As artificial intelligence (AI) capabilities rapidly advance, organizations are exploring new ways to leverage these technologies for creative problem-solving and innovation. A recent HBS working paper, “The Crowdless Future? Generative AI and Creative Problem Solving”, – by Léonard Boussioux, Assistant Professor at the University of Washington; Jacqueline N. Lane, Assistant Professor at Harvard Business School and co-Principal Investigator at the Digital Data Design Institute’s (D^3’s) Laboratory for Innovation Science at Harvard (LISH); Miaomiao Zhang, doctoral candidate and researcher at LISH; Vladimir Jacimovic, co-founder of D^3 and CEO of ContinuumLab.ai; and Karim R. Lakhani, HBS Professor and co-founder of D^3 – investigates how human-AI collaboration compares to traditional crowdsourcing approaches in generating novel and valuable solutions to complex challenges.
In the study, the researchers launched a crowdsourcing challenge to develop sustainable business ideas centered on the circular economy. They engaged 125 global participants from diverse industries and used prompt engineering to facilitate human-AI collaborative solutions. Solutions were generated through two main approaches: human crowd (HC) and human-AI (HAI), in which human solvers partnered with LLMs to co-create solutions. Three hundred external evaluators assessed a random subset of 13 solutions from a total of 234, resulting in 3,900 evaluator-solution pairs. Each solution was rated across five criteria: Novelty, Strategic Viability, Environmental Value, Financial Value, and Overall Quality.
Key Insight: Human-AI Solutions Offer Impressive Overall Results
“When considering all factors collectively, the HAI solutions are deemed superior in quality compared to the HC solutions.” [1]
The researchers found that while HC solutions were rated as more novel, HAI-generated solutions scored higher on measures of strategic viability, environmental value, and financial value. Importantly, when all factors were considered together, the HAI solutions were judged to be of higher overall quality. This suggests that AI-augmented approaches may be particularly effective at producing implementable ideas with tangible business value.
Key Insight: Human Guidance Enhances AI Creativity
“Our results demonstrate that for current LLM capabilities, the single instance configuration with iterative human prompts can effectively increase the novelty of outputs while preserving their perceived value.” [2]
The study compared two approaches to human-AI collaboration: an “independent search” and a “differentiated search.” The independent search used a multiple-instance configuration, in which a human solver supplied an initial prompt, and the LLM, using distinct instances, independently generated potential solutions by leveraging its extensive search capabilities. The differentiated search, on the other hand, employed a single-instance configuration in which a human interacted with a single instance of the LLM iteratively, providing a series of prompts aimed at diversifying the model’s outputs, encouraging it to explore various parts of the solution space.
The researchers found that the human-guided, differentiated search approach produced more novel solutions without sacrificing value, highlighting the importance of human involvement in steering AI creativity.
Key Insight: AI Augmentation Offers Massive Efficiency Gains
“In our specific study, whereas the HC solutions cost $2,555 and 2,520 hours to develop, the final HAI solutions were generated in only 5.5 hours and $27.01.” [3]
One of the most striking findings was the dramatic difference in time and cost between human crowdsourcing and AI-augmented approaches. The human-AI method was able to produce comparable or superior results in a fraction of the time and expense of traditional crowdsourcing. As the data cited above shows, AI-driven R&D approaches reduced costs by 99% and time by 99.8% compared to traditional crowdsourcing methods.
Why This Matters
While the paper notes certain limitations, such as lack of expertise in its crowdsourced evaluators and the use of a single LLM for the study, it still holds massive promise for business leaders looking to innovate in the age of AI and underscores the value of a hybrid approach to creativity in the AI era. While human ingenuity is vital for novel ideas, AI excels at generating high-quality, scalable solutions. Organizations might combine human brainstorming for initial concepts with AI for rapid iteration, refinement, and evaluation. Effective use of AI requires skilled human interaction, including expertise in prompt engineering and collaboration. To remain competitive, businesses must balance AI’s efficiency with human originality, avoiding over-reliance on automation. Embracing AI as a complement to human creativity will drive breakthrough innovations and accelerate the delivery of complex solutions.
References
[1] Léonard Boussioux, Jacqueline N. Lane, Miaomiao Zhang, Vladimir Jacimovic, and Karim R. Lakhani, “The Crowdless Future? Generative AI and Creative Problem Solving,” Working Paper 24-005, (Harvard Business School, 2024): 1-39, 22.
[2] Léonard Boussioux et al., “The Crowdless Future?” Working Paper 24-005, (Harvard Business School, 2024), 23.
[3] Léonard Boussioux et al., “The Crowdless Future?” Working Paper 24-005, (Harvard Business School, 2024), 23.
Meet the Authors
Léonard Boussioux is an Assistant Professor in the Department of Information Systems and Operations Management at the University of Washington, Foster School of Business, with an adjunct position at the Allen School of Computer Science and Engineering. His research combines operations research, machine learning, and artificial intelligence with an emphasis on multimodal frameworks and data-driven decision tools, especially in healthcare and sustainability.
Jacqueline N. Lane is an Assistant Professor at Harvard Business School and a co-Principal Investigator of the Laboratory for Innovation Science at Harvard (LISH) at the Digital Data Design Institute (D^3). She earned her PhD from Northwestern University.
Miaomiao Zhang is a doctoral candidate at the Technology & Operations Management Unit at Harvard Business School and researcher with LISH at D^3. Miaomiao is interested in the role of generative AI in shaping organizational knowledge production, learning, and innovation processes. Her current research focuses on human-AI collaboration in the early stage of the innovation cycle, specifically idea generation, refinement, iteration, and evaluation.
Vladimir Jacimovic is the co-founder of D^3 and an Advisory Council member, as well as the CEO at ContinuumLab.ai.
Karim R. Lakhani is the Dorothy & Michael Hintze Professor of Business Administration at the Harvard Business School. His innovation-related research is centered around his role as the founder and co-director of the LISH and as the Principal Investigator of the NASA Tournament Laboratory. He is also the co-founder and chair of D^3 and the co-founder and co-chair of the Harvard Business Analytics Program, a university-wide online program transforming mid-career executives into data-savvy leaders.