Future of CS Education
Recently went for a workshop on this topic. As a CS undergrad myself, I've been wondering how I'd future-proof myself in the age of AI. Thought I'd share some points I found interersting. Again, they're in point-form - because that's the way I recorded them lol
- Go beyond code syntax. Writing code is cheap, it’s all about designing systems now - how to make it modular, what code should exist at all, etc..
- Proof of Skill > Proof of Attendance…. companies are looking at open-source contributions, project work, etc.. over just GPA/grades
- Specialists & Generalists both need foundational skills (i.e. first principles)
- The future of jobs lie in the ability to learn & learn fast… it’s about having strong adaptability & foundations
- ChulaGenie - genAI application used at Chulalongkorn university
- A chatbot assistant alongside the programming assignment… Not free-form asking (i.e. you can’t just ask the chatbot whatever you want), instead there’s a predefined set of questions that can be asked (e.g. “What’s causing the wrong output?”). Essentially, the AI only gives hints, not the actual answer. Yet, it’s still dynamic because the AI is able to see the most updated code structure, and give its response
- An assignment given at Chulalongkorn: Students are given a piece of reference code, and they have to prompt the AI to produce similar code in as little prompts as possible. This assesses whether students can appropriately describe the task to AI, which itself involves understanding the programming structure (e.g. what are loops, if-else, etc…)
- AI governance must be agile & iterative
- Industry exposure is critical
Most important courses to take in the age of AI?
- Essential Courses: Learning how to learn, learning how to ask, learning how to unlearn
- Macro-level (first principles) courses: What is intelligence? What is computation? What is the philosophy of CS?
- Give students the big picture, then have them go deeper into the ones that interest them…
- Knowing the history behind computation gives students a sense of time, especially in the age of AI where everything is just so fast
- Computing verticals (e.g. databases, computer networks, etc.) shouldn’t be mandatory (since AI will eventually be able to do them all), but passion-driven (i.e. you only pursue them if you’re interested in it)
- Explore more hands-on experiences independently, rather than studying everything in a sandbox
- Crucial CS skills: Tech fundamentals + Business Awareness
- Evolving CS specializations: Human-AI interaction (workflows that give trust, etc..)
- How will assessments change? Shift toward oral defense & code review to assess understanding?
- Shift in focus of education: Move from “how to write code directly” (i.e. syntax) toward “how to write code correctly.” (i.e. verification & debugging)
- Emphasis on supervising. Training students to be “supervisors” rather than just executors
- Can use AI as a: (a) shortcut (for certain tasks where we already know the outcomes), and (b) collaborator (for tasks with uncertain outcomes)
- Things that AI can’t have: The conviction (the why?) and taste (of the chosen problem/solution)