@CGI

I havent read the whole thread but you were asking why this is just smoke and mirrors or not realtime.

The demo was damn impressive handing over a sheet wasnt the most impressive thing also not the interaction between the player and milo.

See whats not possible and wont be possible for a long time is such a fluid conversation with an AI.

lets see there are two kinds of AI a strong one and a weak one. That has nothing to do with the quality of the AI but with the way a AI is designed. A strong one is a AI which is able to learn it learns in a similar way a human does. Those AIs are stupid and we are not even on the level of an insect here. Strong AIs requires atleast 50-100 years to be at a point which could be seen as partially inteligence maybe even more.

Ok so all AIs we have around us are just weak AIs strong AIs are things for laboratories.

A weak AI dont really tries to be intelligent a weak AI tries to trick people to think it is intelligent. They are a lot of work because in the weak AI every situation has to be programmed. There is a socalled turing test. Some decades ago there was a guy who said. If a human can communciate with a machine and cant recognize it as a machine then it passes the test. Until now there was no programm or computer who was able to pass the turing test.

And try to talk with a machine about things if your conversation is on the typical chat level then maybe a machine can trick you but its without a doubt no problem to trick even in 1 minute the world most smartest weak AI.

lets see the video and think about it this kid remembers things which is bigger then people think. The question is how does it remember things. How can the kid remember context how does the kid understand context. Distinguishing feelings is also a big thing but its tiny compared to the other 3 really technically impossible things.

This thing will be smart for todays standards but the communication wont be like in the video. The kid will remember just certain key things. And probably understand just a few things out of its own world but not really understand also you wont be able to freely talk with it. There will be tight boundaries and inbetween those boundaries you still will face problems in the communication between you and Milo.


Dont expect to much of the AI. The video is PR the interactions are possible (sheet of paper and voice scanning after feelings and maybe even some interpretations of your face) Still it will missinterpret what you will do a lot of times and people will be pretty fast bored because people make a wrong picture in their mind about how this milo will react to them.


Those whole conversation was scripted or preprogrammed in a way that the actor knew what he is allowed to say.

It will probably still be amazing (atleast for me) but only if you know that the Ai technology is still at the beginnig. We still dont know how the brains works exactly we are at the beginnig of this technology. But if we compare it with flying: 

The humanity has not even reached the point off building the first plane. We have not even a proven (good) theory and models how flying could work. See Newton is not even born yet we wont build a plane anytime soon.

Milo will remember stuff which has happened in its own world like catching a fish two days ago.
Maybe it will be able to remember certain things the programmers allowed him to remember like your hair color but I dont think it will be able to remember that you have an aunt named kelly or just say remember this words: XXXXXXXXXXXX
and ask him two days later if it remembers what you said two days ago.

But thats not what it is about its just a demo that shows you that you can programm a game that can transfer your words into the right letters and save it and that it can scan certain things and that it can distinguish between different voices and stuff like that. Its not about the intelligence. It will be like a very advanced tamagotchi with some cool features.

The point will be if natal and 360 is close the flawless in interpreting your actions. The furhter processing of informations will be a showcase how far the technology has gone yet. But wont be better then the best projects to simulate intelligence (which are imo not far yet)

don't you guys play ME it's a VI

but beside that.... where did you guys pull out that tech possibility is years away???? heuristic AIs maybe... and the limitation is more from the fact that we don't have the technological know-how than the tech capabilities itself. working VIs already exist in industries in particular.
tech prophet estimate mass produce computer chips "matching" the brain processing power will be available by 2020 and some push it to say that by 2050 we'll match the world population combine.... so tech years away not really...
without saying that voice recognition and pitch analysis have been improving really fast and are almost fully reliable now...

i really don't think tech capabilities is the issue here, but simply that we don't know how to develop a working code for heuristic AI that is complete and efficient....

now Milo and the demo was really basic and I believe that such basic interaction is already possible... they didn't exactly ask him to give us The Answer to the Ultimate Question of Life, the Universe, and Everything

Microsoft promises Milo is the next step.
like HDDVD?

Woe. Peeps. Poor CGI.

Dont get on his back. What happened to not guilty until proven?

M$ said it was a realtime demo. Now. It's possible the tech shown was simpler than what we thought. Maybe a set number of modes for the demo. Maybe the AI could think for itself within a set perameter. But CGI simply said there is no proof against it other than fanboy nonsense.

Yes it does seem to good to be true. Can they actually pull the level of freedom off that they are portraying? Who knows, but if Molyneux is making Milo then we will all know within 12 months.

You know this kind of "game" is not what I got into gaming for. I don't want an "imaginary" friend. Children should be encouraged to go out and play in the real world outside, not talking to imaginary boys and girls on their Xboxes.

Although I can see how this technology could make many other games totally awesome like RPG's. Allowing the characters in your party to actually interact with you would be amazing and actually be "roleplaying" for once!

However, employing this interactivity in a game the size of say, Mass Effect 2, would be a collosal effort of programming that just wouldn't be cost effective and just thinking of the amount of coding involved and the man power makes me shudder.

This may be the future, just not in this decade or probably the next.

I just would like to point out that a journalist went to Lionhead and saw a picture of Milo getting sucked into a portal. So, the game will have fantasy elements in it.

