Summary:

Milo and Kate is Lionhead's newest IP and was announced at the E3 2009 conference as part of Microsoft's project Natal demonstration. Players will take part in the childhood of Milo (or Milly) by interacting with them through the Natal interface. The game will also feature Kate, Milo's dog.

Milo represents Lionhead's latest work on emotional AI and is the culmination of the project formerly known as Dimitri. The demo shown at E3 also incorporated technology from other Microsoft projects, such as voice recognition. Milo will respond to players movements, facial expressions, vocal intonation, as well as interacting with objects the player is holding.

E3 presentation: Gametrailers (SD and HD available)

Hands-on previews & interviews:

Kotaku 02.06.2009
Eurogamer 02.06.2009

Thanks to those people who have contributed to this thread, including: Salimort, Glen Watts, and fablepower.

Video links? Please? Pretty please?

---

Futurama is back, baby. Complete with the original cast!

Are Lionhead going to post a link to the vid they showed at E3? It was very impressive, I really want to show my friends.

ErodeTheSoul:

Video links? Please? Pretty please?

Gametrailers will have it up soon enough... They already got a lot of other videos from the conference just keep refreshing.

---

Salimort:

Are Lionhead going to post a link to the vid they showed at E3? It was very impressive, I really want to show my friends.

Ditto to that. I was astounded by it and they all want to see why I'm so excited. Waiting for the conferences to archive takes too long

EDIT:

maggot:

ErodeTheSoul:

Video links? Please? Pretty please?

Gametrailers will have it up soon enough... They already got a lot of other videos from the conference just keep refreshing.

Oh trust me, I am. I'm also constantly refreshing G$, Gamespot, IGN and Kotaku, looking for anyone with that video

---

Futurama is back, baby. Complete with the original cast!

Oh my god! Lionhead never fail to dissapoint! Aaah, if only there was a release date! So excited! When PM walked on, i literally screamed (almost).

Fable

---

~Chazza~

Milo has evolved the whole concept of NPC to human communication. (which is a thing mind).

---

Competing in the Olympic Hide and Seek event in London 2012.

Fable445:

Oh my god! Lionhead never fail to dissapoint! Aaah, if only there was a release date! So excited! When PM walked on, i literally screamed (almost).

Fable

They never fail to dissapoint ?

---

maggot:

Fable445:

Oh my god! Lionhead never fail to dissapoint! Aaah, if only there was a release date! So excited! When PM walked on, i literally screamed (almost).

Fable

They never fail to dissapoint ?

lol, fail happy comment.
@Maggot: Not your comment.

OOHH, I'm so excited about Milo.

Found a vid:


http://www.youtube.com/watch?v=LgRCToXfef8

Quality is absolutely nasty mind.

Go here

At about 1:12:00 you have peters bit with milo

---

Just in case you don't know you need air to breath.

Kotaku have had their live demo.

Seems like its not scripted
This sounds amazing technology is advancing to fast.
Eurogamer Interview With PM

---

Ooh! Pretty cool. And I was right about the trampoline and the school. Damn I'm .

---

It's interesting that it doesn't appear to meaningfully parse sentences in terms of word content. Well, I assume not at any rate - I think I read in one of the interviews its vocabulary had built up to about 500 words after it had been used for a while, which I suppose is substantially less than text adventures of 25 years ago. But I'm very impressed by how it looks for key words and - as far as I can tell - knows to some extent about intonation of sentences to give them a little more context. Combining two relatively simple inputs to infer more complexed meaning as it were (I'm a little unclear on how much the visual component adds to all this). But I suppose in reality you don't need perfect AI - ultimately you just need something that fakes it well enough to give an impression of meaningful reaction.

I'd be very interested to see how well it does 'cold'. That is, with someone entirely new using it for the first time. I mean, I assume it draws more connections as you interact so you're left with less random moments as time goes along, but can it fake it well enough at the beginning to keep people's interest long enough to get to that stage? For some reason I keep thinking of how you train voice recognition software like Dragon Naturally Speaking etc (although I'm aware it's very different software), and wondering if there's going to be similar periods of adjustment before interesting stuff happens.



Oh, and I can't help but think the Mainstream press will have a field day with this. Interactive Small Children for people to fiddle with is just waiting for a press mauling, no matter how unjustified.

---

Yer I can imagine it getting some complaints by some press (but what doesn't) gamings certainly going to change though over the next few years, if they can really perfect natal how they did in the demo video then we could lose controllers all together, (what they're trying to do) but then what do we do about shooting games I'm scared.

---

Depends. It's basically another controller, so for it to get significant support I suspect it'd have to be standard issue with the next console. If it's an add-on, we're pretty much looking at another EyeToy in terms of market use. That said, the EyeToy clocked up something like 11 million units sold...

