Far from being the killer of all creative thought, AI can empower your innovative processes into tomorrow and beyond. AI is transformative, but its impact will depend on the model of innovation that you are using.

AI as an innovation accelerator

Some innovative processes require surveying a dataset so vast that innovation gets stifled. The pharmaceutical industry offers a good case in point. It has experienced a ‘discovery void’ since 1987 – almost 40 years – in the field of antibiotic drug discovery. Working through millions of potential compounds for one that is successful has resulted in a high failure rate that has historically made the entire drug research industry prohibitively expensive.

Today, Hoffman-La Roche has passed mammal testing with the first AI-identified antibiotic and is currently entering human clinical trials. The technology was used to produce a shortlist of candidates among the millions that appeared worthy of further investigation. Any time you are looking for a one in a thousand event, AI can bring competitive advantages in the early stages.

AI as a catalyst for change

When the rate of internal innovation falls behind the rate of external change, companies will get into trouble rapidly. The banking sector has quickly responded to a new, younger generation of customers following the introduction of banking data interoperability laws that have freed up this conservative sector for innovation. This might include omni-channel services, with a choice of webchat, email, human-to-human phone call, video call, or face-to-face service.

The younger generation are also said to be impatient and perceptions abound that putting them through multiple processes to reach a service will lose them quickly. AI can offer personalisation through an analysis of extensive customer records, reducing the steps needed to access a service. 

In fields such as construction or manufacturing, people want fast, efficient results and the power of AI to reduce error can be a gamechanger. This means rethinking your relationship with customers and suppliers from a one-time sale to an ongoing partnership.

If you agree to share performance data on machinery delivered through sensors to a central data store, it’s also possible to track performance. Customers get early identification of problems, reducing downtime when there’s a fault and the supplier gets valuable feedback on the performance of their machine after the product has left the factory.

AI as the solution

As new AI tools leave laboratories and enter offices, a new innovation cycle begins. The generic generative AI large language model platforms, such as Llama or ChatGPT, are now being fine-tuned with domain-specific datasets. This means that they learn what matters in, say, the insurance field, allowing them to come up with more tailored responses.

We can use therefore these machines to help source datasets we don’t have or sift through relevant research. Provided that any staff involved are well trained in critical thinking, they will be able to look through the technology’s suggestions, remove the junk and focus on potential treasures. Using these tools creates efficiencies in the innovation stage and brings in outside voices. After all, knowledge of your business is a valuable thing, but sometimes a fresh idea from an external voice is essential.

We can also redirect old technologies to new problems now that they are more accessible to smaller businesses. Image recognition, for example, has been used at airports to reduce queues for almost a decade and that same technology is now starting to transform farming. Tractor manufacturers have trained their vehicles to scan fields of mixed crops and identify the crops from the weeds. The machines can then deliver a single drop of pesticide or fertiliser to the weeds and crops, as appropriate, when they are just one cm high. The result is far less chemical use, healthier crops and a move away from environmentally ruinous mono-crop farming – all by re-purposing a technology that was working reliably back in 2016.

Recommendations

However you choose to integrate AI in your innovation cycle, make sure you consider a couple of potential problems. It can be hard to wean your staff off legacy systems, so the earlier you involve your team in discussions, the better. It’s also essential to remember that error is almost always unavoidable and you need a critical approach and plan to deal with it, such as robust feedback loops. These problems are no different from human-based innovation, but a machine can only ever show you ‘what’ – you’re going to need a human to answer questions around ‘why’.