Sometimes, my students reply to my emails with messages that appear AI-generated. This has always annoyed me, but only lately I’ve been able to articulate why. Here are the reasons why using AI to do your correspondence with your supervisor is not a good idea. So, these are the reasons why to avoid using AI when emailing your supervisor. It…
[Edit: added a seventh problem based on Elaine Zanutto’s comment on LinkedIn.][Edit: added a sixth problem based on Chris Chapman’s comment on LinkedIn.] Five problems in synthetic users created using LLMs: Takeaway: Synthetic users are inauthentic users. They could be useful for evaluating systems in an exploratory and limited way when the target user persona is extremely well-defined (e.g., based…
Two very strong trends that will surely impact how qualitative research is done: 1. Automated interview pipelines: AI handles the customer/user interviews through chat or telephone automatically. 2. Automated thematic analysis: AI either applies a pre-existing taxonomy to qualitative text containing people’s feedback or other experiences, or creates one using inductive logic and then applies it to data coding and…
(A message to my students, sharing publicly because it can help others understand what commenting actually is about …. it’s often done wrong.) I realized I didn’t instruct you properly on how to “edit and comment” papers. I want to fix my mistake. When we talk about “edit and comment”, we don’t mean proof-reading. Proof-reading is the least important part.…
ACM and other publishers are increasingly requiring authors to state in what ways (if any) they used AI in the research and writing process. Here’s one version based on Elsevier’s template that you can use (modify where needed): Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work, the author(s) used Open AI’s…
Let’s start from evaluation. When have a computer that does something, we want to see if it does it well. So, this (“seeing if it does well”) is called evaluation. Evaluation is at the core of pretty much every computational technique, such as algorithm, model, system, device, etc. We want these things to work and improve people’s lives. And that…
A major problem in user studies are high SDs (=standard deviation). People generally experience and perceive things differently. This imposes problems for statistical analysis of user study data, because methods are often based on means, e.g., means comparison (t-test, ANOVA). Consider two cases: participant A gives on a five-point likert scale the rating 5 to a system (maximum), participant B…
MIRH: Motivate, Inspire, Relay personal experiences, Hold accountable >> four advantages human educators have in comparison to AI MOTIVATE: People still rely on other people to be motivated. Except if a person has strong intrinsic motivation, then they don’t need to be motivated. But many students need to be motivated and the generic “Hello, I am your friendly AI assistant…
Been playing chess recently. Chess is oddly similar to research. Thinking and writing require steps similar to moving pieces on board, with eventually the paper you’re writing being the board. Concepts and sentences are the pieces. You need to move them into the correct order (position) on the board. And there is a similar goal of making the opponent (reviewer)…
1. Google Ads is nowadays NOT much about selecting keywords >> Google Ads is no longer about keyword selection >> Google Ads is most typically selecting the keywords on your behalf, especially in Performance Max campaigns >> there can be some campaigns (e.g., brand) in which you’d still want to manually select keywords, but this practice is no longer the…