Helen Armstrong_Designers_AI_UX Y'all 2022

Transcript

1. Why Designers Should Care About Artificial Intelligence Helen Armstrong, hsarmstr@ncsu.edu Director, Master of Graphic & Experience Design, NC State
2. ANALYZE D ATA LEARN FROM THE P AT TERN S IT D ETECTS MAKE P REDI CTI O N S Machine Learning Algorithms >  > 
3. Spam or not ???? Vivian or Tess ???? Mad or happy ????
4. Examples of Machine Learning?
5. Why is machine learning growing increasingly prevalent in society?
6. →
7. “ As of 2018, the compute was 300,000 times larger than it was in 2012. ” (available compute increasing by a factor of 10 each year) —John Seabrook Compute: Innovations in chip design, network architecture, and cloud-based resources
8. What does any of this have to do with designers? Why should we care?
9. human-centered research methods
10. Data Scientists: are trained to seek what can be accurately determined from the data at hand. Designers: trained to seek a desired future or a ‘right thing/ experience’ to design.
11. We can design interesting things with ML....
12. Personalization: Recommending and/or filtering info & experiences
13. Predictive Maintenance App In collaboration with an auto parts company, MGD students at NCSU prototyped an intelligent system to anticipate each customer’s automotive maintenance needs. Designer: Harrison Lyman
14. Rare Disease Info Hub Design researcher Rachael Paine collaborated w/ NCSU colleagues in computer science to create an intelligent information hub for caretakers of kids with rare diseases.
15. McDonald’s Intelligent Digital Menu Detect an individual and then prioritize content and suggest complementary items in response to that individual’s buying habits, time of day, local events, weather patterns, etc
16. Anticipatory Design No direct user input 
17. Portrait Blur Bubble Blur Interface Moment Flagged / B Moment Flagged / A Side Blur HearU. This wearable device prototype interfaces with cochlear implants and hearing aids via Bluetooth to adjust microphones to optimal settings thus facilitating small group conversation. Designed by North Carolina State University MGD students: Shadrick Addy, Jessye Holmgren-Sidell, Matt Lemmond, and Krithika Sathyamurthy. Persona: Andrea, an HR assistant, is profoundly deaf with two cochlear implants.
18. Do it for me Do it for myself Vs.
20. Deskilling. What do we want to keep vs what do we want to automate?
22. NeverMind This device strengthens human memory by combining memory palace memorization methods with augmented reality (AR). MIT Fluid Interface Group: Pattie Maes, Oscar Rosello, Marc Exposito Gomez.
23. How is machine learning changing how humans relate to machines.. and how does that affect what we design?
24. Natural Language Processing (NLP) Computer Vision +
25. “ Listeners and talkers cannot suppress their natural responses to speech regardless of source. ” —Clifford Nass and Scott Brave, Wired for Speech
26. In 2019, Facebook’s self-care chatbot, Woebot, was receiving 2 million conversations a week. By 2022 Amazon had sold over 120 million Alexa devices. In 2020, hundreds of thousands of people were each sending around 70 messages daily to Replika, a personal AI companion. Microsoft Xiaoice, a popular social chatbot has over 660 million users, primarily in China.
27. How might these relationships open us up to manipulation? In whose interest is the relationship being formed?
28. Beebop This ironic virtual agent, prototyped by NCSU graduate student Jessye Holmgren- Sidell, trains users to address it properly by manipulating their emotions.
29. “ Even as we treat machines as if they were almost human, we develop habits that have us treating human beings as almost-machines. ” —Sherry Turkle, Massachusetts Institute of Technology
30. “ What happens when your refrigerator is more attuned to your emotions than your [partner] is?” —Yuval Noah Harari, Hebrew University of Jerusalem
31. “ If we do not focus on creating machines with an adroit social emotional intelligence...we will end up with computers that don’t understand us—and don’t need us. ” —AI researcher Subbarao Kambhampati
32. Concept-i Design and technology studio Tellart worked with Toyota’s Advanced Design team to create the user experience for this emotionally intelligent autonomous concept car.
33. How might ML change not just what we design but how we design?
35. Debuild.com Sharif Shameem. This tool allows users to generate the code for a fully functioning app by just describing it to OpenAI’s language generator GPT-3.
39. Maybe the very idea of a tool will grow irrelevant....
40. How else will ML impact design practice?
41. “scaffolding complexity” —Patrick Hebron, Director of Machine Intelligence, Adobe
42. Florence Cathedral: 140 years Burj Kalifia: 5 years
43. “ Machine intelligence will enable creatives to do even more and think even bigger. ” —Patrick Hebron, Director of Machine Intelligence, Adobe
44. “ Working with AI is a lot less like working with another human and a lot more like working with some weird force of nature. ” —Janelle Shane, author of You Look Like a Thing and I Love You
45. David Ha, Google Brain Asked an AI to assemble some parts into a robot to move from Point A to Point B. AI combine parts into a tower that could just fall over and land on Point B. (See Janelle Shane’s TED talk) →
46. “ If we want human thinking, we should probably go to humans for it. There are a lot of them. ” —Patrick Hebron, Director of Machine Intelligence, Adobe
47. The AI has no understanding of consequences. Humans bring that understanding. —Janelle Shane, author of You Look Like a Thing and Love You
48. Critical issues around bias, privacy, manipulation, surveillance....
49. Founded Algorithmic Justice League to fight “the coded gaze”: Joy Buolamwini
50. Sarah Gold. Data Patterns This collection, cataloged by IF, a studio founded by Sarah Gold, shares a range of interaction design patterns to help creative teams address ethical issues around collecting and using people’s data.
51. A Visual Introduction to Machine Learning Stephanie Yee and Tony Chu, the founders of R2D3, use interactive design to express statistical thinking. http://www. r2d3.us/visual-intro-to-ma- chine-learning-part-1/
52. The Library of Missing Datasets Artist & researcher Mimi Onuoha created this installation to draw our attention not to the data that has been collected, but the data that is missing, thereby revealing “our hidden social biases and indifferences. ”
53. Making AI Art Responsibly: A Field Guide An illustrated zine composed of questions and case studies to help AI artists use AI techniques responsibly and with care. Created by Lia Coleman in collaboration with the Partnership on AI. “The guide is structured around questions artists should ask while making AI art. It provides some emerging best practices and checkpoints for artists to try out in their work. In particular, artists can expect to learn responsi- ble practices for using and creating datasets, machine learning codebases, training resources, and publishing their work. ” Partnership on AI Read the full guide
54. We should not trust a technology that has no true understanding of human consequences to take the lead.
55. We can use A.I. as a mechanism for equity and justice through the products/features/experiences we choose to make—or not make.
56. Your skills and vision are sorely needed in this space.