CHI2024 Workshop: Personal AI Assistant Speculations

Important dates:
February 22, 2024 – deadline for submission of position paper to
March 10, 2024 – notification of acceptance or waitlist status

We are encouraging three types of position paper submissions:

  1. Design Fiction – can take the form of a fictional paper about a fictional technology or fictional study, a mock-up or prototype of a fictional product, or the creation of a critical design of some kind. We are not making any intentional distinctions between design fiction, critical design, and speculative design; all submissions that represent ideas that are not meant to be actual proposals for products but, instead, are meant to make us think about our interactions with technology will be considered under this submission type. Preference will be given to submissions that are related to the workshop schemas, proposed questions, or relevant themes.
  2. Science Fiction Analyses – should analyze a personal intelligent assistant from any source of science fiction (broadly interpreted) and its role in the relationships of the characters it impacts. Examples include, but are not limited to: discussing how Jarvis in Iron Man mediates Tony Stark’s relationships with other characters; exploring the relationship between Gibson’s AI systems in Neuromancer or Count Zero and today’s AI systems; describing the impact of the surveillance possible through the Internet-like system in Benjamin Rosenbaum’s The Unraveling on Fift’s 9 parents; and speculating about the new interdependencies opened up by the many automated algorithms described in Ruthanna Emrys’ A Half-Built Garden.
  3. Case Studies – should describe projects about personal AI assistants that authors are working on, hoping to start, or for which they are seeking collaborators. These can include projects they are hoping to build or study themselves, as well as “projects” that involve a critical analysis of a company’s proposed AI relationship imaginary. We will prioritize case studies that focus on the impact of PAIAs on relationships or that otherwise align with the workshop’s schemas, proposed questions, or related themes.

Schemas and Prompts to Think With

In the following figures, we present a subset of many potential schemas for exploring potential human-AI configurations as they relate to our interactions with other people or other PAIAs. In the list below, we use the associated schemas to generatively explore potential use of PAIAs in relationships. Participants are encouraged to generate their own interpretations of these schemas (or of other schemas they generate!) in their position papers and throughout the workshop.

1. Someone consults their PAIA to help them reply to a work email in a professional way, or to re-word harsh feedback.

2. A husband asks his partner’s PAIA to remind them about trash day after their current meeting is over.

3. The family’s shared PAIA is helping them brainstorm costume ideas.

4. Collaborators using their PAIAs as personal knowledge management entities to re-construct a project timeline.%consulting with only their own PAIAs to navigate a tricky conversation about boundaries.

5. Work colleagues collaborating asynchronously while their PAIAs interact directly and continuously.

6. Friends discussing future plans while they offload the logistics of that process onto their PAIAs.

7. A couple attempting to surprise each other with Taylor Swift tickets, which is a delicate enough operation that they do not want their PAIAs to talk to each other for fear of being found out.

8. Non-amicably divorced parents working out the logistics of their split custody for the next month.

9. Two people who know and trust each other very well, but who still cannot figure out what to do for dinner.

10. An older adult who is lonely starts a conversation about their day with the PAIAs of a few people who he feels like he might otherwise be burdening.

11. Someone who recently tested positive for COVID-19 relies on her PAIA to cancel a party.

12. Collaborators ask a few trusted, specialty PAIAs to keep an eye out for patterns they might be missing.

Based on the above—and other—possible interpretations, we hope participants will explore many possible roles that PAIAs might play in our future relationships. Here we list a few questions that participants can choose to use as prompts, though we encourage variations on many of these themes:

  • How can PAIAs optimally balance acting as independent entities and direct representatives of their users?
  • What mechanisms should be implemented to manage the inheritance, memorialization, or transfer of PAIAs and their associated data?
  • How should PAIAs adapt their behavior and responses in situations with distinct social power imbalances?
  • How can they be configured or programmed to actively participate in or drive social and political activism while adhering to legal and ethical guidelines?
  • Should they strive to emulate human characteristics, or openly communicate and act in alignment with their machine nature?
  • How can PAIAs intelligently navigate complex, context-dependent privacy boundaries and ensure secure management of personal information?
  • How should they alter their representation and communication styles based on the relationship and context of the interacting parties?
  • How can PAIAs exchange user preferences to create enriched and personalized verbal and visual interactions while respecting user privacy?
Scroll to top