Creativity, GenAI, and Software Development: A Future Together

Victoria Jackson (University of Southampton, GB, v.jackson@soton.ac.uk)
Rafael Prikladnicki (Pontifiícia Universidade Católica do Rio Grande do Sul, BR,
rafael.prikladnicki@pucrs.br)
Daniel Russo (Aalborg University, DK, daniel.russo@cs.aau.dk)
Margaret-Anne Storey (University of Victoria, CA mstorey@uvic.ca)
André van der Hoek (University of California, Irvine, US, andre@ics.uci.edu)

This Dagstuhl Perspectives Workshop 25412 “Creativity, GenAI, and Software Development: A Future Together” brought together leading researchers and thought leaders from industry to collectively examine the evolving relationship between creativity, Generative AI (GenAI), and software development, and to articulate a forward-looking research agenda at this critical inflection point for the field.

Although much of the discourse, especially in industry, has focused on productivity gains from GenAI, this workshop deliberately shifted attention toward creativity as a central dimension of software development work.

During five days of highly interactive and trust-building activities, participants explored how creativity manifests in software development, how it is shaped by tools, teams, and organizational contexts, and how GenAI can both amplify and erode creative practice. Rather than treating creativity as an abstract or purely individual trait, discussions emphasized its situated, socio-technical, and collective nature, spanning problem framing, exploration, sensemaking, and decision-making throughout the software lifecycle.

A key insight emerging from the workshop is that GenAI, in disrupting software development by reconfiguring who does what and in changing how value is created and assessed, creativity is simultaneously more important than ever and seriously challenged. Participants surfaced optimistic futures, in which GenAI expands creative reach and lowers barriers to exploration, and more cautionary trajectories, where over-automation risks flattening diversity of thought, narrowing solution spaces, or displacing reflective practice. These tensions highlight the need for research that goes beyond tool performance to examine creative quality, meaning, values, and long-term impacts on people and organizations.

The workshop culminated in the identification of five interrelated research directions that form the basis of the forthcoming Dagstuhl Manifesto, provisionally titled “Creativity is the New Productivity.” These directions call for: (1) developing a critical perspective on GenAI and creativity, (2) deeper characterization of creativity in software development; (3) stronger evidence of how creative work generates value; (4) exploration of the design space for GenAI-driven tools and processes that responsibly amplify creativity; and (5) synthesis across fragmented understandings, perspectives, and emergent publications to effectively promote knowledge sharing and generate impact.

Beyond the Manifesto itself, the workshop fostered a shared language, durable collaborations, and a collective commitment to shaping futures in which GenAI supports – not supplants – human creativity in software development. Taken together, the outcomes of this workshop argue for reframing success in software engineering: from efficiency alone toward creative, humane, and value-aligned systems of work.

The high-level objective of this Perspectives Workshop was to bring together thought leaders from academia and industry to develop a comprehensive research agenda on the intersection of creativity, GenAI, and software development.

Creativity is a bedrock of software development: without it, innovation would stall and the day-to-day problem solving that leads to new features and improvements to existing features would be gravely hampered. Yet, whereas creativity in other disciplines such as arts, engineering, and architecture has been the subject of a vast amount of research, little is known about the nature, importance, and effect of creativity in software development.

The advent of GenAI has thrust the role of creativity in software development to the fore. Some envision a gentle evolution toward GenAI-driven tools that amplify the creative work of developers and make them more effective. Others see a radical near future in which GenAI usurps all rote work, with the sole job left for developers to creatively direct the tools to produce innovative software and features. While a major portion of research, studies, and discussions in the literature and media have concentrated on the impact of GenAI on productivity, this Perspectives Workshop considered that understanding the short-term, medium-term, and long-term impacts of and connections among creativity, GenAI, and software development warrants equal, if not greater, attention.

To discuss, anticipate, and proactively shape the future of creativity augmented by GenAI in software development, this Perspectives Workshop brought together leading academics and practitioner thought leaders to address a range of questions, including: what is the nature of creativity in software development; how might GenAI shape, enhance, and perturb software creativity; what does that mean for the resulting software products; which kinds of novel development tools may be needed; how might organizations and their teams adapt; and what unintended consequences may result?

The workshop had five specific goals:

• $\mathrm{■}$

  Raise awareness of the important role of creativity in software development and the opportunities and pitfalls that may arise with the advent of GenAI.

• $\mathrm{■}$

  Articulate different futures in terms of how GenAI may impact creativity in software development, as bracketed by other relevant factors such as value, productivity, and developer experience and wellbeing.

• $\mathrm{■}$

  Use these futures to develop a comprehensive research agenda that helps shape the future of creativity, GenAI, and software development.

• $\mathrm{■}$

  Ideate new GenAI driven tools that amplify creativity across different software development activities and tasks in positive ways while minimizing potential risks.

