Response to Data Science & AI Level 3 Qualification discussion paper

The RaspberryPi blog post about it is here.

The discussion article itself is here.

Below is the response I sent to RaspberryPi.  I've learned so much from previous discussions on the CAS site - I would love to hear other perspectives, especially if you think I've got this all wrong!

 

I am an experienced Computer Science teacher in an academically selective high-achieving independent school.  I have spent some months exploring how to bridge the gap between activites like Micro:bit CreateAI and the UK Olympiad in AI.  Coming from a software engineering background, this has been the most difficult thing I have tried to learn in many years.

I wholeheartedly support the suggestion that Computer Science A-Level cannot accommodate this subject.  Even the language required to understand AI/ML is alien, let alone the deeper concepts.

In the long term, the only way we will do justice to this field is by having a fully regulated A-Level in it.  

The breadth of concepts required for a solid foundation could not be met by an individual student's independent research.  Even with Computer Science A-Level, it can be a struggle to get a student who is experienced in programming to break their pre-learned moulds in order to properly appreciate the deeper concepts.  Getting more academically-inclined students to develop their engineering skills is a similar struggle.  Leaving students to direct their own research into AI risks them avoiding the necessary discomfort/struggle of learning outside their previous experience, and thus not developing that solid foundation.

Rushing in to specifying an A-Level before a proper pedagogy has been explored by teachers, students and subject experts could leave us stuck with any mistakes/misjudgements for many years to come.  The discussion paper seems to suggest that an EPQ-style qualification would be an interim stepping-stone towards specifying an A-Level.  I whole-heartedly agree with this approach.  I still fear that once an EPQ-style qualification becomes more settled, there will be inertia and/or active resistance to moving to (or adding) a more rigorously structured A-Level.

Your 'cohort B' does not make much sense to me, but I suspect that might be due to the nature of my school.  Medicine at university is so competitive, there is essentially no room in a student's week for anything that will not directly help them obtain a university place.  Your examples of physicists and biologists makes more sense, but it would be competing for an option slot against Further Maths.  As so many university entrance exams seem to be maths focused, I suspect a new A-Level might struggle there too.  I am sure the national cohort will be very different to that at my one school but there is already a struggle convincing students that A-Level Computer Science is valuable when not even university Computer Science courses require it!

In terms of delivery of the course, I think lessons can be taken from the NCCE's successes.  A central scheme-of-work with high quality resources will help teachers self-train ahead of teaching the students.  

Students will ask questions the teachers are not able to answer, so it will be important to pro-actively support those teachers in learning those answers and explaining them effectively.

Providing high-quality CPD courses (preferably face-to-face) before the course starts could also help overcome confidence barriers for both teachers and school managers.

By its very nature, responding to an article like this risks sounding depressing and defeatist.  So I want to finish by saying that I find it really uplifting and encouraging that this new field is being explored and discussed in such a thorough, detailed manner.  It is clear to me, even in the midst of the current investment bubble, that AI/ML is a world-changing tool.  We have a huge opportunity to enable young people to understand and use it effectively and responsibly.  Thank you for leading that discussion.