Students will be assigned to groups and 3 groups are expected to read AI/ML articles on a specific topic while considering real-world consequences (hopefully based on their interests) and prepare a 15-minute lightning presentation on that topic with a 3-5 minute Q&A. Students not presenting are expected to ask questions and actively engage in talks. We will take note of students participating in class which will impact students’ participation scores for the class.
- Student Lightning Talks 1: Amazon AI recruitment system; COVID-19 Diagnosis and Triage Models; OpenAI’s GPT-3 based Chatbot Samantha
- Student Lightning Talks 2: Google AI Diabetic Retinopathy Detection; Amazon’s Rekognition; Sentient Investment AI Hedge Fund
- Student Lightning Talks 3: Microsoft’s Tay Chatbot; IBM’s Watson
Here are some presentations tips to help students present in a clear and concise way. All members of the student group are expected to contribute equally to the presentation. The outline of group presentations is shown below:
- Topic: Describe the topic of the article, the key problem discussed, and why this problem is important to enable or improve HRI.
- Prior Work: Discuss prior work relevant to the proposed work. How is prior work different and similar to the proposed work?
- Research Gap: What gap is the paper addressing and why should the HRI field care about this research?
- Methodology: Describe the proposed method, how it works, assumptions, inputs, outputs, and study design.
- Evaluation: Discuss the evaluation procedure used in the paper including metrics, whether it is well-suited for the problem, and comparative approaches (is applicable).
- Results: Summarize the results in a manner that audiences with and with technical backgrounds understand.
- Impact for AI/ML: Use critical thinking to discuss your thoughts on how impactful the paper is compared to what is stated in the paper. In what situations might the proposed research fail? What are the ethical or societal impacts of the research?
Here are additional tips borrowed from the HRI class in MechE:
- Read the paper in full before preparing your first slide or talking point.
- First things first: summarize the main ideas in a few sentences. Your audience wants to know what they’re in for, from a bird’s’ eye view. It’s very reassuring for your listeners to know what is about to happen.
- Before going into the details, say a few words about the authors. Who are they? What is their main line of research? What’s the academic background they are writing in?
- Then build the story, along the lines of the paper. You can go on tangents, for example if the paper includes a reference to a seminal or particularly interesting other reading.
- Whenever you present an idea clearly, visual illustration is better than text on the wall.
- Where relevant, explain the research methods employed (qualitative/quantitative, experimental/descriptive, objective/subjective), including pitfalls and limitations.
- Most often, results are best shown through graphs and tables.
- Think of your personal critique. This could be of the methodology, the theory, or the conclusions of the paper.
- If you found any references to be interesting, mention them briefly at the end of the presentation.
- Tie the ideas in the paper to broader topics. How can the paper’s points be applied to your research interest?
- Think about discussion points to encourage class participation.
- Be concise, engaging, and clear.
Additional reading: Lera Boroditsky’s tips on giving research talks, in turn borrowed from Gordon H. Bower’s original tips.