The thoughts on how important user friendly interface is itself an ongoing issues in all types of software, but one common thing unites all of their respective users old or new alike is that we all have to “google/search” for the answer and instruction when problem arises
task such as opening browser > google > type issue > find best match > search x forum > post question > WAIT etc is 2nd nature today, but can we save a few of these steps and make it easier for users? Can we have some form of ways to solve such mundane problems locally? like how typo/syntax error/code completion can be dealt automatically and immediately without any further ado
I felt that a personal assistant AI train inside of blender/any software or cloud that could answer or solve problems based on “learning” the data that has already existed here on BA forum will have the benefit of greatly reducing the effort of community members on an already solved issues, saving both time and bandwidth for everyone thus improving users experience and productivity
Imagine a basic chatbot interacting with new users, guiding them
user > press spacebar > “Where is x button?”
response > “it’s inside x mode, properties menu”
or an advance google assistant like AI that can perform actions inside blender
user > press spacebar > “make this object walk with wasd keys”
response > “done”:evilgrin:
With the ongoing machine learning revolution, this sound like a necessary next step for all software
so I’d like you all to share some of your thoughts on this here and discuss how this could be done
Also I’m currently working and exploring the possibilities of such but while there are a lot of chatbot libraries out there, its the training data that I’m having trouble here, so it’d be great if anyone can share the best practice to collect or data mine this site for all solved question for training
This definitely is an interesting idea. I think that if that the blender devs could make a thing in blender that collects usage information about issues people experience in blender and how they solve them, there could be a native implementation to help blender users solve problems dynamically.
@Lostscience
Although I’m learning tensorflow and keras atm, I’ll mainly be using python chatterBot, the AI will only be readily trained to answer short question related to the bge operations, so its operating under a retrieval model inside specific(bge) domain, since we already have tons of data here on BA and perhaps including the rest of web(reddit stackoverflow etc), this should be a cake… when I figure things out
Thanks for the links and lol youtube! it’s so 2nd nature these days that I strongly believe an ever increasing AI aided and driven environment will continue to empower and make us even smarter
@Nicholas_A
an add on or something may be sufficient but it’s still in its early stage, I’ll see how far I can take and learn with this small project, it won’t win Loebner prize but it should help increase user friendliness and efficiency in using the game engine