In the two years since the debut of generative AI, companies in various industries have found these text- and image-based tools can help their innovation teams invent new products and create new designs for their existing products and services.

It stands to reason. In general, top innovators are vastly more successful at creating business value from their investments in tech and R&D, according to recent McKinsey research. Generative AI (GenAI) represents a fresh addition to existing efforts. It has proven valuable in accelerating drug discovery, as EY notes. It’s also strengthened Toyota’s next generation of safer car designs.

But is this work as easy as firing up ChatGPT or Gemini and waiting for it to spit out the next business-changing idea? No. You’ll need to start with a good understanding of how to coax the best from GenAI. These AI tools, by themselves, can’t provide magic answers; they require expert supervision. They can propose new designs and products by quickly scanning data and creating new (and sometimes unexpected) combinations for you to evaluate.

These guidelines will help you get started using GenAI to help your product and innovation teams develop fresh ideas they can refine and sharpen. Two recommendations to start: Use your own data to retain your ownership of the results. And focus your experiments on real-world customer problems.

Your creative collaborator

As the next big thing in AI innovation assistance, GenAI offers companies a fresh way to develop new product ideas and novel designs for existing products. In the former, product developers input text prompts and receive back text containing ideas. In the latter, designers use text prompts to generate images of products that don’t yet exist but represent advances in the current generation of products.

In both situations, prompt engineering – the process of creating input texts to guide the responses of the AI models – is paramount. However, in each case, the exercise of using GenAI is fundamentally different. Blue-sky brainstorming has the purpose of narrowing the field of possibilities from a wide aperture down to a few promising ideas. On the other hand, extending the functionality of current products through image generation is about applying GenAI’s ability to be a visual designer. Whatever your specific innovation goal, you should now be using a large language model (LLM) – such as ChatGPT, Google’s Gemini, Claude, and a growing list of others – as your “creative co-pilot” that helps you come up with new ideas, faster.

For example, Wharton professor Christian Terwiesch was astonished by the results of an experiment he conducted with his MBA students using ChatGPT 4.0, as noted in this paper. They found that the model, when prompted to brainstorm product ideas for college students that would retail for less than $50, came up with 200 new product ideas in about 15 minutes. The LLM scored higher than the students on efficiency, quality of product ideas, and cost of generating the ideas. It performed especially well when shown examples of good products.

Large language models such as ChatGPT, Google’s Gemini, Claude, and a growing list of others can serve as your “creative co-pilot” that helps you come up with new ideas, faster.

A follow-up academic research paper that discussed Terwiesch’s experiment promoted the approach where humans ideate in concert with GenAI. “Overall, a co-creation mode where humans propose the initial ideas, followed by a GPT-assisted revision process, can achieve greater balance between idea quality and diversity (than relying on humans or GPT alone),” wrote the paper’s authors.

But, as with any new technology, you’ll need to ensure guardrails are in place for experimentation and usage. We have tips on how to set up the process. We help identify how organizations can evaluate what GenAI offers to make the most of time and resources.

Three ways GenAI can be used for new product ideas

Use GenAI to enhance existing idea generation processes by speeding things up and adding a data-informed voice to the brainstorming.

1. To shorten the brainstorming cycle.

Terwiesch, who is co-director of Wharton’s Mack Institute for Innovation Management, writes that everybody should be using ChatGPT to help them generate ideas. “It’s cheap. It’s fast. It’s good. What’s not to be liked? Worst case is you reject all the ideas and run with your own. But our research speaks strongly to the fact that your idea pool will get better,” wrote Terwiesch. In terms of provoking thinking and getting ideas flowing, GenAI is invaluable when speed is of the essence, which it always is.

2. To augment human creativity by promoting divergent thinking.

One of the most important aspects of this creative process is to seek a diversity of viewpoints. Traditionally, you’d get the most varied group of people you could find—different ages, different expertise, different experiences—around a conference table and see what happens. GenAI can shortcut that process and supply the necessary range of opinions without even the need to order lunch.

“Rapidly and inexpensively, producing a plethora of designs in this way allows a company to evaluate a wide range of product concepts quickly. For example, a clothing company that uses generative AI to create new designs for T-shirts could stay on top of trends and offer a constantly changing selection of products to customers,” Tojin T. Eapen, a senior fellow at the Conference Board and his coauthors of a Harvard Business Review article write.

Still, it is best to view GenAI outputs always with the caveat in mind that it is, “It can catalyze divergent thinking from the designer or the engineer but it, in and of itself, is not innovative,’ says Bryce Booth, associate director of industrial design for McKinsey & Company.

3. To elevate product engineering.

AI is accelerating engineering, reducing costs, and enhancing the efficiency of the product engineering process, consultancy UST notes. This is especially useful when it comes to innovating current product lines. For example, GenAI tools can be set to incorporate user feedback and manufacturing cost in modeling and to design more varied and specific tests.

Product ideas from airplanes to consumer-packaged goods

Since the debut of GenAI at the end of 2022, improvements—especially the advent of image-generation tools like DALL-E and Midjourney—have expanded the use of these tools.

