Improved user experience for Display AI-assisted text categorization
backlog
T
Timothy George
Displayr AI-assisted text categorization is a leap forward, but I think the user experience can be improved significantly to avoid situations like the following:
- A user uses Displayr AI to code a large number of open ends (e.g., > 1,000), which Displayr says "may take a few minutes." The user's held captive for a while thinking it'll be done in a few minutes like Displayr says, but it ends up taking more than an hour.
- A user uses Displayr AI to code a small number of open ends (e.g., < 100), which Displayr says "may take a few minutes." The user steps away for 15 minutes thinking it won't be done for a few minutes like Displayr says, but it's done in 30 seconds.
It's all about setting realistic expectations, so users can act and plan accordingly. Telling all users that Displayr AI-assisted text categorization "may take a few minutes" isn't acceptable. I know it's difficult and involves a lot of factors outside of our (Displayr's) control, but we need to give users a better time estimate based on the volume of text data they’re trying to code w/ Displayr AI. Displayr's all about saving time, and Displayr AI-assisted text categorization, although very powerful, currently puts users in bind and can cause them to waste a lot of time. In short, saying it "may take a few minutes" is way too vague.
J
Jamie Morris
backlog
S
Simon McManis
A change is in the works that will significantly increase the speed of AI-assisted text categorization (roughly 25x faster).
It is possible to give an estimate on the time taken to do the categorization based on the number of responses. We have the calculations ready.
T
Timothy George
Simon McManis Awesome to hear!