What we are doing here requires absolutely zero intelligence, or at least only as much as it does for walking, driving, or lying flat on your stomach.
That, however, doesn’t mean we just dump whatever we find and hope for the best. The vast majority of data, even for the English language, is unstructured and not immediately usable without some effort put into it beyond curation. For Hmar, this is true as well, and it is even more critical that we take the initiative to collect, create, and structure it to some level, even if it isn’t standard and ready to use out of the box, because there is very little data out there for our language in a form that is ready for this purpose.
And while this may seem complex or intimidating for the less technologically savvy, it’s simply due to ignorance of the simplicity of the task. YouTube tutorials tend to mask the fact that 90% of “making” an AI is in the boring work of collecting, scanning, digitizing, and structuring data, with the remaining 10% divided among generating scripts, fine-tuning, and testing the AI.
There is no intelligence required in the whole pipeline anymore because so much intelligence has been built into these systems that, for what we wish to accomplish and enable, we just need to follow a predictable routine. The technology has matured enough to enable people like me to have the courage to undertake the initiative, and that comes from recognizing the progress we have made as a collective. And I think it’s fair to say that this isn’t a technical initiative so much as it is organizing human effort.
Anything that actually requires actual thinking requires a team of scientists and a whole lot of money. But we don’t have to worry about that aspect of developing AI. That work has already been done for us. All we need to do is make sure that we build that bridge so we could cross over.