Research Topics
In Research Areas, you narrowed your interest to one or two AI research areas. In this section, you will go one level deeper into those areas to commit to a specific topic for the semester, with a clear understanding of why it is worth pursuing now.
A topic here is a theme or subfield, which is narrower than an area but not yet a concrete, executable research problem. That final step, choosing a specific task, comes later in Task Selection, once you've formed a team.
Topic Discovery
Finding a topic that is motivating, active, novel, supported, and feasible takes a deliberate process. Work through the steps below in order, each with a short exercise, so that by the end you have a specific research topic to commit to for the semester.
1. Personal Motivation
Start from the area(s) you narrowed down, and identify a topic within them that excites you. Ownership matters: a topic you pick because it feels alive to you is far more likely to sustain a full semester of motivation than one chosen purely because it looked promising on paper.
Exercise (5 min): Propose a specific AI research area and a topic within it that excites you. Write 2-3 sentences justifying your choice: why this topic, why now, and what about it excites you.
2. Landscape Analysis
Now validate that pick: survey your area's traditional (well-established, studied for years) vs. emerging (newer, gaining momentum recently) topics to check whether the topic you are drawn to is active in the field, and to see what else is out there in case it is not. Useful signals are:
- Keynote and invited talks at recent venues — what are established researchers highlighting as important right now?
- Awarded and highly-cited papers — traditional topics tend to have a long citation history; emerging topics tend to have a recent, fast-growing citation count.
- Workshops and tutorials — new focused venues can be a strong signal of an emerging topic gaining traction.
- Paper volume in recent years — is this topic growing, stable, or shrinking?
Exercise (10 min): Using the conferences listed for your chosen area(s), check whether your step 1 topic reads as traditional or emerging, and identify one alternative topic in the other category as a backup. Justify both with the signals above in 2-3 sentences each.
3. Novelty Check
Verify there is still room to say something new in your chosen topic. An emerging topic can still be crowded, while a traditional topic can still have open sub-problems.
Exercise (5 min): Skim 2-3 very recent papers in your chosen topic. Note challenges or limitations they are still struggling with, or explicitly list as future work.
4. Support Check
Look for a faculty member, lab, or active line of work connected to your topic. You do not need a formal advisor to start, but knowing that expertise or resources exist nearby will make it much easier to get help when you hit a wall.
Exercise (5 min): Name 1-2 faculty members from AI Faculty whose work connects to your topic, and note briefly how.
5. Feasibility Check
Before you commit, do a final reality check: is there data available (or creatable) for this topic? Is the compute realistic for a semester project? Can this be meaningfully scoped down to fit the time you have? If the answer to any of these is no, revisit your shortlist rather than forcing the topic.
Exercise (10 min): Write 1-2 sentences on each check (e.g., data, compute, scope). Does it sound feasible for a semester project? Too broad? Too narrow?