Humans have to work with a very large vocabulary. sometimes in acronym form, sometimes in unpronounceable terms and yet others times in overloaded terms and/or reused acronyms and terms. Properly understanding requires context in addition to vast memorization capability. In addition the ability to comprehend concepts is invaluable.
AI/ML is able to accomplish this but without true comprehension rather, in the case of most current models, a probability based semblance of understanding.
Some humans excel at this, some less so than others. Essentially, as with most things, there is a bell curve representing how humans are dealing with this knowledge phenomena.
The AI/ML phenomenon extends beyond just language. Depending on the modality—whether it’s text, image, audio, or another form—and the way data is tokenized and embedded into tensors, these systems can process a vast range of input types. Their capacity to generalize across modalities reinforces the illusion of comprehension. Unlike humans, they can do this across datasets of a scale and diversity that no individual could ever realistically engage with.
Leave a Reply