Ethical co-creativity with AI: A hands-on workshop at the International Conference on Computational Creativity (ICCC 2024)

Abstract

This hands-on workshop explores how software engineering techniques used to develop AI systems can be translated into creative practice to support and facilitate human - AI co-creativity. The workshop introduces participants to techniques for identifying ethical concerns around AI, and explores how these concerns can inspire creative output across different domains. As part of this workshop, participants will be offered the opportunity to develop a creative output such as a story or piece of art inspired by AI ethical concerns, using AI as a co-creative tool. Participants will then work collaboratively to analyse and improve these outputs using a modified version of existing AI development hazard analysis techniques.

Registration for the workshop

To register for the workshop, please email the organisers (details below), providing the following information:

• your name and affiliation;
• the type/s of output you are interested in creating, e.g., short story, flash fiction, artwork, music;
• whether you intend to co-create these outputs with an AI/s;
• confirmation of your commitment to the workshop (see below);
• any questions you have; and
• any other information you think the organisers should be aware of, e.g., dietary requirements, neurodiversity, physical dis/ability.

If you have questions, prior to registration, please contact the organisers (see below).

Organisers

Prof. Austen Rainer, Queen’s University Belfast, Northern Ireland (email; Primary contact)

Dr. Catherine Menon, University of Hertfordshire, England (email)

Description of the event’s purpose and scope

This workshops aims to explore how we can use structured analysis techniques drawn from software safety engineering to support both the ethical creativity of individual humans and the ethical co-creativity of humans with AI.

We’ll begin by presenting participants with information about an example AI system and scenario, such as an assistive robot intended to help vulnerable users live independently in their homes. The workshop then explores how structured creative practice can facilitate the identification and discussion of ethical hazards and concerns with the proposed AI systems.

We’ll use two expansions of HAZOP in this workshop: the extant CHAZOP methodology and a new proposed methodology, EHAZOP, which is directed towards the identification of ethical hazards. Together, the workshop explores how creative practice can be used to examine the space of ethical hazards associated with the AI system provided.

Expected commitments of the expected participants

Indicative number of participants are between six and eighteen participants. If there is an unsufficient number of participants, we may need to either cancel or substantially re-organise the workshop.

We ask that participants:

• Have a background in engineering, software, computation or safety assessment
• Will create a supportive space for co-creativity;
• Are willing to rapidly produce preliminary creative outputs in a time-constrained situation;
• Are willing to share those creative outputs with others in the workshop;
• Are interested in stories in their various forms, e.g., flash fiction, short stories;
• Are open-minded to the unpredictability that arises with creativity.

Note: Our main focus will be on written stories however we would like to try to accommodate other kinds of story-telling (e.g., music). If you are interested in producing a non-literary output, such as music or artwork, please contact the organisers so that we can seek to accommodate you.

Indicative format and duration of the workshop

Activity
Welcome and introduction
Activity 1: scenario, ethical hazards and EHAZOP
- Introduction to the AI system / scenario which will be used as the seed for creative practice in the workshop
- Identification of potential ethical hazards associated with this system / scenario, using the EHAZOP (Ethical HAZOP) methodology
Break
Activity 2: creative output inspired by the scenario
- Participants may choose to create different kinds of output. We will focus on short, written stories however we will also seek to accommodate other kinds of creative output.
- Participants will have the option to create outputs individually or in collaboration with generative AI tools, such as ChatGPT. When co-creating the story with AI/s, the human participant will be responsible for managing access to their chosen AI tools (if any), and must clearly identify the contribution/s made by the AI to the creative output.
Lunch
Activity 3: CHAZOP workshop on creative outputs
- Groups of participants collaboratively workshop a selection of the outputs from Activity 1 using the CHAZOP or EHAZOP methodology
- Depending on the number of participants and the types of output, participants may be organised into groups based on type of output, e.g., one group reviews a written story, another group reviews a musical output.
Break
Activity 4: Open discussion and evaluation
Items to discuss include (these are indicative):
1. Does xHAZOP encourage creativity?
2. Does xHAZOP encourage ethical creativity?
3. How might we assess xHAZOP’s effects on ethics?
4. How might we assess zHAZOP’s effects on creativity?
5. Does xHAZOP apply to different kinds of creative output, or is it more applicable to specific kinds of output, e.g., linear outputs, or written stories?
Close
Note: there may be the opportunity to organise a spontaneous celebration/performance of some or all of the outputs at the end of the workshop

Indicative format of submissions

• When participants register for the workshop, they will be asked to indicate what kind of creative output/s they would like to produce, e.g., written output, musical output. This is to help us more effectively organise the workshop.
• Participants are not required to bring any pre-prepared work to the workshop.
• Participants are required to bring any resources they might want to use in the workshop, e.g., a laptop to give access to ChatGPT, or to a server running some AI.
• Participants considering musical outputs are asked to bring headphones.