by a developer who was sitting nearby, and I couldn't tell if he was merely calibrating the game or how much he was pulling its strings.
Again, it's early on, but I'm still skeptical of all the tech.
Substitute boy for woman with huge breasts that is naked and saying dirty things and you suddenly realize how much money they could make off this thing.
"O no! You caught virtual herpes! Once a month, your avatar will look like a street hooker with a giant sore!"
Sex sells, and while it's sad, any company that manages to bring something for all the lonely geeks of the world is going to be filthy rich.
Google will fill in the rest for the curious.
I think this does have possibilities in areas such as, like stated previously, RPGs.
Let's say I want to buy a Sword of Awesome +2. I'd walk into a shop and say to the owner "Hey, have anything for sale?" The keyword "hey" would signify that he could return the greeting, and "sale" would put him into 'selling crap' mode. Tone of voice could also be used as a tool for haggling - if he's a jovial kind of person he'll respond well to a lighthearted "you know, another five gold off that price would sell it...", or perhaps he's more stubborn, requiring fast-paced and brash negotiation. Even the most basic of NPCs would get another dimension to their personality, to say nothing of informants or interactions with party members.
While obviously still years off solid implementation as much more than a gimmick, there's a valid line of reasoning for attempting the technology.
Edit: Not to say this hasn't been tried before. It just hasn't really worked all that well due to limitations in technology. With enough work (in years, maybe even decades), I could see this being a much more natural and quick way of interaction compared to radial menus and written dialogue options (just like head-mounted displays for vision).
Posts
a. something on a tv or computer
b. "censorable behavior" i.e. "Hey milo I fucked your mom and she never loved you"
if the kid makes a
The guy can't even get a dog's AI right.
Color me apprehensive.
I know there seems to be a lot of optimism in this thread but if history has told us anything it's that this stuff is all going to be a joke. Just like the Power Glove and everything else like it.
If microsoft does this properly I think the Wii 2 will have to at least have something similar.
Also if microsoft wants to capture any of the Wii market with this device they are going to have to market the shit out of it. Commercials every five seconds.
Thirdly if Milo is anywhere near as intelligent as they are making him out to be I will be amazed.
His games always promise more than they ever achieve, and he usually misses them big.
Yea, Molyneux's team solved all those issues AI researchers have been trying to make even a bit of progress in for the past half-century.
Or, you know, it was scripted.
Whichever.
Yeah, I don't think the Milo thing is worth pinning much hope on. Frankly I don't really care. Actual AI is not something I'm yearning for. A severely limited approximation of AI is fine by me, as long as it's capable of dealing with the entire scope of whatever application I'm using it in.
But Project Natal itself looks promising. I'm definitely looking forward to what comes of it.
I wonder how Milo will react to goatse.
Tofu wrote: Here be Littleboots, destroyer of threads and master of drunkposting.
People don't seem to realize that NLP requires a certain knowledge of the world to work though. It can't work in a vacuum.
Depending on accuracy, Natal plus Forza 3 could be bitchin'.
Har, someone asked this very question on the E3 megathread. Here's what I posted:
The TIME article is here: http://www.time.com/time/arts/article/0,8599,1902208,00.html
Here is the quote:
So I actually got to see this demo-ed a few times over the past few months (one of the privileges of working at Xbox LIVE). The Burnout demo that we saw was the thing that impressed me the most. He did this awesome brake-turn-spin maneuver which would be difficult to do on a controller, but easy to do with a wheel or, in this case, a fake wheel.
Not sure why they didn't show that demo in the press conference, when they're obviously showing it to reporters. I guess they just didn't have enough time or something.
- Don't add me, I'm at/near the friend limit
Steam: JC_Rooks
Twitter: http://twitter.com/JiunweiC
I work on this: http://www.xbox.com
Yeah, I understand the current limitations of AI. I should have been a little clearer. I'm assuming that any application intending to use natural language processing is going to have some pretty stringent, obvious limitations. I'm not talking about having deep and meaningful discussions with your Xbox. I'm talking about telling your Xbox to play a video, ordering your squad to move to point C, etc. Selecting an option from a finite list. Not so much AI as speech recognition, which is basically what the Natal trailer showed.
It's worth bearing in mind that even if we somehow moved hundreds of years ahead in AI research and developed a flawless NLP system, it would be useless unless we had a game to go along with it, capable of adapting to any possible command. So even if we somehow came up with this NLP system sometime in the foreseeable future, that limits it to pretty much an AI 'buddy', that can chat with you. I really don't think it's worth worrying about 'real' AI right now - it's just not feasible.
