Search, AI, and the need for deep concern and caution

I have been monitoring the AI-sphere since the Nov-Dec 2022 release of AI-generative tools such as Chat-GPT et al. I have developed strong opinions about what has happened and what will happen regarding AI-assisted search and returns.

My deliberate, cautious approach, especially the need for vetting information with a jaundiced eye, will become increasingly necessary as the ready-or-not rush to assimilate disruptive AI-assisted search, summation, and presentation technology affects our knowledge base. Whether we want it or not, we are told all this is necessary for unimaginable future opportunities and boundless progress. “Move fast and break things (and maybe fix it later),” and “it is better to ask for forgiveness than permission” is their SOP attitude. Submission is not optional. There is no opt-out. The genie is out of the bottle; the toothpaste is out of the tube. There is no going back.

I could write hundreds of pages about what I have observed, read, and seen presented as truth, “received” knowledge, and visual reality— what we/they think we/they know that isn’t so. But for my purposes here, I will limit my comments to information retrieval and why I have this unapologetic sense of dread in this field.

We have been dealing with so-called “hallucinations,” as well as pre-vetted and biased, preferred, truncated returns programmed by humans with agendas—don’t ever forget that. This has accelerated since the inception and public introduction of these tools and toys. Have you consciously realized and recognized this?

Almost daily, we can read (and I do) about some new problem with AI. Of particular recent note (early 2025), is that the training models are heading for a brick wall: They are running out of available material to scrape and are looking for more and more sources because their training is incomplete. This has been obvious to anyone who has wasted time with these tool-toys.

Putting aside the ethical and legal concerns over their training methods to date, a newer concern is that scraping and circular training on their own corrupt content is evident and will continue. We are told not to be concerned, that the models are constantly improving and will be a diminishing concern. Even if true, the circular training presently occurring, and that will continue as the problem allegedly diminishes, will continue. This is a disaster in the making.

Will you know the difference? Good information used to make essential decisions requires real intelligence, experience, and disciplined skepticism. The content and the tech must be vetted all the time.

Not all available information is available on ‘the web.’ Thus, “the web” is a vast but limited resource. With the “help” of AI, it is becoming a noxious cesspool at warp speed. Be aware and beware.