Maybe Microsoft are planning to do a Wii on us - their next console dumping traditional controllers in favour of Natal. It'd be a hugely bold move for a traditionally quite conservative company though

Still, they're not exactly the only company working with cameras in gaming right now, with both Sony and Ninty having their own offerings of various flavours. And it's undeniably a direction I really want to see controllers taking. The more freeform and less gamepad/mouse/keyboard we get the better in my opinion. Once the fine control aspects are cracked the potential is huge. Still a fair way to go yet though.

---

matneee:

I think I read in one of the interviews its vocabulary had built up to about 500 words after it had been used for a while, which I suppose is substantially less than text adventures of 25 years ago. But I'm very impressed by how it looks for key words and - as far as I can tell - knows to some extent about intonation of sentences to give them a little more context. Combining two relatively simple inputs to infer more complexed meaning as it were (I'm a little unclear on how much the visual component adds to all this).

Well, in the demo it seemed to recognise that Claire was smiling. And in one of the interviews Peter says that it can recognise bags under your eyes and will respond with 'You look a little tired.' But I just wonder to what extent? If I painted black under my eyes would it be tricked? If I wore a Joker mask would it think I was always happy?

---

The only thing worse than beating a dead horse is betting on one.

First thoughts;
amazing,

now please get microsoft to join this with  wolfram research's "wolfram alpha engine", stick it on supercomputer and we have multivac, (or the first steps towards skynet/HAL/...)

read this if you dont' know what multivaac is:
 http://filer.case.edu/dts8/thelastq.htm


Second thoughts:
now on pc please, ideally integrated into a future version of windows.

This is an interesting AI simulation, but I'd like to see more substantial "games" demo'd with Natal, things less "Wii Fit" or tech expo demo and more "Fable" or "Halo", games that are not driven by the tech they use. Yes, it is a great leap in terms of technology, but can it be successfully applied to current game genres in a timely manner? That said, Dmitri, now Milo, is what several of us guessed it might be, an unparalleled AI simulation.

---

Has the world ended already? Oh dear, I must have missed it. . .

deliriousstudios:

matneee:

I think I read in one of the interviews its vocabulary had built up to about 500 words after it had been used for a while, which I suppose is substantially less than text adventures of 25 years ago. But I'm very impressed by how it looks for key words and - as far as I can tell - knows to some extent about intonation of sentences to give them a little more context. Combining two relatively simple inputs to infer more complexed meaning as it were (I'm a little unclear on how much the visual component adds to all this).

Well, in the demo it seemed to recognise that Claire was smiling. And in one of the interviews Peter says that it can recognise bags under your eyes and will respond with 'You look a little tired.' But I just wonder to what extent? If I painted black under my eyes would it be tricked? If I wore a Joker mask would it think I was always happy?

Wel on the hardware demo It showed how you could take any real life thing and show it to get scanned , My hope is that it could use that tech to see the difference between what's real and what's not ,like taking a general "picture" of how your face is suppossed to look the first time you saw him , then taking note by telling what's natural (on your face) and what's not, so who knows? , it might be able to tell what's human and what's artificial

---

Will I find the light?
Or be swallowed by darkness...
I sense hope...
But foresee harm.
Will I know the blade,
And taste the blood?
I see a figure,
I see a shadow,
I see myself...

It won't work. You realize that there are tens and hundreds of thousands of words and can be combined in millions upon millions of different sentences. How much of those words can you program into it and how many sentences will it be capable of properly replying to?? You can probably ship a little booklet of phrases it can respond to but even then there's context to each possible phrase. For example if I say "That's it." or "That's it?" will it respond properly? Now think how many other ways there are of saying just about anything. Can it recognize sarcasm, anger....etc.

I don't think so. If Fable II is any indication Milo will run like a tape repeating the same 10 responses over and over and over again.

And I haven't even touched the variety of events that could be done with it. It will quickly get boring if you can only do the same 5 things a trillion times.

---

I wonder how much it would cost to impliment, in terms of money, time, and by proxy actual game content? God I sound scroog-y

---

Large Hadron Collider - because fleshlites are for wusses

Great questions! I am highly anticipating more details about the technology behind Milo and Natal. How much of the demo were uniquely Milo, and how much was enabled through the Natal API? Mr. Molyneux stated that Natal was doing the image recognition and Milo the interpretation (that is, during the fish drawing part of the demo) of the fish, or rather, the color it was drawn in But did Milo also recognize the drawing was a fish? What would he have said if it wasn't, and how does my handwriting have to be? I don't know if my human friends would recognize a drawing of a fish? How would Milo handle a bad drawing? Words of encouragement? Playful banter about it barely being a fish drawing? Either depending upon previous interactions and current interpretation of mood?

What it comes down to is most of us just don't know. Until the SDK and other tech data leaks out, our information is mostly from what we received at E3 and the few interviews that have come since. I think many wait in earnest.

Poor Earnest, he must be full.

---

Roliel/Tolanas/RR

The demo video described the picture vocally beforehand and while drawing, possibly cuing a reaction. I would like to know what Milo can actually do, such as having someone else interact a bit differently and see what changes, then reset and repeat with another person in the same video.

---

Has the world ended already? Oh dear, I must have missed it. . .