I've been keen on AI research since thinking I made progress on the fundamentals of knowledge. The search for answers was the great joy. But the chances of being right must be vanishingly slim. Nevertheless I patented compression algorithms and structures in US 6,414,610 and US 5,748,955 (later cited by Google, Oracle, IBM and others) related to the progress I thought I'd made. I also felt LLMs are on the wrong path.
Personal case: I'm currently trying to get about $150,000 wrongly withheld by the largest and most corrupt organ of state hereabouts. By "most corrupt" that's mainly a comment on the quantity of corruption, not the quality.
My MSD case started in November 2016, and a quick overview of 2017's conspiratorial fraud gives a brief outline of the first of the nine years. The way justice works around here, I discovered, is that politicians (apart from the current prime minister) don't give a toss unless they think votes are involved. That means making grievances public even thought doing that can be incredibly hard.