03.09.2018| Phase 3 | Generative Research
During this week, we finished our generative research and present our research finds, insights, and design concepts in the class. Arnold also gave us very useful feedback during our presentation.
At the beginning of the week, we conducted several speed dating with our defined concepts.
The process and goal of Speed Dating are as followings.
- Pitch each of the three possible design concepts to potential users
- Have potential users to identify shortcomings and merits of the design concepts within a short period of time
- Feedback incorporated back into design concepts
- Smart recommendations that connect international students with American students
5 initial design concepts
We mainly came up with 5 design concepts based on our previous generative workshops, here are our design concept descriptions as well as feedbacks from our speeddating.
1.Smart recommendations that connect international students with American students:
Imagine a social network that connects incoming international students with American students on the same campus based on similar interests, to allow for an exchange of cultural information between locals and immigrants. The system would act as a facilitator to help make difficult conversations easier and promote mutual learning. It would also make recommendations for other people to connect with. The system would “listen” and learn from the conversations and build on that data to train and help other users outside the conversation (or during).
Feedback: less likely to connect with someone we dont’ know, even if there is mutual interests. Really cool, meeting people based on interests was nice, and also to meet ppl from outside class. Good to connect with ppl outside major, too. More helpful once you’re here, not before arrival. Effective, but worried about privacy issues (that it will listen to your convos — creepy and invasive). Nice to find common ground with ppl, and also give suggestions for things to talk about. But how does it work, how does it learn from our conversation? Does it replace conversation with a chatbot then?
Suggestions: maybe machine can know how narrowly or broadly to search. Maybe bot facilitates doesn’t replace. And gives suggestions on how to respond to people.
2. AI Digital Assistant
Descriptions: Imagine an app on your phone or a wearable that recognizes your location and prompts relevant information based on your location. So, for example, if you are in a restaurant, it will send you a notification that you need to tip the waiter, and it will help you calculate the amount to tip. Like Siri, you can also ask it specific questions related to cultural norms and it will advise you on how to react or respond.
Feedback: Least helpful. It takes opportunity away from user to engage in a conversation, and for really important information, they’d rely on a organization / agency. Area of intervention should be thought through. User should control notification frequency and choose when. Maybe it can help complete more complex tasks like in a hospital or filing taxes — situations where you’d be less likely to talk to a person b/c of embarrassment or privacy concerns.Maybe it can identify false assumptions, and AI can learn and adapt based on that.
3. Trivia Game
Descriptions: Imagine a smartphone game like Google Trivia that quizzes you on certain cultural norms (such as how to pay for the bus or how much to tip) to help you learn and get acclimated to American culture. Using machine learning, it will remember which questions you answer correctly so that you are constantly learning new things and not repeating old topics.
Feedback: Game should be very engaging and interesting. I don’t want to play with a machine; what if i can play with friends? Need to show how knowledge improves / changes behavior. We need motivation for playing the game. She doesn’t need to know everything about being an American; she still identifies as ‘indian.’ social norms are tricky. She doesn’t want to be quizzed about it and worry about being wrong.
4. Simulating scenarios in advance
Descriptions: Imagine an AI assistant that helps you practice conversations in advance. You are giving your first presentation in class tomorrow and not feeling entirely confident, instead of worrying about it, you can practice your delivery with the AI assistant. When you feel stuck, tell the assistant what you want to say in your own language and it will translate to help you fill the missing gaps.
Learning Motivation: Avoiding embarrassment, better delivery/grades, richer conversation
Learning advantage: (observe performance — feedback — direct practice)
Feedback: It should only be used for complex scenarios/tasks, but also should be general and not too specific. It’s practical but more relevant for specific scenarios, so need to figure out which scenarios are most important. Could be really helpful for conversations. What else can it do? Make suggestions for how to do something better?
5. Conversation “wing man”
Descriptions: Imagine an AI assistant is constantly listening to the conversation, and become a facilitator of the conversation, which can recognize when the user has difficulty in understanding others or conveying his/her ideas, then prompt and help the user in that moment.
Feedback: Intrusive privacy concerns, too much power, happy with a person but not a machine doing the help. Helpful, but creepy when it listens.
After speed dating, we ranked our design concept with the feedback from our participants of speed dating. We finalised three possible design directions that we would like to focus on.