• $\mathrm{■}$

  Promote an interdisciplinary community of researchers that cares about, understands, and drives how creativity and GenAI will shape future software development.

Participants in the workshop were specifically selected to bring together expertise in a wide range of backgrounds, including creativity, GenAI, innovation, software analytics, productivity, human-AI teams, design and expertise, software tools, problem solving, and human cognition, among others. To effectively exploit this wide variety of expertise for purposes of future agenda setting, the workshop was run in a highly interactive fashion, without any formal presentations. Instead, participants engaged in many different collaborative activities that were carefully structured and sequenced to build trust, collectively take stock and create common ground, generate many different ideas, articulate possible futures out of those ideas, and refine the most promising directions for future research.

The remainder of this document details how the workshop proceeded, and accompanies the Manifesto that is to be published separately in the first half of 2026. Next, we first provide an overview of the workshop’s approach (section 4), followed by a description of each of the workshop’s working sessions (section 5). The report then concludes with a short preview of the Manifesto’s planned contents (section 6).

The workshop’s overarching goal was to discuss, anticipate, and proactively shape the future of research into the intersection of creativity, Generative AI, and software development. To facilitate discussion, including the identification of future research directions, the workshop was purposefully designed to be interactive and fun, incorporating small-group activities. Many of these activities were drawn from Liberating Structures⁶⁶6https://www.liberatingstructures.com/. Liberating Structures are a set of facilitation strategies designed to involve participants in meaningful discussions by breaking conventional patterns, encouraging fresh insights, and collaborative problem-solving. Several of the organizers had prior experience using Liberating Structures in workshops and found them helpful for fostering vibrant discussions for shaping research strategies and directions.

A key output from a Perspectives Workshop such as ours is a Dagstuhl Manifesto. The workshop organizers wished to have an outline of the Manifesto’s contents by the end of the week, which could then be worked on asynchronously by all participants. This required some lightweight structuring of the workshop to move from the many diverse perspectives on creativity among participants to a concrete set of future research directions. The workshop was thus organized with a daily goal, with each day building upon the work and insights from the prior day. The first day’s goal was firmly scene-setting, getting participants to discuss their perspectives on creativity and start thinking about where the future lies. The second day explored the amplification of creative software work through GenAI. The third (short) day began by identifying future research directions, which were subsequently refined on the fourth day, before agreeing on the Manifesto’s contents and starting to contribute to it on the final (short) day.

Table 1 summarizes each of the working sessions that took place during the workshop. As can be seen, there was more structure in the early part of the week, with many sessions to help participants explore the topic. The latter part of the week offered participants an opportunity to provide more details on their preferred research direction.

While the organizers had initially planned the week’s activities, they accepted that the plan might change depending on the outcomes of each day. The organizers met at the end of each day to reflect and decide whether any changes would be required to the upcoming plans. This reflection led to some minor changes, introducing additional Liberating Structures to the planned ones.

On the first day with all the participants, the organizers explained that the workshop was designed to be interactive and stimulating. Liberating Structures were introduced, followed by guidance for participants. Firstly, to create a safe space where participants could openly share their experiences and ideas, the workshop adopted Chatham House rules⁷⁷7https://www.chathamhouse.org/about-us/chatham-house-rule: “Participants are free to use the information received, but neither the identity nor the affiliation of the speaker(s), nor that of any other participant, may be revealed.” Secondly, the participants were encouraged to engage actively as their insights were crucial to the collective success. We thus asked them throughout the week to contribute by sharing freely, listening intently, and building on others’ ideas. Notably, the whole week’s agenda was not shared with participants to allow the organizers to re-plan as needed.

We hope that by sharing the approach taken to facilitate the collective dialogue necessary to produce a Dagstuhl Manifesto, future workshop organizers can learn from our experience when designing their own Dagstuhl Perspectives Workshop.

The next section shares details of each of the working sections shown in Table 1.

This section describes the various working sessions held during the workshop (as listed in Table 1). The majority of the sessions used techniques drawn from Liberating Structures.

One of the organizers (primarily Daniel Russo) acted as a facilitator to each session. The facilitator introduced the session’s purpose and format, and ensured it stayed within its allotted time. Each session is now described. Accompanying this narrative are images of several artifacts produced during the workshop, as shown in Figure 1.

Importantly, to help grow personal connections and to build positive interactions, each day began with an Impromptu Networking session and ended with a Positive Gossip session. Impromptu Networking asked each participant to pair up and discuss provided questions, such as “What does creativity mean to you?” for several minutes before switching partners to discuss a different question. “Positive Gossip” asked each participant to first reflect on their own upon the day and identify an idea, moment of curiosity, insight, or a personal quality of someone they appreciated. Participants then formed triads to share “positive gossip” about people outside their triad. The whole group then got together to share emerging themes.

