question about the future of AI

Comments

• wolf359 Posts: 3,940

  February 2

  Generative AI is a separate ecosystem.
• sidcarton1587 Posts: 62

  February 6

  Toobis said:

  I'm pretty new to AI generally as most of us are but I ewas jkust wondering and partially even worrying that as it gets more and more powerful over time, will it be to the point that product vendors would be portentially losing their income due to AI daz (and any other program like Daz) simple but powerful content creation tools? I maybe talking space-age here but could it really be a case of typing what you want the AI to create for you in a Daz scene for you to work with and it comes up? for example a certain car you might want to add and then add genesis characters to etc to create a final render. What is the end game with AI and content creation please. Will it really be possible? maybe I've just got my head in the clouds. lol.

   

  It may sound trite, but it's a question of "what do you mean by AI?"

  The initial hype around AI focused a lot on generative AI -- submit a prompt and the AI produces content for you based on the prompt, whether it's text or images or sound or whatever. Some people view this as competition for the 3D ecosystems which exist to model and create, in most cases, the same type of end-product that generative AIs do. I say the same type of product, not the same product. 

  As AI has evolved, however, we are adding an entire class of problems that AI is now starting to address. The popular term is "agentic" AI. Basically you're creating agents that will solve a part of a problem -- fetch X, get me the best price for Y, summarize Z, email X, etc. What this means for 3D ecosystems is still up in the air, but what you are describing is getting closer to reality, but it's not there yet. The reason is that the applications themselves are not exposing their interfaces in a way that lends themselves to being used by AI agents. 

  DAZ is a good example of this. It's meant to be a desktop application driven by a human. It has a scripting and plugin capability that can automate some of the work that a human would do, but it doesn't create new content on its own. It resolves the repetitive, tedious tasks that a human would have to sit there and click 50 times. It can't yet take a natural language scene description and populate and pose the 3D assets and render and environment settings into a DAZ scene. I say not yet, but it's something that you can see on the horizon. It's already possible to take a scene description and at have AI least identify the assets that could be used to populate the scene, for example. As asset descriptions and metadaata are more fine tuned to be read by AIs, for example, searching for them using narrative descriptions will become more and more accurate. We can do similar things for posing, expressions, lighting, environments, and so on. The missing pieces are primarily the description and metadata to make them truly searchable in a way that lends itself to AI scene builders. 

  The AI scene builder is the next big frontier, taking all these AI-enabled blocks around asset finding to actually build a scene. This has a number of hurdles that are still not easy to overcome. For example, if your search for an asset, pose, or expression returns multiple results, how do you refine it? What does the human interaction look like at that point?

  It's an interesting problem, and I think as the AI landscape evolves we're likely to get closer to that the next big frontier.