This article was originally published on The Conversation, an independent and nonprofit source of news, analysis and commentary from academic experts. Disclosure information is available on the original site.
___
Authors: Camille Grange, Associate Professor, Department of Information Technologies, HEC Montréal; Mickaël Ringeval, Postdoctoral Fellow, Department of Information Technologies, HEC Montréal; Simon Bourdeau, Professor, Département d'analytique, opérations et technologies de l'information, Université du Québec à Montréal (UQAM), and Théophile Demazure, Assistant Professor in Information Technologies, HEC Montréal
Generative artificial intelligence (AI) tools like ChatGPT or Dall-E are changing how creative work is done, particularly in industries that rely on innovation.
However, AI use in the innovation process requires careful considerations. Our research shows that the key to success is understanding and leveraging the distinct but complementary roles that both humans and AI play.
Innovation is vital for any business that wants to succeed today. In fact, 83 per cent of companies see innovation as a top priority, yet only three per cent are ready to turn this priority into action. This shows how much companies need to improve their approach to innovation.
Innovation is about solving complex problems that result in real improvement. It’s not just about coming up with good ideas — it also involves knowledge work, which is the process of using information to create something valuable.
Generative AI can help businesses get ready to innovate by making knowledge work easier, but its full potential in this area is still not fully understood.
Design sprints
Our team, which includes academic researchers with expertise in emerging digital technologies and a practitioner experienced in leading human-centred innovation projects, conducted a detailed study of how generative AI was used in design sprints at three organizations. (The study is available as a pre-print and has been submitted to a journal for peer review).
A design sprint is a fast, structured process for solving important problems that helps teams test if a product, service or strategy will work. Sprints are useful because they reduce the risks and costs of traditional product development
During a design sprint, a small team of five to seven employees from different areas works together intensely for a few days to solve a problem. Their work is co-ordinated by a facilitator, who organizes activities, guides the team, keeps track of progress, makes sure the goals are clear and that time is used efficiently.
The first stage of a design sprint focuses on understanding and defining the problem, while the second stage is about creating and testing a solution. Both stages require teams to use two key types of thinking:
- Divergent thinking, which means coming up with many different ideas and possibilities.
- Convergent thinking, which means narrowing those ideas down to identify priorities or solutions.
Our study examined how the facilitator used generative AI tools like ChatGPT, DALL-E 3 or Uizard to help the team engage effectively in both divergence and convergence.
AI and humans working together
In divergent thinking activities, we found two main benefits of using generative AI. First, it encouraged teams to explore more possibilities by providing baseline ideas as a starting point. Second, it helped to rephrase and synthesize unclear ideas from team members, ultimately leading to better communication within the teams.
One participant told us:
“Sometimes we had a lot of ideas, and the AI summarized them into a concise text. This allowed us to wrap our head around it. It gave us a base, there were many fragmented ideas that everyone had contributed, and now we had a text we all agreed on. This way, we started from the same base which served as a springboard to move forward.”
The real value of generative AI was thus not in contributing brilliant new ideas itself, but in the valuable synergies that emerged from the process. Team members used their contextual knowledge and stayed in charge of the process while the AI helped to better convey their ideas, expand exploration and address possible blind spots
Making better informed decisions
We noticed different dynamics in convergence activities where teams had to make decisions after demanding sessions of idea generation. By that point, team members were usually mentally exhausted. Generative AI was especially helpful for doing the heavy lifting during this part.
The AI helped manage the information-intensive tasks necessary for team alignment like reformulating, summarizing, organizing, comparing and ranking options. This reduced the mental strain on team members, allowing them to focus on important tasks like evaluating ideas. In this process, the team was responsible for:
- Checking AI’s outputs to make sure the content was accurate and useful. For example, ChatGPT and Uizard helped create draft scenarios and prototype drafts to validate their concept, but the team still had to refine them to meet project goals.
- Adding their own insights and contextual nuances to guide final decisions, considering factors like feasibility, ethics and long-term strategic impact.
One participant said:
“Sometimes, the AI would focus onto details that were insignificant to us…Sometimes we needed less general synthesis and more personalized input.”
Overall, this form of human-AI collaboration in convergent activities helped the team make better informed and more confident decisions about which problem to focus on and which solution to pursue. This made them feel in control of the sprint’s final outcomes.
One participant said:
“For pivotal phases like making decisions or voting on something important like a success factor, if we relied solely on AI to determine what is important, there would be rejection. We are better positioned to know. We are the employees who will execute the final solution.”
Challenges and opportunities
Consistent with research on cognitive automation and intelligent automation, we found that generative AI was of great help in handling cognitively demanding tasks like reformulating poorly articulated ideas, summarizing information and recognizing patterns in team members’ contributions.
A key challenge with using Generative AI in innovation is ensuring it complements, rather than replaces, human involvement. While AI can act as a useful companion, there’s a risk it could reduce team engagement or ownership of the project if overused.
The design sprint facilitator told us:
“Feasibility needs to be balanced with desirability. You could technically automate most of the process but that would kill the need for pleasure, interaction, and humans’ doubts won’t be addressed; plus humans need to own the problem — all these are essential elements in a human-centred innovation process.”
Consequently, regularly assessing AI’s impact within this process is crucial in order to maintain a healthy balance. Automation should enhance creativity and decision-making without undermining the human insights that are central to innovation.
As AI continues to develop, its role in innovation will grow. Companies that integrate AI into their workflows will be better equipped to handle the fast-paced demands of modern innovation. But it’s important to understand both the strengths and limits of AI and humans to ensure this collaboration is effective.
___
This article was co-authored by Cédric Martineau, CEO and innovation management consultant at Carverinno Consulting.
The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.
___
This article is republished from The Conversation under a Creative Commons license. Disclosure information is available on the original site. Read the original article: https://theconversation.com/generative-ai-can-boost-innovation-but-only-when-humans-are-in-control-240637
Camille Grange, Associate Professor, Department of Information Technologies, HEC Montréal; Mickaël Ringeval, Postdoctoral Fellow, Department of Information Technologies, HEC Montréal; Simon Bourdeau, Professor, Département d'analytique, opérations et technologies de l'information, Université du Québec à Montréal (UQAM), and Théophile Demazure, Assistant Professor in Information Technologies, HEC Montréal, The Conversation