A key result from the workshop is a Dagstuhl Manifesto detailing future research directions and places these research directions into a larger societal context. The workshop identified five future research directions to form the basis of the Manifesto (to be published in the first half of 2026). This Manifesto will be collectively authored by all of the workshop participants. The Manifesto is important, but equally important are the connections formed during the workshop. These connections led to various participants planning to collaborate on more detailed research ideas that were identified during the week. Indeed, at the time of this writing, some joint papers have already been authored and submitted. Over time, the group plans to stay connected and to continue to discuss and advance research and practice at the intersection of creativity, GenAI, and software development.

The five research directions to be included in the Manifesto are:

• $\mathrm{■}$

  Critical perspectives: Understanding how to develop a science of critical GenAI use in creative software development where biases are understood, intentional architectures are used, multicultural accounts of creativity are incorporated, sustainability is a design parameter, and organization practices and processes guide developers in appropriate use that limits harm.

• $\mathrm{■}$

  Characterizing creativity: Understanding what creativity in software engineering looks like for different software development roles and for different aspects of software development, how it plays out in the day-to-day work of software developers as well as at a larger, more organizational scale toward innovation, and how this may differ across individual character traits, team practices, and organizational structures.

• $\mathrm{■}$

  Evidence of value: Understanding the value that creativity brings to individuals, teams, and organizations in terms of both process and the resulting product, under what conditions it flourishes or may be hampered, and how it can be effectively promoted and practiced.

• $\mathrm{■}$

  Design space: Developing a design space for GenAI-driven tools and processes that promote creativity in development, together with identifying interaction and other design principles that can guide the design of future novel tools and processes that augment and amplify creativity in software engineering.

• $\mathrm{■}$

  Synthesize research: How to bring together and package research results and practical knowledge in GenAI, creativity, and software development so to increase the change of widespread impact; this entails effectively synthesizing and communicating emerging results and knowledge in an easily-accessible manner.

The organizers wish to thank Schloss Dagstuhl, the Leibniz Center for Informatics for hosting and supporting the workshop. Schloss Dagstuhl is a fantastic venue for fostering the lively discussions and collaboration necessary to move research forward. All of the staff was truly awesome in supporting whatever the workshop needed; we are extremely grateful.

• $\mathrm{■}$

  Aamir Amin – University of Portsmouth, GB

• $\mathrm{■}$

  Souti Chattopadhyay – USC – Los Angeles, US

• $\mathrm{■}$

  Sarah D’Angelo – Google – Mountain View, US

• $\mathrm{■}$

  Peter Dalsgaard – Aarhus University, DK

• $\mathrm{■}$

  Jeanette Falk – Aalborg University Copenhagen, DK

• $\mathrm{■}$

  Guo Freeman – Clemson University, US

• $\mathrm{■}$

  Jonas Frich – Aarhus University, DK

• $\mathrm{■}$

  Kiev Gama – Federal University of Pernambuco – Recife, BR

• $\mathrm{■}$

  Cat Hicks – Pluralsight – Draper, US

• $\mathrm{■}$

  Maliheh Izadi – TU Delft, NL

• $\mathrm{■}$

  Victoria Jackson – University of Southampton, GB

• $\mathrm{■}$

  Michael Keeling – Kiavi – San Francisco, US

• $\mathrm{■}$

  Amy Ko – University of Washington – Seattle, US

• $\mathrm{■}$

  Raymond Li – ServiceNow – Montreal, CA

• $\mathrm{■}$

  Olga Lvova – JetBrains – Yerevan, AM

• $\mathrm{■}$

  Mary Lou Maher – University of North Carolina – Charlotte, US

• $\mathrm{■}$

  Adriana Meza Soria – MIT-IBM AI Watson Lab – Cambridge, US

• $\mathrm{■}$

  Gail C. Murphy – University of British Columbia – Vancouver, CA

• $\mathrm{■}$

  Karim Nakad – Meta Platforms – Bellevue, US

• $\mathrm{■}$

  Marian Petre – The Open University – Milton Keynes, GB

• $\mathrm{■}$

  Rafael Prikladnicki – Pontifícia Universidade Católica do Rio Grande do Sul – Porto Alegre, BR

• $\mathrm{■}$

  Daniel Russo – Aalborg University Copenhagen, DK

• $\mathrm{■}$

  Zohreh Sharafi – Polytechnique Montréal, CA

• $\mathrm{■}$

  Mary Shaw – Carnegie Mellon University – Pittsburgh, US

• $\mathrm{■}$

  Margaret-Anne Storey – University of Victoria, CA

• $\mathrm{■}$

  André van der Hoek – University of California, Irvine, US

• $\mathrm{■}$

  Bogdan Vasilescu – Carnegie Mellon University – Pittsburgh, US