What Meta is looking to do with ImageBind is to expand this into a more board input range. ImageBind can link text, video, temperature, depth, and even motion data all without the need to train the model before the input. The goal is to move the generation of an environment to a more human-like one. Right now, your brain is taking in thousands of data points and processing them to inform your current environment. Meta wants image bind to do the same thing and are even interested in including brain fMRI signals into data points that ImageBind can ingest (what could go wrong there).
Eventually ImaageBind would be used to create hyper-realistic environments in VR… meaning the Metaverse. Now, I hope I am not the only one that is thinking about “The Matrix” right now. Or, perhaps you are thinking of something more sinister like the IOI Loyalty Centers from “Ready Player One”. Don’t get me wrong here, I am fascinated by the work and the concept. Still, we are talking about a system that is capable of perceiving and creating an environment without any ability to appreciate it or truly understand it. The way humans perceive something is not just sensory input or ones and zeros, there are other factors that are added in such as an emotional response to what we see. As of yet, no AI is capable of mimicking human empathy or emotion which makes their rate of advance concerning to say the least.
Meta and other AI proponents sees tools like ImageBind as a benefit to content creators, and as a path to creating exceptionally immersive games and virtual realities. I am not opposed to graphical improvements and certainly do enjoy a good VR experience; Valve’s Alyx is still one of my all-time favorite games. The challenge with AI like ImageBind is the same as all AI tools; how they are programed, the uses to which they are put, the people who control them, and the potential for abuse either intentional or unintentional. With AI being the next big shiny thing and part of a new technology race, the groups that are developing these tools have an even greater responsibility to slow down and ensure they are not missing something in their mad dash to be the first. It would not be the first time rushing a project resulted in something that the developers wish they never created in the first place. I just hope this rush to AI does not end up as one we regret but are not able to do anything about.
