Reflections on defending my thesis in the age of LLMs

The landscape of academic research is shifting rapidly. As I stood before the committee, I couldn't help but wonder how generative AI will reshape the very definition of original contribution in computer science doctorates.

The New Standard

It used to be that writing the code was 80% of the battle. Now, with Copilot and GPT-4, the implementation details are becoming commoditized. The real value is shifting towards:

1. Problem Formulation: Asking the right questions.
2. System Architecture: Designing robust interactions between components.
3. Evaluation: Rigorous testing that goes beyond standard benchmarks.

What this means for future PhDs

I believe we are entering an era where "solo" research will become much more ambitious. A single student can now build systems that previously required a team of engineers. This raises the bar for what constitutes a "contribution".

"The tool is only as good as the hand that wields it."

We must ensure that we are training students not just to code, but to think computationally and critically about the systems they build.