All thoughts my own and constructed from what I’m reading.
I’ve struggled writing this post a few times, because of two reasons. The first is it’s just so negative I was worried about sharing it, and the second is I didn’t feel brave enough in the vision I saw to share it. That’s changed now, my wife told me if I see something coming, I should share it and let the other people make their decisions, and also after chatting with a good colleague today, he’s helped vindicate my thinking. You never, know, I could be wrong.
So here goes. Two reasons why the instructional design job will be dramatically changed in 1 year, and mostly gone in 2 years.
- With a series of prompts, and an API connection to your VLE, we can output a full course of poor to decent quality in an hour.
Doesn’t sound massive, but it is. With just writing text, now, it’s possible to output a course in a matter of an hour. Just text, nothing more. Spend a few more hours refining those prompts, and we have a course of decent quality.
The outlay is 5 hours, not 10 days. it’s 20$ dollars of API spend and not hundreds on day rates.
- AI isn’t going to get worse, it’s getting better exponentially
“Oh but, it doesn’t do constructive alignment, oh but your prompts don’t include assessment, oh but it can’t do XYZ”… yes you’re right, but only very temporarily.
These prompts nearly do these things already in chatGPT3.5 and now AI is industry led, not academic led, it’s only going to get smarter and faster.
To conclude
The instructional designer’s job is very exposed to Large Language models, generative AI, knowledge curation, html activity creation, video and presentation creation and using rubrics to check and improve work.
Instructional design is going to change massively in the next year. I can’t say it won’t, because I don’t believe that.
I know a lot of people will attempt to soften the news with, “AI won’t replace you, people using AI will”, however that’s not true, I believe AI will replace many instructional design jobs.
So what can we do about it? Frankly, after you’ve gone through the grief cycle of the role loss, you’re welcome to do like me and upskill in AI/ML, but with AI/ML without coding :). , or the newly release raspberry PI and Deepmind collaboration. Do ping me on Linkedin if you want to chat.
And yes, this might be seen as a negative post, but, personally for my learning journey I needed to accept the fate to move on to conquering it.
Potential prompts
Here is a set of prompts to chain to start you off. Nothing here is new, it’s just one step in front of the other like any AutoGPT tool would go. It’s a v1. All v2 costs is a little more writing, and the course quality jumps up.
- You are an instructional designer. Based on the following course level ILOs, please create a rough course outline and rough weekly ILOs for 6 weeks of learning for adult business professionals. The course is remote only. [ILOs] By the end of the course the students should have achieved the course level ILOs
- Taking that result, now rate each week’s ILOs to check they are measurable from 1 to 3. Any which score 1, list the potential improvements to make it a 2
- Taking that result, now re-write the weekly ILOs which scored 1 to make them a 2 or 3
- Output the structure, prompt user to continue or not.
- Take that result and now, list out weekly activities which add up to 6 hours study time a week, and include 1 hour of introductory reading time. The activities should follow Dr philippha hardman’s DOMS, and “every week participants learn by solving a problem (rather than consuming content) is as effective online as it is in the flesh.”. The first activity should always be a light opener for students to self assess their early knowledge, the main activities should push towards students learning the skill/background briefly, but move quickly to working and applying the knowledge to case studies and the group problem solving through discussion and application. The final activity in each week should be a student applying it to a work case study they bring, and the tutor offering feedback.
- take that result, anywhere a video would make sense, please use [VIDEO SERVICE API] to generate
- On every other page, generate a nice picture which frames the content [chat gpt4]
- Take that result, now score each week against DOMs by Dr Phillipa Hardman’s approach and if the week “a session where participants learn by solving a problem (rather than consuming content)” out of 1 to 3 and list areas for improvement.
- Take that result, any areas with a score of 1, please action the items and re-output
- Take that result, and score each area against David Hopkin’s rubric, with a 1 to 3. https://www.dontwasteyourtime.co.uk/ai/learning-design-checklist/. anything with a 1, list one change to make it a 2
- Take that result, and action anything with a 1.
- Take that result, now and write it to the VLE using [YOUR VLE API]
Also, pretty clearly, this post was not written with the aid of AI 🙂