This is Peter Molyneux were talking about here. I harbor no ill will towards the guy. I loved B&W 1 AND 2, as well as both Fables.
But he's like... He's a child with his toys.
Coincidentally, Stephen Totilo (previously of MTV Multiplayer fame, now on Kotaku), just blogged about his brief demo with Milo: http://kotaku.com/5275204/testing-molyneuxs-milo-a-virtual-boy-with-yes-a-dog
So there are definitely some elements to that presentation that aren't "fake". It'll be curious to see how Milo interacts with the other "VIP" reporters.
(And FYI, while I knew about Natal months ago, Milo was just as big a surprise to me as it was to everyone else today)
- Don't add me, I'm at/near the friend limit
Steam: JC_Rooks
Twitter: http://twitter.com/JiunweiC
I work on this: http://www.xbox.com
This coupled with Molyneux, plus the fact that it is a damn kid means this is fucking awful straight from the get go.
Seriously. What part of any of this is appealing to any one? The dude needs to have a lobotomy or a developer license system created so we can take his away.
That's pretty cool.
The author guesses at a price point around $200.
It still looks neat - if you scan your own things into games designed for it, it'll be cool. You can customize the software to be more personal, which is neat. The whole game show thing with the families also looked neat honestly. The skype-type video conferencing was pretty cool too. It's obviously peppered in with some very obvious bullshit (the girl puts the dress on and it fits perfectly on her picture, Milo, flawless voice recognition) but it still looks neat.
they don't it be like it is but it do
They have more than one hurdle to overcome but if they succeed then it is interesting technology.
(I'm...half-kidding.)
Oh. Scanning that in would be epic.
"What do you think of the size Milo?"
"Ever been in a Turkish Prison, Milo?"
So it's an illusion then, but a clever illusion?
Also that joke sucked. Bad Milo. Bad.
Of course it is. What did you expect?
As was stated earlier in the thread AI has a long way to go. So you have to rely on tricks like this in order to make up for defecites in real understanding on the AI's part.
I imagine it being similar to trying to communicate with someone who doesn't speak or understand your language. You look for clues, their facial expressions, their tone of voice (I'd like to see if Milo can pickup on deadpan humor.. let Bill Murray at it and see how it does, would be interesting), body language etc. I believe that's the kind of "tricks" Peter is talking about.
Tofu wrote: Here be Littleboots, destroyer of threads and master of drunkposting.
Ultimately it seems like an impressive tech-demo (assuming it wasnt all scripted at any rate) but kind of pointless... why would I want to spend money on it?
Substitute boy for woman with huge breasts that is naked and saying dirty things and you suddenly realize how much money they could make off this thing.
"O no! You caught virtual herpes! Once a month, your avatar will look like a street hooker with a giant sore!"
Sex sells, and while it's sad, any company that manages to bring something for all the lonely geeks of the world is going to be filthy rich.
they don't it be like it is but it do
This really doesn't matter in the slightest until you get it to perform as a practical application. Talking to a boy that is half gimmicky, half unfinished, and all nothing doesn't really inspire me to believe. I can see where it would work, but it needs to work before I'm convinced it's anything more than an elaborate expectation script. Give me a practical example where this works BETTER than already existing systems, and then we can talk.
Again, it's early on, but I'm still skeptical of all the tech.
Google will fill in the rest for the curious.
I think this does have possibilities in areas such as, like stated previously, RPGs.
Let's say I want to buy a Sword of Awesome +2. I'd walk into a shop and say to the owner "Hey, have anything for sale?" The keyword "hey" would signify that he could return the greeting, and "sale" would put him into 'selling crap' mode. Tone of voice could also be used as a tool for haggling - if he's a jovial kind of person he'll respond well to a lighthearted "you know, another five gold off that price would sell it...", or perhaps he's more stubborn, requiring fast-paced and brash negotiation. Even the most basic of NPCs would get another dimension to their personality, to say nothing of informants or interactions with party members.
While obviously still years off solid implementation as much more than a gimmick, there's a valid line of reasoning for attempting the technology.
Edit: Not to say this hasn't been tried before. It just hasn't really worked all that well due to limitations in technology. With enough work (in years, maybe even decades), I could see this being a much more natural and quick way of interaction compared to radial menus and written dialogue options (just like head-mounted displays for vision).
Old PA forum lookalike style for the new forums | My ko-fi donation thing.