Over the last two years, I’ve been working with Harvard SEAS faculty and cloud providers to determine how we can best facilitate students use of cloud technologies in the classroom.
Over this time I have found that there are a few principles that cloud providers (e.g. AWS and GCP) who are hoping to penetrate the higher education market would do well to adhere to.
If cloud providers follow these principles when developing their tools and services for educational use then I believe they have a much greater chance of being adopted in the classroom.
The aim of most courses that utilize the cloud is not to teach the use of a particular tool or platform, but to teach a computing concept and to demonstrate its application.
When teaching these concepts, teaching time is invaluable. If you spend 2 weeks trying to make sure everybody in a class has an account and basic knowledge of how to use your platform, then that time is lost and cannot be made up.
For this reason, onboarding for the tool should be as seamless as possible. There should be a very clear pathway for students to learn how to navigate your platform. A dedicated set of short videos aimed at students that could be assigned as a homework 0 task would be ideal.
Consider integrations with common LMS platforms to simplify adding students to resources and easily enable SSO.
Provide an administrative interface from which teaching staff can monitor students’ spend.
Every semester we see students using advanced, CPU and memory intensive operations earlier and earlier in their academic journey.
Many entry level tasks these days involve GPUs which cost upwards of 1$ per hour. It is not unreasonable to expect a student to use 200 - 300 hours of GPU time over a 4 month course. So, credit limits should be flexible.
It should be very easy for teaching staff, or even students themselves, to extend their credit limit if necessary.
I personally believe that asking students to sign-up with a credit card is potentially discriminatory.
In the past, we have seen students overspend on their accounts (for example, by leaving a GPU running, unused). This has led to many hours of teaching staff negotiating on their behalf to have bills cancelled. Making it as easy as possible for students and teaching staff to monitor student spend would help here.
Tools should also enable limiting the resources that a student can use, but these limits should be flexible or customizable.