[PPT] - Designing People+AI Systems Human-AI Interaction Luigi De Russis PowerPoint Presentation

SLIDE 1

Designing People+AI Systems

Human-AI Interaction Luigi De Russis

Academic Year 2019/2020

SLIDE 2

2

AI: Risks, Benefits, and User Tolerance

Human-AI Interaction

SLIDE 3

Human-AI Interaction Fall 19 .

Uncertainty & Unpredictability for Users

3

Relinquishing control to an AI/ML agent can be helpful, but can be much harder to correct or understand

if things go wrong

“Unpredictability” can joyful in one kind of experience, and a terrible idea in another

SLIDE 4

Human-AI Interaction Fall 19 .

Risk: Severe Failure

4

SLIDE 5

Human-AI Interaction Fall 19 .

What was the error?: Severe Failure

5

Tay’s earlier version XiaoIce ran on China’s most widespread instant

messaging app Wechat … without any major ethical incidents

What makes Twitter a different environment?
Tay had no moral agency. To her, words like Hitler or Holocaust are

not different from words like chair or Oklahoma

SLIDE 6

Human-AI Interaction Fall 19 .

Mitigating: Severe Failure

6

2017 Tay used some black-listing of ‘bad words’ but could make no moral judgements. 2018 Zoe uses both black-listing of ‘bad words’ and makes moral judgements.

SLIDE 7

Human-AI Interaction Fall 19 .

Mitigating: Severe Failure

“It’s easier to program trigger-blindness than teach a bot how to recognize nuance. But the line between casual use (“We’re all Jews here”) and anti-Semitism (“They’re all Jews here”) can be difficult even for humans to parse.” … “Zo’s uncompromising approach to a whole cast of topics represents a troubling trend in AI: censorship without context” - Chloe Rose

Stuart-Ulin, Quartz

7

SLIDE 8

Human-AI Interaction Fall 19 .

Uncertainty & Unpredictability for companies & designers

8

It can be very hard (sometimes today impossible) to predict all the kinds of scenarios your system could

wind up in

An unknown-unknown is a situation out there in the world that your system won’t handle correctly but

won’t know it’s wrong

Model performance will likely change/degrade as time passes
Because users change how they interact with the system
Because it sees new data that looks less like it’s training data

SLIDE 9

Human-AI Interaction Fall 19 .

What are some everyday errors we can expect?

9

SLIDE 10

Human-AI Interaction Fall 19 .

ML/AI error: Poor model performance

Usually solvable by acquiring more training data for the situations the model is weakest at
Data is expensive to collect, and your company or organization has limited resources. Prioritizing what

specific data to collect is essential

Designers can use rule or non-ML based fallbacks to still deliver the user some value when model

performance isn’t good enough for some cases

10

SLIDE 11

Human-AI Interaction Fall 19 .

ML/AI error: Low confidence or false High confidence in a prediction

Low confidence predictions can mean that the model has lower performance, or the phenomena itself is

just… less predictable

Communicating with the user or providing good non-AI/ML fallbacks is key
High confidence (when the model is really wrong) is worse
Unkown unkown errors
Need to give the user some error correction or feedback method to deal when this happens

11

SLIDE 12

Human-AI Interaction Fall 19 .

ML/AI error: Relevance errors

Airbnb suggesting ‘fun local activities’ when you’re traveling for a funeral
Exercise app suggesting ‘time to get up and walk!’ when you’re seated on a long flight
Amazon suggesting products that you are allergic to or can’t eat

12

SLIDE 13

Human-AI Interaction Fall 19 .

ML/AI error: Multiple users and kinds of input that look the same to the system

1. Use Spotify to play 1970s pop jams at your Mum’s party 2. Use Spotify to play your favorite study jams 3. Use Spotify to hate-listen to Ariana Grande (sorry) with your roommate 4. Your roommate also controls the same Spotify account to play their favorite study jams

What music should Spotify recommend this account play?

13

SLIDE 14

14

User: high stakes

AI causes active harm (e.g. recidivism prediction or hiring prediction)
AI reveals information someone wanted kept private
AI shows offensive content

User: low stakes

AI/ML feature is annoying or interrupting
AI/ML feature is often wrong
AI/ML feature is useless

Product/Service organization

Users stop using your app/service because of poor AI/ML performance
Bad press or legal troubles
Bad reviews discouraging others from using the app/service

To recap: what are the stakes for AI failure?