Actually, the effort might not be as colossal as you might imagine. This is like a big step in creating more realistic game character AI. If a developer can work out the problems with speech synthesis, speech recognition (someone mentioned 75% or 85% or something) and perfect these things, and then develop an AI that can learn and "think" and understand (but don't give it power/control please, we don't want SKYNET), they could license the AI the same way Epic licenses the UT engine. It's just the effort it will take to get to that point will colossal and may take 1 or 2 or 3 decades, but after that, things may seem easier.

The cool part is that programmers/audio engineers are working on speech recognition/synthesis outside of the industry, so it's not like it would be upon a game developer to figure everything out themselves. The same with AI research. Then again, science is progressing so rapidly anyways that anything could happen.

There's a key part to all of this that some people are missing/skipping past, according to Molyneux, you're not interacting with a true AI, but it's the illusion that you are interacting with an AI. But this is the first time something of this magnitude is being put in a game, hence it's a big step.

Here is an interesting article on voice recognition.  A Live connection could enable incredibly accurate voice recognition.

http://www.msnbc.msn.com/id/36151627/ns/technology_and_science-tech_and_gadgets/

Now, the availability of 3G-enabled mobile devices with fast, always-on Internet connections and the ability to train voice-modeling software with millions of phone users — a process called crowd sourcing — is helping fuel a new breed of mobile speech-recognition apps that work quickly and are amazingly accurate.

 

 

 

"The power of the masses
Mobile voice-recognition apps also have other advantages over their older desktop counterparts.

One is the ability to communicate with powerful central computers, or servers, that can combine information from millions of users and then make broad generalizations that help improve the apps' overall ability to recognize words.

"The first time you speak to the phone, we put a cookie" — a kind of digital tag — "on your device and when you say something we call up your personal language model from our servers and use it to get better accuracy," said Dave Grannen, president and CEO of speech recognition software maker Vlingo, which also has an app for the iPhone.

An individual's voice model contains information about his accent and unique way of pronouncing certain words, among other things.

The servers can combine the voice models of several speakers who have similar accents to improve the accuracy for that population.

"If you're from India and speaking English as a second language on Vlingo, we work pretty darned well. If you're from Germany speaking English, it doesn't work so well," Grannan told TechNewsDaily.

The reason? Vlingo has many more Indian-speaking users that German-speaking ones, so the voice model for Indians is generally better than that for Germans."

No you are overestimating what was done until now.

The computing power of the brain is not the point. See I can explain you whats the problem. First a Brain works complelty differently we wont ever  be able to emulate a brain with a standard Computer never ever. Because its impossible. The brain is the complexst thing in universe (as far as we know) we cant name a thing which has a similar complex structure. 

So problem one the achitecture to emulate brains is not there. We dont have a computer which can be as fast as the brain. All those numbers with 10 TFLOPS o 100TFLOPS are complete bullshit. 

Because the brains speed cant be measured in those numbers. We have way more neurons then a Chip transistors. We have trillions of neurons in our brain and the problem is that those neurons are working at the same time all together. And we dont even know how they work because it seems that there are quantum effects in the brain because the neurons start working without any interaction so we dont have the slightest idea how our brain works. We can name some areas and we have a simple model which to describe some stuff but even this stuff istn described properly.

AI is still a dream we dont even know whats intelligence. No computer has passed the turing test no programm nothing. And to do that you dont even need real intelligence. we are nowere AI development made just a little progress. Strong AI development made no prpgress at all. See a Xbox360 may crunch numbers better then the brain like most computers but still it isnt better in any way then the brain. 

This milo demo and the talking between the both was in between strict boundaries and the actor knew what she was allowed to say. 

If you dont believe me read some medical books about the brain and invest a little time to read what the industry has truly acomplished until now. You will be disappointed. the computer technology as it is today wont ever be able to realize intelligence not even dog like intelligence. 

Computer technology is just not able to execute a truly intelligent AI. Because it cant emulate the processes which are talking place in the brain. All transistors if they would equal one single neuron and would be connected with the other trillions of transistors directly then there would be a base for intelligence to work. But even if we would have such a magical machine we wouldnt be able to programm it because we dont know how. All the hollywood stuff all the promises out of the tv is like the promises they made 50 years ago with fusion reactors, Today it will take again 50 years to realize the concept and in 50 years another 50 years. People cant see the difference but to create true intelligent beeings is an even bigger problem then to bring a piece of the sun down to earth and use it as an almost infinite power source. We will be dead a long time before we will see truly intelligent machines. But our hope relies on weak AI:

maybe weak AIs will be so complex in future that they will be able to do some things Like communcating with a person about a certain theme we will see that in 20 years those programms will be there everything will be preprogrammed and the development will cost millions. 

In those programms programmers will try to programm all possible situations the AI could be in later. And Milo will be such a thing but those programms are not as far as people think they are even in the web you can google them and talk with them milo will be on a similar level hopefully significantly more advanced. And with some cool features the point will be the interaction.

But I cant help me I am sure some people will be pretty disappointed when they play this game. It will be amazing but dont foucus on the intelligence too much because if intelligence would be so far like the video tried us to show then we would see AIs in way more areas AIs would be everywhere talking with people and trying to solve their problems but they arent the industry is still not existing.