We have reached a point in the AI evolution where we're being told this technology is going to transform education, revolutionize creativity, and fundamentally change how humans work.
And yet apparently we're struggling with the extremely advanced task of teaching a five-year-old the alphabet without asking a chatbot to do it for us.
And then, somehow, we still manage to get the alphabet wrong.
The instinctive reaction to this is to ridicule AI technology with all it's many flaws or errors, instead of looking squarely at the individuals making these judgement calls.
Two news stories caught my attention recently. The first was a Colorado school district that purchased the 'Bluebonnet Curriculum' from Texas. Parents immediately pushed back on certain content and adjustments were initially made, but the entire program was ultimately pulled just 3 days into the school year because the material was riddled with errors and irrelevant information. Many of the errors were suspected of being produced by AI.
Now I have thoughts about using AI to write education programs, but that aside, where is the human judgment in all this? Why was this material not thoroughly vetted before launch into ANY school but also, what was Colorado thinking?
The second story I came across involves an influx of AI slop into the Teachers Pay Teachers marketplace. If you are not familiar, TPT is a platform for teachers to sell the resources they design for their classrooms in order to make a little money that they can then use to supplement their meager teacher salaries or, as is more probably the case, buy additional resources for their classrooms.
Unfortunately the site isn't restricted to only teachers, anyone can sell here, and many unqualified people do. Hence the influx of AI slop, I'm assuming. The news segment featured alphabet poster designs that were quite remarkably wrong. Not a little wrong... a lot wrong.
Which begs the question... wtf is wrong with people?
Someone generated them, looked at them, decided they were good enough, uploaded them, and put a price on them. And clearly, does.not.care.
I am not anti-AI. I think it is a remarkable technology that could be used to our advantage. Instead, it's being squandered in the worst ways possible.
A responsible classroom creator might use AI to brainstorm lesson ideas, generate variations, create practice questions, suggest activities, help structure material or overcome creative block.
The 'responsible' part involves that person following up with checking to see if the work produced is actually correct. Is it appropriate for the age in question, does it teach what you think it teaches, are the examples accurate, does the activity make pedagogical sense, are there weird artifacts or... did the AI hallucinate something?
And do you know how you do that? You use your own brain.
And this where I think most anti-AI feelings are likely triggered. The sheer stupidity of the output.
But it's not AI's fault for that. We have this bizarre expectation that AI is going to be some sort of all-knowing oracle, when in truth, it only knows what WE feed it. And we mostly feed it a junk food diet.
Instead of asking “How could AI do this?” Ask: “How did the humans responsible for this not catch it?”
That's the better question. Because what we are looking at here is purely a quality control issue.
The fact is AI has dramatically lowered the cost and effort required to produce stuff.
Which is terrible if the model is: Generate → minimally inspect → upload → repeat.
The economics reward volume and if people don't care about the added value of quality control they are not going to insert into the production process. Because they don't have to. There is currently no consequence for skipping it in many industry sectors.
The truth is, AI didn't invent the desire to make money with minimal effort. It simply made the minimum effort required much, much smaller.
Basically, we invented machines capable of producing enormous amounts of information, and then decided the appropriate response was to stop checking whether any of it was good. Why? Because doing so has a cost, and many people are choosing to forgo that cost.
Before AI, producing something required effort. That effort acted as a natural filter.
If creating a 200-page curriculum required months of research, writing, editing, illustration, layout, fact-checking and review, you couldn't casually decide: “I'll make one this afternoon.”
AI removes enormous amounts of the friction involved in production. That's wonderful when the friction was unnecessary. But some friction exists because quality takes work.
The danger here is that we've removed the friction without replacing it with a quality-control process.
So instead of: Create → review → revise → publish. We get: Generate → publish.
And that is where the garbage comes from.
Bottom line... AI shouldn't take the blame for human laziness.
If someone uses AI to produce something inaccurate, the person is responsible. If a seller knowingly floods a marketplace with junk, the seller is responsible. If an administrator approves an unvetted curriculum, the administrators are responsible. If a publisher releases material full of errors, the publisher is responsible.
AI isn't a moral agent. It doesn't have greed. It doesn't have a quarterly sales target. It doesn't decide that fact-checking takes too long.
Humans do.
The solution isn't to throw away the technology. It's to stop throwing away our judgment.
We need to start holding people accountable for their decisions and their judgement.
The most frightening future isn't the one where AI becomes smarter than humans. (Skynet) It's the one where humans become so accustomed to letting AI do the thinking that they just stop noticing when the thinking is wrong. (Idiocracy)
0 comments