Workshop
2017 February Bots & Automation
California College of the Arts
San Francisco, CA Chris Hamamoto
Interactions with non-human actors are becoming commonplace. Increasingly we are interfacing with artificial intelligence in situations where we once would interact with other people. The line between human and machine is being blurred. From customer service chat bots to self-driving cars, we are moving towards an automated society in which human labor is being outsourced to machines. AIs are being personified as personal assistants, such as Siri and Cortana, and anthropomorphized in the Weather Cat bot Poncho—with the expectation that these sound and text interfaces will replace many of the web interfaces we use today.
As our interfaces become increasingly tied to communication platforms and ‘non-graphic’ interfaces, these developments raise questions of identity, morality, and the role of graphic design. In this workshop—with the help of Google, Siri, and Alexa—students will examine these questions by developing tactics to engage with bots and investigate automated methods to distribute their findings.
Schedule
Friday, February 10
7:00–7:30PM: Lecture
7:30–8:00PM: Identification Discussion
8:00–9:00PM: Identification Analysis Exercise
Saturday, February 11
Lunch will be provided.
1:00–1:15PM: Quick Recap of Identification Analysis
1:15–1:45PM: Review of Artist Examples & Distribution Methods
1:45–3:00PM: Individual Meetings
Sunday, February 12
Snacks will be provided.
1:00–3:00PM: Final Critiques and Presentations
Step 1: Identify
Task:
Identify the AI you interact with on a daily basis.
Examples:
Siri is both on your phone and desktop.
Your messages app intuits your conversations with autocorrect and typeaheads.
Your phone’s camera auto detects points of interest and human features.
Youtube videos generate suggestions for related content, etc.
How can you detect bots masquerading as humans?
Online Customer Services Agents
Email Auto-Responder
How can you detect humans masquerading as bots?
Google Books Scanning
Mechanical Turks
Horse E-Books
Step 2: Analyze
Task:
Begin analyzing the limitations of your selected AI.
Working in groups of 2-3, analyze your identified bots and try and answer the following questions:
Who programs the bots (the companies, the gender and race of the individuals), and what are their motivations (to sell you things, map language, etc.)?
What are the visual artifacts of your chosen technology?
Are there moments the bots stop performing as you’d expect? In interacting with your technology, when do you realize that they are bots? When do they stop feeling like bots and more like humans?
Consider your interactions with this intelligence and document (write down or with imagery) the nature of these interactions:
What type of information do you discuss?
How do you use this technology?
Identify 2-3 ‘breaking’ points where your bots begin to fail and describe how this manifests.
Bring your research to our mid-day check-in on Saturday, 1PM
Step 3: Reveal through breakpoints...
Task:
Develop a set of actions that expose the breaking points of this technology.
Examples:
Determine the minimum features necessary to create a human face that is detectable by face recognition.
Create a dictionary of words that Siri fails to recognize because of your colloquialisms or particular accent, etc.
Create a speculative Chrome plug-in.
... and/or emulation
Task:
Use algorithmic/automatic methods to emulate a non-human entity.
Examples:
Act as a "bot" and develop an action you perform routinely.
Create methods for preventing routine behavior in humans.
Develop a design algorithm, printing process, or performance that embraces automation.
