Think about the early days of digital photography. When digital cameras first emerged, expert critics from the photograph field were quick to point out issues l

Think about the early days of digital photography. When digital cameras first emerged, expert critics from the photograph field were quick to point out issues like low resolution, significant noise, and poor color reproduction—imperfections that many felt made them inferior to film. Yet, those early digital cameras represented a breakthrough: they enabled immediate image review, easy sharing, and rapid technological improvements that soon eclipsed film in many areas. Just as people eventually recognized that the early “flaws” of digital photography were a natural part of a revolutionary leap forward, so too should we view the occasional hallucinations in modern LLMs as a byproduct of rapidly evolving technology rather than a fundamental flaw. Or how about computer graphics? Early efforts to move 3D graphics hardware into the PC realm were met with extreme skepticism by my colleagues who were “computer graphics researchers” armed with the latest Silicon Graphics hardware. One researcher I was doing some work with in the mid-nineties remarked about PC graphics at the time: “It doesn’t even have a frame buffer. Look how terrible the refresh rate is. It flickers in a nauseating way.” Etc. It’s interesting how people who are actual experts in a field where there is a major disruption going on often take a negative view of the remarkable new innovation simply because it isn’t perfect yet. One day, they all end up eating their words. I don’t think it’s any different with LLMs. The progress is nothing short of astonishing, yet very smart people continue to complain about this one issue of hallucination as if it’s the “missing framebuffer” of 1990s PC graphics…