Data quality didn't suddenly matter because of Gen AI

I like how people keep talking as if data quality just became valuable because of Gen AI, and not a truism for data applications always.

That framing also pre-supposes how Gen AI will be leveraged (and often just assumes Gen AI has to be implemented 🙄). I can think of many Gen AI use cases that do not touch a company's existing data (context based routing of new customer cases being an obvious one).

There are dozens of these posts, all bolting two claims together: clean up your data, and you need AI. The first is true. Clean up your data because we're professionals and should act like it. The second is an assumption smuggled inside the Trojan horse of the first. "Clean your data" quietly becomes "clean your data so you can do the AI thing I've already decided you have to do."

But thoughtfully assessing your strategy, pain points and current state doesn't lend as well to a quippy statement like "AI projects fail because of data". Software projects fail because they're not planned strategically and designed thoughtfully.

The consultancies skipping discovery and handing you specifications for data remediation and a pre-supposed software project are one-trick ponies. Your problem is assumed because the solution is the only thing they sell. Software that gives you a competitive edge is grounded in your reality, not a consultant's hypothesis.

We don't trust people proposing snake oil remedies before they've even asked where it hurts. Apply that same healthy skepticism to anyone offering you a custom software solution before they've taken the time to understand your company.