Presentación de Muse - El primer modelo generativo de IA diseñado para idear juegos - Vídeo Dailymotion

3djuegos

Ver Presentación de Muse - El primer modelo generativo de IA diseñado para idear juegos - 3djuegos en Dailymotion

Transcript

00:00Hey, it's great to be here.

00:02You might have seen like some exciting news came out about

00:06work that our teams are doing on the world human

00:09action model.

00:10So Katia and Dom Katia from our Microsoft Research team.

00:13Dom from Ninja Theory. We're here to talk about this

00:17interesting work and how it impacts our kind of future

00:20in research or future in game development.

00:23It's also interesting note that this model was just published

00:27on Azure Foundry.

00:28The weights for the model were just published on Hugging

00:31Face.

00:32Because we want to make sure that the industry of

00:35creators and researchers are able to engage with this work

00:38and we can all learn together.

00:40We definitely look at this work as a unique opportunity

00:43to kind of leverage the intersection of art and science,

00:46which is what video games are to grow with the

00:49creator community.

00:50That's what these tools are about.

00:52We think there's a really unique opportunity to use technology

00:56to do things that they've never done before.

00:58We have a great great opportunity with two teams that

01:02happen to be frankly right next to each other in

01:05Cambridge in the UK in our game studios and Microsoft

01:08Research and it's been awesome to see the collaboration with

01:11between those two teams. So what I'm going to ask

01:14is I want to have Katia talk about how they

01:17see this technology and the work that they've done.

01:20Absolutely thank you so much.

01:22Fantastic to be here and reflect on the work that

01:26is coming out today.

01:27A key to this work from the very beginning was

01:31to try and understand how we can really ground the

01:35development of generative AI models in an actual understanding of

01:40what users need.

01:41In this case, what are the kinds of model capabilities

01:45that would be needed in order to really empower creatives

01:49to let them explore ideas more quickly,

01:53more deeply than they might be able to do with

01:57traditional tools.

01:58And so in this work we grounded our generative AI model

02:02development in a user study with 27 game creatives and

02:07we were able to tease out that aspects like iterative

02:11tweaking and divergent thinking,

02:13which have already been well established in the computer supported

02:18creativity research. Those continue to be really key when creatives

02:23want to engage with these types of models,

02:26but often they're an afterthought in model development.

02:31So we turned that on its head.

02:32We started from those capabilities and then try to really

02:36systematically push on the model capabilities that might be needed

02:41in order to support tweaking,

02:43ideation, rapid interaction. And so that is how we

02:48developed WAM and focused specifically on those model capabilities of consistency,

02:53diversity and persistency. And you can read a lot more in

02:56the research publication and the demo videos associated with that.

03:00I'll tell you one of the things that I get excited

03:04about.

03:04You know, one of the things we care a lot about

03:07at Xbox is game preservation,

03:09and I think about an opportunity to have models learn about

03:13older games,

03:13games that were maybe tied to unique pieces of hardware

03:17where that engine on that hardware is kind of time

03:20will erode the amount of hardware that's out there that

03:23can actually play a game.

03:25But you can imagine a world where from gameplay data

03:28and video that a model could learn old games and

03:31really make them portable to any platform where these models could

03:35run.

03:35I think that's really exciting.

03:37We've talked about game preservation as an activity for us

03:41in this.

03:41These models and their ability to learn completely how a

03:45game plays without the necessity of the original engine running on

03:49the original hardware,

03:50I think opens up a ton of opportunity.

03:53But the fact that this is research that was done

03:56on top of bleeding edge for Ninja Theory Dom,

03:58why don't you talk about how bleeding with the role

04:01that bleeding edge plays and how you as a creator

04:04think about this technology?

04:05Yeah, thanks Phil. Yeah, so as you mentioned,

04:08catchier and catches team and our team here in Ninja

04:11Theory.

04:11We happen to be in the same city in Cambridge

04:14in the UK.

04:14We know each other so when we joined Xbox

04:17Game Studios we were introduced to catcher and we started

04:20to think about how could we collaborate and we've been

04:24working together now for a few years and catcher asked

04:27if they could use bleeding edge as a multiplayer catalog

04:30title as a testbed for their research and for this

04:33technology.

04:34We were happy for them to do that and what

04:36we're looking at today is really the fruits of the

04:39work that catcher and the team have done.

04:42And I think for us as a as a game

04:45creator,

04:45the thing that I'm always really interested in is how

04:49can we help our teams do more?

04:51How can we help them to push their creative ambitions

04:55and their creativity even further so?

04:57Technology like this for me and I think for our

05:01studio here at Ninja Theory is not about using AI

05:05to generate.

05:06It's about using AI to generate the results that we

05:10want.

05:11It's not about using AI to generate content,

05:14but it's actually about creating workflows and approaches that allow

05:19our team here of 100 creative experts to do more

05:23to go further,

05:24to iterate quicker, to ideate quicker,

05:27to bring their what's the ideas in their heads to

05:30life in a tangible form for others to to see.

05:34So you know, although this is technology that we haven't

05:38used in the creation of our games and we don't

05:42intend to use this technology for the creation of content,

05:46I think the interesting aspect for us that is exciting

05:50is how can we use technology like this to make

05:53the process of making games quicker and easier for our

05:57talented team so that they can really focus on the

06:01thing that I think is really special about games,

06:05which is that human creativity.

06:08Games to me are really creators using the medium of

06:12games to talk to their audience,

06:14to communicate through the medium of interactivity,

06:18and that is core to game creation for me.

06:21So what's important with this is that we have the

06:25human creativity,

06:26but through this technology we can empower them to do

06:29more and push their own dreams further than they have

06:33been before.

06:34Yeah, like we're always thinking about how do we put

06:38our creativity in the hands of creators so that they

06:41can create things that they haven't been able to create

06:45before? I know you and the team will continue to

06:48make progress. We obviously have Game Developers Conference coming up.

06:51It's always a unique opportunity for us to talk about

06:54more of the innovation that's happening between our creative teams,

06:58our platform teams, and our research teams,

07:00and I'm really excited about what we will get to

07:03show next.

07:04But congrats again on all of the progress.

07:06Thank you very much and thanks for the fantastic collaboration.

Presentación de Muse - El primer modelo generativo de IA diseñado para idear juegos

Categoría

Transcripción

Sé la primera persona en añadir un comentario

Recomendada