Human-AI Interaction

SLIDE 15

15

To recap: what AI limitations may look like

During Development

Insufficient data
Poor model accuracy
ML task is more expensive than it’s

worth

User scenarios not sufficiently mapped

What to do § Start with scenarios, involving different stakeholders (no tech push) § Start with de-risking § Purposefully design (quantitative) metrics to match scenarios and user studies Deployed in the wild

ML features fails silently
ML feature reduces or does not add

engagement What to do

Design performance metrics for

deployment around engagement, use, accuracy

Collect (private and appropriate) user

data

If possible, do a field study with a special

group of beta users

Human-AI Interaction

SLIDE 16

16

§ What is the role of an AI feature? § Should it be:

Critical or Complimentary
if a system can still work without the feature that AI enables, AI is complementary
Proactive or Reactive
Proactive feature: it provides results without people requesting it to do so
Reactive feature: it provides results when people ask for them or when they take

certain actions

Visible or Invisible
Dynamic or Static
how features evolve over time

(source: https://developer.apple.com/design/human-interface-guidelines/machine-learning/overview/roles/)

User Tolerance To AI Failures

Human-AI Interaction

SLIDE 17

17

§ In general, the more critical an app feature is, the more people need accurate and reliable results § On the other hand, if a complementary feature delivers results that are not always of the highest quality, people may be more forgiving

User Tolerance: Critical or Complimentary

Human-AI Interaction

SLIDE 18

18

§ Proactive features can prompt new tasks and interactions by providing unexpected, sometimes serendipitous results § Reactive features typically help people as they perform their current task § Because people do not ask for the results that a proactive feature provides, they may have less tolerance for low-quality information

such features have more potential to be annoying

User Tolerance: Proactive or Reactive

Human-AI Interaction

SLIDE 19

19

§ Proactive features can be helpful

in small amounts
at the "right" moment
if they are easy to dismiss

User Tolerance: Proactive or Reactive

Human-AI Interaction

SLIDE 20

20

§ People's impression of the reliability of results can differ depending on whether a feature is visible or invisible § With a visible feature, people form an opinion about the feature's reliability as they choose from among its results § It is harder for an invisible feature to communicate its reliability - and potentially receive feedback - because people may not be aware of the feature at all

User Tolerance: Visible or Invisible

Human-AI Interaction

SLIDE 21

21

§ In addition to the frequency of a system updates, static or dynamic improvements affect other parts of the user experience § For example, dynamic features often incorporate forms of calibration and feedback (either implicit or explicit), whereas static features may not

User Tolerance: Dynamic or Static

Human-AI Interaction

SLIDE 22

22

§ Do not overuse feedback requests or users will get annoyed

People won't like to feel like the AI is so stupid that it needs their help

§ Save for high stakes failure, is possible

User Tolerance To Give Feedback

Human-AI Interaction

SLIDE 23

23

§ How should an AI system best react to failure not to lose the user's trust?

or not to be turned off/abandoned

§ Which roles AI features should have? And when? § Examples:

YouTube Recommendation are Visible + Complementary + Proactive
Smart home features could be Invisible + Critical
some of them could be reactive, other proactive

User Tolerance To AI Failures

Human-AI Interaction

SLIDE 24

24 Human-AI Interaction

source: https://pair.withgoogle.com/worksheet/user-needs.pdf

"To AI or not to AI?"

SLIDE 25

25

Choosing the People+AI Path

Mitigating Risks, Increasing Tolerance, Highlighting Benefits

Human-AI Interaction

SLIDE 26

26

Guidelines

Human-AI Interaction

§ By Microsoft Research

https://www.microsoft.c
m/en-

us/research/project/guid elines-for-human-ai- interaction/ § Saleema Amershi et al. Guidelines for Human-AI

Interaction. ACM CHI 2019
https://doi.org/10.1145/32

90605.3300233

SLIDE 27

27

§ Slides with the "Human-AI Interaction Fall 19" banner are taken from the Human-AI Interaction class at Carnegie Mellon University

http://www.humanaiclass.org

§ All the other sources are reported when they first occurred

Sources

Human-AI Interaction

SLIDE 28

28

§ These slides are distributed under a Creative Commons license “Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)” § You are free to:

Share — copy and redistribute the material in any medium or format
Adapt — remix, transform, and build upon the material
The licensor cannot revoke these freedoms as long as you follow the license terms.

§ Under the following terms:

Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were
made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses

you or your use.

NonCommercial — You may not use the material for commercial purposes.
ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions

under the same license as the original.

No additional restrictions — You may not apply legal terms or technological measures that legally restrict
thers from doing anything the license permits.

§ https://creativecommons.org/licenses/by-nc-sa/4.0/

License

Human-AI Interaction