Recently, I’ve raised questions about how students might demonstrate mastery in the age of ChatGPT and put forth a new vision for higher education. The first focused on the mechanics of testing and assessment, and the second on the whole undergraduate experience. The focus of this essay is on overall higher ed learning objectives in the age of ChatGPT.
What should an undergraduate student learn?
Note well: This assumes the continued existence of an “undergraduate student”. Even if this cultural role doesn’t continue to exist, we can ask an equivalent question that should spark the same debate: “What should someone learn as they prepare to enter adulthood?”
To be clear, this is a question for special committees composed of experts from across the higher education spectrum. I am but one man, but here I put forth a few thoughts.
General knowledge
No matter how competent and capable LLMs are (or ever will become), people need a certain amount of knowledge across a wide range of subjects (I’m thinking about the kind of topics in a humanities education) in order to both interpret what the LLMs says and assess its validity.
The issue of trust will not disappear any time soon. While the LLM can be a good partner, people will need to adopt a policy of “trust but verify” with much of what the LLM says until we have more experience with it.
Decision making
People will continue to need to make decisions, and also to make decisions in cooperation with an LLM assistant. The person should need to know about and how to use effective and appropriate decision-making processes.
Argumentation
Related to the above, a person should know how to construct an argument in favor or against a position held by a human assisted by an LLM. This should entail learning about the structure of valid and sound arguments as well as how to make them.
Assessment
People are going to need to be able to work with and without LLMs to assess decisions and arguments made by others. This will involve applying what is learned about decision-making and argumentation to themselves and others. It will also involve learning to assess if premises (“facts”) are true.
Information processing
Related to much of the above, people should know how to retrieve, process, and assess information in concert with an LLM and related computational tools. This will require gaining expertise in organizing, assessing, using, and distilling vast quantities of data and information so that it might be used in decision-making and argumentation.
Group work
Just as workers have been trained over the past century/millennia to work with other people in order to be more productive, tomorrow’s workers will need to be trained in how to work effectively with groups of people (with their LLM assistants) and supervisory and assisting LLMs from an organization. This is—clearly—an evolving field but should not be ignored as it will significantly affect our lives quite soon.
Conclusion
There you have it. My thoughts about higher ed learning objectives in the age of ChatGPT.
Do you agree or not? What did I miss? I would love to hear your thoughts.
