Research Foundations › Module 1: What research is
Basic, applied and development work
Lesson 2 of 3 · 2 min
Research is often split into three kinds. The split comes from the OECD Frascati Manual, the guide most governments use to count research.
| Kind | Main aim | Example in AI |
|---|---|---|
| Basic research | Understand how something works, with no use in mind yet | Why do very large neural networks generalise when they could memorise their training data? |
| Applied research | Solve a specific, practical problem | Can a small model detect crop disease from phone photos taken in poor light? |
| Experimental development | Use known knowledge to build or improve a product or process | Turning that crop-disease model into an app that works offline on cheap phones |
- Basic research
- Main aim: Understand how something works, with no use in mind yet
- Example in AI: Why do very large neural networks generalise when they could memorise their training data?
- Applied research
- Main aim: Solve a specific, practical problem
- Example in AI: Can a small model detect crop disease from phone photos taken in poor light?
- Experimental development
- Main aim: Use known knowledge to build or improve a product or process
- Example in AI: Turning that crop-disease model into an app that works offline on cheap phones
They feed each other
These kinds are not ranked. Basic research often leads to uses nobody expected. Work on how neurons in the eye respond to edges influenced early designs of image-recognition networks decades later. Applied problems also raise basic questions: when a model fails in the real world, someone has to ask why.
Which one is right for you?
For a first project, applied research is often the easiest start, because the goal is clear and you can tell when you have made progress. Basic research questions are wonderful, but they can be hard to scope.
Whatever you choose, be clear about it. A reviewer will judge a basic study by how much it explains, and an applied study by how well it solves the problem. Mixing the two aims in one vague goal is a common reason first projects drift.
RememberAsk yourself: "Am I trying to understand something, or to solve something?" Write the answer at the top of your plan.