Here are some prominent examples of GenAI in action:

• Manufacturers like BMW and Autodesk, which worked with Airbus, are using GenAI to design transportation experiences of the future. Autodesk used GenAI tools to develop the Airbus Concept Plane, envisioning the jetliner of 2050, which will feature tailored flight experiences with reduced environmental impact, according to the Master of Code consultancy.

• Bentley Systems, which makes construction-industry software, is developing generative AI-powered tools for creating drawings, such as site plans, and systems that could propose infrastructure designs with better climate resiliency. This functionality will be included in future versions of its software for clients to use, The Wall Street Journal reported.

• Consumer packaged goods companies are among the most enthusiastic users of GenAI for product ideation, leveraging both image-based and text tools. Nestlé is at the forefront of product innovation in the consumer packaged goods (CPG) sector, using a proprietary GenAI tool to validate new product ideas—another way to use GenAI in developing ideas. “In our early efforts, we have seen the tool accelerate the production ideation process from six months to six weeks,” Veeral Shah, chief digital and e-commerce officer, writes. The company is reportedly using a generative AI platform called Tastewise, which provides insights into restaurant items, home cooking trends, and real-life consumption moments, Consumer Goods Tech reports.

Four ways GenAI can help redesign products

When product innovation is on the agenda, it pays to keep GenAI’s natural strengths in mind. LLMs excel at everything from summarizing content to answering questions, researching topics, translating languages, creating images; personalizing responses, and coding. All of these actions can play a role in product innovation—when you use four basic best practices:

1. Start with a customer problem.

Make sure you start with a customer problem you are trying to solve, not just the “something new,” advises Andrew Binns, director at consultancy Change Logic and co-author of Corporate Explorer: How Corporations Beat Startups at the Innovation Game. “You have to be careful with how well you frame the use case. With any newly emerging technology, the temptation to go be a solution looking for a problem is very powerful. When you're under pressure to demonstrate action, and it's really about framing a use case that is about a customer problem.”

For some companies, he adds, the use case will naturally present a good fit for GenAI, as in the case of Elsevier’s Scopus search tool, which gives a GenAI front end to a massive volume of scientific papers. For others, the use case may not be immediately apparent. Concentrate on what customers truly value, advises Binns. People tend to think they need an “idea funnel” from which to winnow down useful concepts. But what they really need is a “customer problem funnel” that can be used as a front end to GenAI queries. Look for places where you know there are currently inadequate answers and inadequate solutions available, he adds.

2. Give prompt engineering due attention.

Customer service chatbots are one of the most obvious first places GenAI has been used. For example, Italian small-appliances manufacturer De'Longhi is using GenAI to manage 1.5 million customer interactions annually while perfecting the customer experience by providing a single source of customer information. Seeking innovations in that area means the prompts used will be crucial, says Binns. “You need to try to consider the different eventualities where you might want to apply it. What will consumers say to you in different situations? How will they respond? These are situations where setting up the right prompt with an understanding of the limits of what it can and cannot do is essential.”

Getting the prompts right is part art, part science, and makes all the difference in the quality of output you derive. In some contexts, your team can use prompts like “Tree of Thought” or “panel discussion” to get GenAI to challenge, evaluate, and strengthen its own new product ideas. Basically, this involves prompting the AI to do a step-by-step evaluation, from a variety of perspectives, refining its answer as it works through each step, as Vellum points out. This approach fits well with situations that require complex reasoning and multivariable analysis.

Keep your good ideas to yourself

It pays to be wary about conducting product and design innovation projects through a public GenAI platform because many of them use prompt inputs to further train the model. In theory, this means the idea developed in one session could then leak out to another random user.

If your organization is serious about using GenAI for product innovation, investing in a proprietary platform is a best practice, argues McKinsey’s Matt Banholzer, a McKinsey consultant, in a March 2024 podcast. He says that innovation leaders build their proprietary data sets. “We have GenAI tools that are wired into some of our proprietary databases on company performances, market size, and so on, so the answers are synthesized in the right way, and we can sift through data that others don’t have.” That keeps ideas relevant—and your own.

3. Rely on people to make the qualitative judgments.

The current versions of ChatGPT and its ilk are superb at summarizing content. But that doesn’t mean they know which elements are the most useful to real-life customers.

Binns warns you’ll need to keep real-world customer needs top of mind going into the experiment. “ChatGPT can help us synthesize the results of a number of interviews, for example. You record the interviews, and then you can get a summary of those [conversations]. That's pretty helpful. But it's a generic summary, it doesn't necessarily tell you what's most important.” On matters of qualitative judgment, you need a human to be in the loop. “It does not have the ability to distinguish between the quality of different ideas, because it is purely descriptive. It is not actually intelligent, right?” says Binns.

4. Consider your sources.

If somebody brings a “revolutionary new product idea” from ChatGPT 3.5, your antenna should go up. Anybody could have generated the same thing. If the idea comes from a gated GenAI that works with your own proprietary data, it has a higher chance of being truly new. If you want people to create strong ideas, give them the strongest tools.

Generative AI has ushered in a whole new world of product development. Tools like ChatGPT can help you unlock the secrets of customer desire—and then reap the rewards of catering to those desires. But it’s not as easy as just firing up the latest GenAI and throwing out a few prompts. A methodical approach is key.

Read more