Human creativity is the crux of innovation, spurring inventions that have shaped the world as we know it. From the first stone tools to generative artificial intelligence (AI), we have continually improved the way we work, learn, and live by design. In today’s digital era, how can we maximise data and technological advancements to maintain our creative design streak?
That is a question Professor Jianxi Luo has been addressing for the past 10 years. “We generate a lot of data every day in our work and life. If we can properly mine, analyse, and make sense of such data to inform design decisions or inspire designers, we can potentially make the design process more creative,” he said.
Professor Luo is the Director of the Data-Driven Innovation Lab at Singapore University of Technology and Design (SUTD). He established the lab in 2013 with a simple mission—to make the design process more informed, inspired, and intelligent. By leveraging data science and AI, the lab aims to develop theories, methods, tools, expert systems, and knowledge infrastructure that can enhance human creativity for design innovation.
His interest in human creativity began early in his career when he was an engineer in a traditional field. He found himself preferring to innovate and create original designs, but had a hard time doing so. The difficulty in conceiving something both original and useful frustrated him for many years. He then joined SUTD, where an idea took hold.
“As a professor at SUTD—established to educate technically grounded innovators and leaders—I have overseen hundreds of innovative design projects by our students. It’s very exciting to help them generate ideas and develop prototypes,” he noted. “I find it exciting because I can inspire my students to design, create, and innovate. But I only have one brain and 24 hours a day. So I thought to myself, imagine a computer with both broad and deep knowledge and creative thinking capabilities that can replace me to help inspire my students generate more and better ideas for their design projects at SUTD.”
Determined to support his students, Professor Luo started researching ways to develop data-driven methods, tools, and expert systems that can improve design idea generation for innovation. The result was the Data-Driven Innovation Lab, built on the concept of data-driven innovation, which he defines as “the process of innovating that draws information, knowledge, and inspiration from data”.
“Intelligentising” the innovation process
Innovation is often a lengthy process beset with uncertainties that designers and engineers can find challenging. According to Professor Luo, four actions in the design process contribute most to potential innovation from the process: opportunity discovery, opportunity evaluation, design generation, and design evaluation.
In the context of innovation, opportunities refer to unmet human needs that new technologies, products and services are designed to address. Innovation involves identifying needs and designing solutions to address those needs.
“Traditionally, we use human social approaches to discover and assess opportunities via interviews, surveys, natural observations, expert panels, etc., and to generate and evaluate design solution ideas through expertise, intuition, brainstorming, idea crowdsourcing, physical experiments, etc. These have been helpful, but the world is changing with the advancement of data science and AI tools,” expressed Professor Luo.
For instance, instead of conducting interviews with multiple individuals to understand their needs, we can use unsupervised machine learning to probe their digital footprints and uncover their latent needs and preferences more effectively. We can also use supervised learning to evaluate potential design opportunities more efficiently. Then, generative algorithms trained with data from prior arts and knowledge can produce many and more novel design solutions ideas. Again, supervised learning may help rapidly assess and select machine-generated design ideas for implementation. AI tools can help us go beyond the limitations of human capability in innovation.
“I believe AI advancements can augment human creativity for the better. Various kinds of machine learning and machine creation capabilities are available to augment the different activities in a design process,” said Professor Luo.
“We don’t actually understand human creativity well, and research on creativity is limited. I believe AI is revolutionary for creativity. Our experiments show that, compared to human brains, cutting-edge, data-driven AI has been able to learn faster from unstructured datasets and synthesise different parts of prior arts to generate many more and more original design ideas in high dimensional spaces,” he added.
Using AI to our advantage
With AI being an inevitable addition to our lives, Professor Luo believes innovators have to embrace it for creative tasks. “Creative AI (CAI) used to be a theory, but real-world capabilities are now in front of us. We have to think about how to use CAI as a safe tool and redesign our workflows. CAI is unavoidable, so we have to be better at managing it to improve the design process,” he opined.
Since 2021, the Data-Driven Innovation Lab has explored the capabilities of generative pre trained transformers (GPT) in generating innovative solutions to complex problems that require multidisciplinary knowledge or difficult-to-learn creative heuristics. For example, the team fine-tuned GPT-3 into an AI design expert specialising in generating biologically inspired design concepts.
“Solutions inspired from nature are exciting and novel, but often hard to develop. AI and GPTs can access and combine various types of knowledge—real-world, common sense, engineering, biology—to conceptualise possible solutions based on nature. Our team has demonstrated that this approach can be used in a number of industrial cases, from bio-composites to flying car skeletons,” described Professor Luo.
Another area that the team is exploring is artificial empathy. “We’ve already developed capabilities in using AI to generate solution concepts to address existing problems. How do we define the right problem to solve based on the needs of people? We believe AGI (artificial general intelligence) has the potential to recognise emotions and empathise with people in different and broad contexts. So we’re developing a controllable AI system with artificial empathy to be able to better understand latent and high-dimensional human needs, and then generate designs to meet such needs,” he explained.
The applicability of the data-driven innovation model is evident. Companies and innovators worldwide have approached the lab to adopt its tools and systems. Professor Luo believes that the industrial relevance of the lab’s research outcome is attributable to the “use-inspired basic research” approach adopted from the very beginning.
“I started the lab to solve the challenge of generating creative ideas that my design students and I faced when we tried to innovate. But I first needed a fundamental understanding of human creativity. My team researched design heuristics and how people use unrelated knowledge in domains with varied knowledge distance to innovate before we applied what we’ve learned to design our methods and tools for data-driven innovation,” he said.
“As it turns out, engineers in industries also wanted to generate creative ideas and found innovation frustrating. They came across our papers and became interested in our work. While we didn’t build our tools specifically for them, our solutions are relevant because they address the same fundamental issue of creativity,” he continued.
The one concern Professor Luo has in relation to AI is the polarisation it might create in society. While AI tools can inspire and inform the creative process, they could make some people overly dependent on them—which is why he advises aspiring innovators to take charge of the design process.
“Embrace AI, but don’t rely on it. If we are led by AI to create and innovate, then we lose our human value. We need to continue exploring ways to remain relevant and dominant, while using AI as a tool, in the design process,” he suggested. “Instead of just focusing on low or operational-level problem-solving, which AI can do a better job at addressing, young innovators will have to think hard about self-motivation, leadership, and their values.”