In the cube next to you, you hear, “Aw man, which version of the company logo is official?”

or

“Why do we have 18 versions of Maslow’s Hierarchy of Needs that our own graphic artists produced for separate courses?”

Then someone says, “Hey, wouldn’t it be great to share all this content? You know, like, pick the best and share it? We’d save so much time!”

Maybe. It depends. You might not. You must design the order to get the order. It doesn’t happen on its own. Throwing books, images, pamphlets, and PowerPoints into a dumpster of a shared drive is easy. Building a library that expedites user search is not. The steps you need to take to order/catalogue/rate digital data is EXACTLY the same as print data. Only the medium is different. And computers can do whatever you can tell them to do faster than a person. So, they do both dumb and smart things very fast. Humans define best, not computers. Well…computers (AI, Gen AI, ChatGPT etc.) will if you let them.

Also, there may be order that is just not apparent to people outside a team, unit, department. Don’t throw the baby out with the bathwater just so you have something to talk about at the next Dean’s conference.

Can AI help organize 1 million files that go back 13 years?

Definitely. Well hold on…do you care about meaning and usefulness and not creating a minefield of errors your users will have to discover in the middle of a PMP’s critical path on WBS? Ok, walking back definitely–Probably it can help, if you know how to write prompts, have either a method to spot check that is statistically sound to root out find hallucinations OR have emply an AI that includes citation so a person can ‘keep the first AI honest”, or finally, have another AI to check the first somehow. We’re definitely getting there. But AI is not the answer if you don’t know what the ‘manual’ methods for ordering content are.

Keywords for search

How deep is your search? File titles or a ‘deep’ look into all the content of the file. Say you have 10,000 MS Office files. Does your search only look at the file titles? Titles and medata? Or does it read the file itself? I like deep search because I don’t trust assigned keywords, especially if this was crowdsourced. If a librarian with a MS in Library Science did the keywords, I’d feel better about file title and metadata only search.

Folders with useful names for searching in

Kind of the same as keywords, but you can only put something in one folder. Or, you have the slightly more confusing possibility of having the same file in multiple folders. Or, you have keywords and also put things into a folder for the primary knowledge domain of this. It’s like the card catalog system where there was a title card and a subject card for same book. I think. It was 1993 last time I used a card catalog.

Pattern Library, an .html standard for looks and function

Is 508 important? Yes, in general. Is it important for your organization? Are you going to conform? You might do a risk assessment or cost/benefit analysis before you start down that path; it’s a long path if you have a lot of stuff to make accessible. It’s a serious task that involves reading comprehension and writing at the SME level, and likely some graphics/imagery skill. You can fool an accessibility checker into reporting compliance while alienating your vision impaired users because you didn’t actually make the information accessible.

[insert example of compliant but not accessible here; (if i don’t this is ironic example of bad alt text)]

And some user groups will just never need it. You make training for firefighters? They can all see and hear; they’re not using screen readers. Bona fide job qualifications can narrow your audience. But talk to an attorney, not me.

Do you have page or resource tracking on? Web analytics? Does anyone EVER actually use the thing you’re about to make accessible? If no one uses it, don’t waste your time.

But if you know the content will be used, pick a standard like WCAG 2.2, get a pattern library and enforce its use. You’ll save time in the long run. BTW, the bracketed placeholder above is analogous to bad alt text. You, the user, knows something better should be there, you just can’t access it because of the way I built this page. I might come back to this and fix it.

SCORM, Tin Can — kinda related

Next cubicle says, “Hey, let’s bundle features and get an interoperability standard for learning objects, or an API, that reports test scores and bookmarks with a taxonomy of content! Then education will be like legos!” That was SCORM, which is still useful to a point.

Trying to get from SCORM 1.3 and 2004 to the next level broke their brains. It was too hard to get to the next step; which was adding in taxonomy of content. Tin Can didn’t do it either, it’s just a more robust standard as eLearning moves beyond desktop PCs to mobile. Let’s boil it down. Education/training is, writ large, at scale, every idea anyone ever had worth passing on. Oh wait, you can’t boil it down because you’re trying to boil the sea. I mean, humanity COULD do it if we had a committee big enough. Or small enough? And was Wittgenstein was joking when he wrote Tractatus Logico-Philosophicus? He was trying to document how everything one could say meaningfully and it didn’t really work for me.

Currently, you still need SMEs working with instructional designers OR experienced instructors to pick/write the content, sequence it and create the assessments. Or you buy a curriculum where someone else did all that stuff.

Can AI do this now? Maybe, some of the time, but how do you know? What’s your safety net if it goes off the rails and you don’t have the SMEs and instructional systems experts to tell you it has? Maybe hire some human experts. Or maybe you just have a folder labeled Style Guide and put old stuff in a folder called Style Guide (old, don’t use).