WP7: Human-centric AI for Creative Problem-Solving

This work package is about using creative AI to solve real problems. That can mean designing a product, where art is part of the job, or a rescue operation, where people must find a clever way out with what they have at hand.

This work package explores how a person or a group can use creative AI to solve a specific task. The tasks range from industrial design, where artistic methods are part of the process, to emergency rescue, where a pressing problem has to be solved by using the available resources creatively.

This WP explores how a person or group can use Creative AI for task-specific problem-solving. The tasks considered in this WP range from industrial design, where artistic methods are integral to design processes, to emergency rescue scenarios, where imminent problems must be solved by creatively using available resources.

Description

We work with people who create for a purpose: designers, film makers, people who run machines in industry, and emergency workers. We build on the ideas about creative AI that the whole centre develops.

For these people, AI has to give freedom and surprise but also control. They must be able to judge what the AI suggests, keep the good parts and throw away the rest. Sometimes that happens step by step, in a conversation with the machine; sometimes it happens live, with a robot or a device.

The AI has to respect real limits: the laws of physics, the law itself, safety rules. We build those limits into the tools from the start. A hard part is the clash between doing something creative and doing what the job requires.

The work package addresses creative practitioners who work towards a goal, as different as designers, film makers, operators in industrial settings and emergency personnel, and it adopts the principles of creative AI that the centre develops through its three perspectives.

In a goal-driven setting, new frameworks are needed that give creators freedom, surprise, control and inspiration, and let them evaluate, select, discard and apply what the AI contributes. The process has to be interactive, from step-by-step exploration through prompts to real-time collaboration with an AI that has a body.

The process should help creators meet the requirements of their profession and other practical demands, finding workable solutions rather than avoiding them. A central challenge is the tension between artistic freedom and the practical requirements the creator has to fulfil.

The AI systems developed here operate within stricter limits than elsewhere, from the laws of physics to legal frameworks such as requirements for realism, and those limits are encoded into the principles the systems are built on. Together with the other work packages, the work produces new knowledge about the algorithmic, personal and societal challenges of applying creative AI in practice.

We target creative goal-based practitioners, including professionals as diverse as designers, filmmakers, human operators in industrial settings, and emergency service personnel, and we adopt the principles and ideas of Creative AI that MishMash co-creates through its three perspectives.

To achieve this in a goal-driven context, new frameworks must be developed that give creators freedom, surprise, control, and inspiration and allow them to evaluate, select, discard, and apply AI’s contributions. The creative process must be interactive, ranging from iterative prompt-based exploration to real-time collaboration with an embodied AI.

The process should support creators in addressing their professions’ requirements and other practical challenges, finding pragmatic solutions instead of circumventing them. A key challenge here is potential conflicts between artistic freedom and the practical requirements that the creator must fulfil.

AIs developed in this WP will be limited by stricter boundaries than others, from the laws of physics to legal frameworks, such as realism constraints. We will go beyond informed AI approaches in our focus on encoding these limitations into the development principles for Creative AIs. WP7 will generate new knowledge of algorithmic, personal and societal challenges when applying Creative AI in creative practices in collaboration with other WPs facilitated by MishMash’s cross-cutting perspectives.

Research Questions

How can creative AI help people solve problems while they stay in control and keep their own style?
Can an AI that seems to understand how you feel make you feel more capable and more in charge?
How should AI tools be designed so they are clear, easy to use and open to more people?
How far can we adapt creative AI to each person and each job, and still build one shared framework?

RQ7.1: How can creative AI be brought into creative problem-solving so that human agency, control and personal expression are kept and developed?
RQ7.2: How can empathy emulated by an algorithm give people a stronger sense of agency and capability?
RQ7.3: How can interfaces for AI and AI-supported services be designed to improve the user’s experience and transparency, and make co-creative processes more accessible?
RQ7.4: How far can creative AI systems be personalised and adapted to a wide range of uses and settings, and what generic frameworks does that allow?

RQ7.1: How can we integrate Creative AI in creative problem-solving processes to maintain and develop human agency, control and personal expressive capabilities?
RQ7.2: How can we use algorithmically emulated empathy to achieve a greater sense of human agency and empowerment for humans?
RQ7.3: How can interfaces for AI and AI-supported services be designed to enhance user experience and transparency, making co-creative processes more accessible?
RQ7.4: To what extent can we personalise and customise Creative AI systems that extend to a wide range of applications and contexts, allowing the definition of generic frameworks?

Approaches

CREATE: AI that helps people be creative towards a goal, while staying within the limits of physics, law, safety and good sense.
EXPLORE: how creative AI can support people while they keep the feeling that the work is their own.
REFLECT: on what creativity means when solving problems: for the AI, for the creator, and for whoever looks at the result.

CREATE: AI algorithms that strengthen human creativity in tasks with external goals, while staying within physical, legal, emotional, social and other constraints, learned or given.
EXPLORE: how creative AI algorithms can support people while preserving their sense of agency and ownership, by observing, understanding and adapting to human expression.
REFLECT: on how creativity in problem-solving is perceived, from an AI system’s ability to tell recreation from creation, through the creator’s intent, to how a user or an observer sees the result.

CREATE: AI algorithms that can empower human creativity in pursuing tasks with external goals while remaining within (learned) bounds regarding physical, legal, emotional, social and other constraints.
EXPLORE: how Creative AI algorithms can empower humans while ensuring their sense of agency and ownership by observing, understanding, and adapting to human expression.
REFLECT: about the perception of creativity in problem-solving, ranging from the ability of an AI system to separate recreation from creativity, the creator’s intent, to the perception of a user or an observer of the creation.

Sustainability

This work helps with one of the United Nations’ goals for a better world. Tools for people and machines to create together, built for real jobs with real limits, support innovation in the sectors that take them up (SDG 8).

The centre’s other commitments on sustainability, openness and reuse are on the open research page.

This work package contributes to one of the UN Sustainable Development Goals directly. Task-specific tools for human-machine co-creation, built for practitioners working under real constraints, foster innovation in the sectors that adopt them (SDG 8).

The centre’s other commitments on sustainability, openness and reuse are on the open research page.