How to Use AI to Identify Much-Needed Solutions for Your Business Entrepreneurs can unlock new opportunities and identify innovative solutions using AI to solve problems in one field using ideas and solutions from unrelated fields.

Entrepreneurs are at the forefront of innovation and economic development. Researchers have started to use artificial intelligence (AI) to discover new proteins and develop new sorting algorithms. Following this lead, entrepreneurs can unlock new opportunities and identify innovative solutions using AI to solve problems in one field using ideas and solutions from unrelated fields. This cross-industry innovation is not new to entrepreneurs.

Kanban from Toyota

One of the well-known examples of inspiration from one field impacting another field is "The Kanban methodology." It is now essential for modern agile software development teams. However, this traces back over half a century to Toyota's innovation, inspired by supermarket stocking methods. As supermarkets optimize their inventory to meet consumer demand, Toyota adopted the model to optimize material inventory to meet production demands. The process involved using a card system, "Kanban," to inform the need for materials to enable just-in-time manufacturing. The system has evolved over the years and is central to the Kanban board used in Software Development. This has improved collaboration and the overall software development process.

Related: Beware the Duplicity of OpenAI — 4 Strategies to Safeguard Your Brand in the Age of AI

Checklists from the aviation industry

Another well-known example is the use of checklists to ensure safety in the aviation industry. Pilots and flight attendants use checklists for the completeness of pre-flight operations. This is adapted in healthcare. The most notable example is the World Health Organization's Surgical Safety Checklist. Introduced in 2008, it is a simple tool used in surgical operations to ensure everyone knows vital information such as patient's identity, planned procedure, etc.

It also helps confirm that safety checks are completed before the procedure begins, fostering better communication among surgical staff. In both Aviation and Healthcare, checklists serve as a cognitive aid in complex, high-stress situations, so no critical step is missed due to human error.

Related: 4 Reasons Why Most Entrepreneurs Still Hesitate to Use ChatGPT

More examples

Other examples of such cross-industry innovation resulting in significant impact include

• The Six Sigma process from Motorola reduces defects and errors, minimizing variance.
• Randomized Control Trials from clinical research inspired A/B testing run by consumer internet companies to decide email marketing, product pricing, and many more.
• Engineers redesigned the Shinkansen bullet train based on how the Kingfisher birds can slice through the air to catch their prey.
• Fractal theory, which shows how to build complex shapes from simple repeating patterns, has inspired the construction of compact and complex antennas.

We can see a combination of process adoptions and technical solutions from one field to solve the problems in another. There are various mechanisms through which such cross-industry inspiration and innovations happen. It could be individuals switching disciplines, bringing their knowledge, and applying the techniques to solve problems in the new field. In other cases, dedicated interdisciplinary teams seek inspiration from other industries to tackle a particular challenge. With the rise of the Internet and digital communication platforms, it's easier for experts to learn from each other and share best practices of their respective fields, enabling cross-pollination and accelerating cross-industry innovation.

Leverage LLMs to identify opportunities

Large Language models (LLMs) trained on vast datasets across the internet are used for creative tasks such as poem writing and image creation. It is possible to fine-tune and train them to identify such cross-pollinated cross-industry innovations. This requires training in a large language model focusing on multidisciplinary understanding and analogical reasoning.

Until such an LLM is trained, a domain expert could leverage existing Large Language models by sharing key research papers and prompting them to identify a solution. Their domain expertise could later verify the viability and feasibility of the solution. For example, prompting GPT4 to read scholarly articles about "Physics of Protein Self Assembly" and "Carbon Nanotube Fabrication (CNT): Challenges" resulted in GPT4 proposing a solution to improve the CNT fabrication by potentially developing a solution to control the chirality and diameter of Carbon nanotubes. LLM tries to inspire the idea of designing anisotropic interactions between the catalyst particles and carbon precursors, guiding the directional growth of Carbon Nanotubes. Another inspiration is to leverage the patchiness and interaction specificity of proteins, then engineer to control the growth directions of Carbon Nanotubes.

Related: How to Use AI to Amplify the Potential of Your Team

After repeating the experiment with Claude 3 Opus, we can see that the idea of anisotropy interaction was validated again. Below is the excerpt of the last paragraph generated by Anthropic Claude for the same prompt.

"In summary, borrowing the concepts of interaction anisotropy and assembly kinetics from protein physics could point the way to rational catalyst design for controlled, efficient CNT growth. The protein physics article provides a useful framework for thinking about self-assembly that could translate productively to the CNT fabrication challenge with further research."

Hallucination is not a bug

It is entirely possible that the solution is infeasible and that LLMs are hallucinating. Even if the LLM's proposed solution is not practically feasible, it might inspire experts. LLMs could act as copilots, analogous to Jarvis and Ironman. Hallucination is not a bug but a feature of forced cross-pollination to inspire innovation. Entrepreneurs should invest in learning how to leverage LLMs to identify groundbreaking zero-to-one cross-industry innovation.

Retrieval Augmented Generation applications built on top of Large Language Models could be used to identify innovative solutions with humans in the loop to avoid hallucination and ensure practicality. Entrepreneurs are best positioned to do this by working with domain experts.