Why so bad?

Because for the typical office - documents are strewn about on random network drives and are not formatted similarly. This combined with the inability to nail down 100% accuracy on even just internal doc search is just too much to overcome for non-tech industry offices. My office is mind blown if i use Gemini to extract data from a PDF and convert it to an .xlsx or .csv

As a technically minded person but not a comp sci guy, refining document search is like staring into a void and every option uses different (confusing) terminology. This makes it extra difficult for me to both do my regular job AND learn the multiple names/ways to do the exact same thing between platforms.

The only solution that has any reliability for me so far are Gemini instances where i upload only the files i wish to search and just keep it locked to a few questions per instance before it starts to hallucinate.

My attempt at RAG search implementation was a disaster that left me more confused than anything.

It's in the name: generative AIs.

There are very few use cases at companies where you need to generate something. You want to work with the company's often very private disparate data (with access controls etc.) You wouldn't even have enough data to train a custom LLM, much less use a generic one.

Turns out that garbage text has very little intrinsic value

In my experience is that LLMs get you 80%of the way to a solution almost immediately but that last 20% when it comes to missing knowledge, data, or accuracy is a complete tar pit and will wreck adoption. Especially since many vendors are selling products that are wrappers and provide generic, non-customised solutions. I hear the same from others doing trials with various AI tools as well.

Any consumer facing AI project has to contend with the fact that GenAI is predominantly associated with "slop." If you're not actively using an AI tool, most of your experience with GenAI is seeing social media or Youtube flooded with low quality AI content, or having to deal with useless AI customer support. This gives the impression that AI is just cheap garbage, and something that should be actively avoided.

I think one reason for this is that LLMs are sort of maximally if accidentally designed to fuck up our brains. Despite all the advancements in the last five years I see them as still, fundamentally, text transformation machines which have only very limited sort of intelligence. Yet because nothing in history has been able to generate language except humans, most of us are not prepared to make rational judgements about their capabilities and those of us that may be also often fail to do so.

The fact that we live in an era where tech people have been so investor pilled that overstating the capabilities of technology is basically second nature does not help.