Ranga's Take on AI operators
Why we need to think more critically about online tasks
The game has changed, again! It seems like barely 5 minutes since I last said this. So what’s different this time?
Well, for one, I don’t think we can any longer use usage stats to influence purchasing decisions based of online tools, which is a shame, because they had potential to be really useful in preventing unnecessary spends in education.
We’ve all seen or heard claims like “86% of pupils use our tool for at least 15 minutes a day” or “I need to renew this tool because my kids use it loads” - but do they really?
Now that AI operators like Proxy from ConvergenceAI are easily accessible to all, for free, that might not be credible data any more.
With these operators, a user can provide a link (such as one provided via Google Classroom) and get the work, quiz, assignment completed by AI in the background while the user reads blogs, watches Netflix or plays Minecraft or Call of Duty.
The operators complete the tasks in reasonably realistic time scales, so the teacher can’t use that as a ‘tell’. These operators can even log in, complete and submit the work for you with no way of a teacher knowing.
That in itself is an issue, but add to that the following complication:
The tool the teacher is using to setting work applies algorithms to determine the next questions and tasks a pupil should complete based on their results. Inflated results from an operator may result in the root tool then providing pupils with further work that is beyond their ability.
It also means the teacher is being provided with data about understanding and mastery that is inaccurate, directly impacting that teacher’s planning and impacting their workload and the access and success rate of future lessons. Watch a short example using the button below.
So, we really need to think more critically about not just how we set online tasks, but also at the validity of the insights from them. Let’s not assume, because that makes an ass out of u and me.