IDEO Executive Director Sergio Fregoni on how to get the most from AI in business

With AI hype continuing to dominate the business agenda, there are still questions around the technology’s greatest promises. Consultancy.UK spoke to Sergio Fregoni, executive director at IDEO, about the material benefits the technology can offer, and how CEOs can ensure the best results.

After two years of hype, there have been some big expectations of AI and GenAI – and few seem to have been met. Where are the largest shortcomings – and can AI firms get past them?  

We have to remember that AI is not a new technology. It has worked its way into almost every industry over the last 10-15 years. From the content we consume at home to the way we work and the transport which gets us between the two, AI is everywhere.

The first Generative AI, the ELIZA chatbot, is dated 1961 and since then the growth and research have been nothing but astonishing. GANs (Generative Adversarial Networks) were introduced in 2014 and, already today, their widespread use helps create data sets that improve diagnosis in medical imagery or increase the resolution & quality of our smartphone cameras.

The thing with AI is that it’s a catch-all word, a black box that people like to use to describe something that looks a bit like magic but no one understands properly. And once an actual application is found, we stop calling it AI, but rather things like Computer Vision, Recommendation Engine, or self-driving cars.

Like many technologies, Generative AI has its limitations; current models hallucinate at times, meaning they can say something spectacularly wrong with amazing confidence. But if we understand the technology, the explanation is actually quite simple. All they do is try to statistically predict what word (or pixel) should follow another, without actually understanding the meaning of the final output. Explained in this way, it sounds crazy that we trust this technology so much, but the results can be amazing when trained and used well.

Like any technology, focusing on the shortcoming won’t get us far; at the end of the day, every technology creates opportunities while having limitations. Instead, we can flip the question and defer to the human expertise and intentional design that determine AI’s most productive uses. The solution lies in embedding AI not as a replacement, but as a tool to deepen human insight. When we approach AI as a support for human creativity, decision-making, or problem-solving, we move past the limitations and open up real, transformative applications.

How are CEOs currently implementing AI for innovation?

Almost every CEO is using AI to rethink the way their business works, but not always with innovation in mind. Most leaders are still exploring AI as a tool to automate and streamline what they have always done. This can yield good incremental progress and cut costs but it’s not truly innovation – it’s the baseline investment required to keep up.  

Using AI to help drive growth and innovation, however, is a very different type of challenge. It requires CEOs to invite AIs into the boardroom as collaborators. Collaborators who can cover their blind spots and surface the needs of customers, societies and supply chains which they may have never considered before. It is ultimately a tool to extend CEOs’ capacity to understand a greater range of perspectives and see a greater range of opportunities.

We conducted research this year which found that UK businesses using AI in any capacity saw a 2.4% increase in growth and 0.24% reduction in costs as a result. And, those businesses using AI consistently reported benefits across customer experience, efficiency and product development.

But what is far more significant is that businesses who use AI to intentionally drive innovation see 38% larger impact on growth than businesses who use AI for any other purpose. Leaders need to learn how to embrace AI to create fresh ideas and innovative business models, not just ever more efficient versions of the businesses which came before.

Are there any case studies you could point to? What material benefits have they encountered – and how did IDEO help them?

A good example of this in practice is our work with H&M Group. H&M was seeking to tackle its overstock issues to minimise the volume of excess inventory that ended up in landfills.

We used machine learning to incorporate perspectives from multiple departments and functions of the organisation, and created an algorithm which could better predict demand and shorten the interval between sales and production. As a result, H&M improved sales of specific lines by a third while reducing stock by a fifth – an undeniably positive outcome enabled by human insights augmented by data and AI.

Another great example may be a more recent project we run in collaboration with an edtech startup. Lots of schools are reacting to AI by banning it from the rooms, not realising or questioning how the learning experience could be different when redesigned with AI. IDEO helped them create an AI-driven tool to support secondary students and teachers in enhancing writing skills, rather than replacing them. We crafted a user-friendly platform that allows busy tutors to provide more personalised feedback to students, allowing teachers to better engage their classes and monitor progress, contributing to a notable improvement in writing competency. The platform has since scaled, demonstrating AI’s potential to make educational support both more accessible and impactful.

Why must the sources used to train AIs be broad to ensure best results?

When trained on the right, diverse data sets, AI helps us to access more divergent and complex viewpoints. It can empathise on a scale that is humanly impossible. We will need to include the broadest possible set of perspectives when training AI to ensure marginalised voices are not excluded. This is essential so that the future, not least on climate action, is equitable and just.

How can AI help with the sustainable transition?

The barriers to net zero are no longer primarily technological, but an issue of behaviour. There is no shortage of viable climate solutions, what we must focus on now is driving their uptake by businesses and consumers. Ultimately, we need to make sustainable behaviours more desirable across entire circular supply chains.

AI can be used to consider a far greater range of human factors which dictate the adoption of sustainable solutions. AI’s ability to synthesise many competing perspectives can help leaders to understand the people in their value chains on a much larger scale.

One of my favourite examples here is in the use of satellite imagery combined with AI. Several organisations are using computer vision with increasingly sophisticated models built to detect deforestation, predict wildfires, or recommend the right amount of fertiliser and irrigation to farmers in water-stressed areas. Digital tools, often with very simple interfaces like text messages, can help individuals combine their local knowledge with global interconnected insights.

Currently the technology uses huge amounts of energy, and has regularly been said to have an alarming carbon footprint. Is it really worth it for the analysis it can offer up?

AI’s energy and water consumption is significant, and we clearly have to match investment in new AI data centres with investment in renewable energy.

However, it is hard to imagine us achieving the change that a circular economy requires without the help of AI. We need to redesign products, business models and entire supply chains for a world which aligns commercial and environmental objectives. CEOs who are used to thinking about their customers need to consider everyone who could be affected by the manufacture and use of that product. CFOs who are adept at measuring the financial health of their business must also consider the health of the ecosystems they rely on.

We recently undertook research to quantify how collaborating with AI can address divergent and complex sustainability challenges, revealing that AI collaboration can increase innovation in the design of sustainable products and services by 56%. The value of AI’s imaginative abilities in problem-solving for the climate crisis should not be underestimated.