Did we forget to think commercially about AI?
Everyone is talking about AI. But is AI universally delivering value? We know that the largest tech companies such as Amazon and Facebook, are regularly announcing new AI projects that are “revolutionising” their businesses, while AI is being pitched to add $13 trillion to the world economy – an increase of 16% of global output by 2030 (). Yet, there is very little evidence that many of the organisations that have invested in AI projects to date are actually seeing any measurable benefits from them. In fact, there’s a wide spectrum of “AI maturity” as reported in a conducted by ElementAI. Most organisations get stuck at the exploring and experimenting stage, never able to deploy AI into production systems.
There are many reasons for this, including technical, as reported in a recent paper I’ve co-authored on the “” with researchers at Cambridge University. However, the most fundamental is that few organisations, to put it bluntly, think about how the project will make, or at the very least, save the organisation money.
Unlike other serious investment decisions, investing in AI projects seems to be undertaken without a thought for how this will be used commercially. The result is typically a proof of concept that never makes it to deployment in front of real users and stakeholders, which, at best, is a publicity stunt for the corporation. At worst, it is a drain on technology investment that could have been better spent on upskilling the workforce. However, if done properly, AI projects implemented for the right reasons and in the right environment can have a positive impact on a business’ bottom line.
Using AI to drive actionable insights or predict business trends is the most well-known use of the technology. However, it is still rare for a business to proactively feed these insights into their decision-making process. This means the initial investment into AI is not realising its full commercial potential or delivering hard business benefits.
A by the research house Gartner, found that until 2022, only 20 percent of analytic insights will ever actually deliver business outcomes. Using AI to process large volumes of data and provide insights can help businesses to guide strategic decisions but if these insights are not used or were not designed to influence decision making, then what is the point?
You can think of insights generation at the macro ambitious level but also at the micro down-to-earth level. For example, at the macro level, a new initiative with Microsoft, which will involve building a cloud-based platform that will leverage data analytics to monitor and optimise field performance. At the micro level, a single employee can develop a summarised dashboard to save everyone time and effort by providing access to the information for all - as exemplified at GSK with . This idea of reducing time to make informed decisions is important because it could allow organisations to take advantage of market price ahead of competition or get recommendations to key stakeholders in a timely manner so it can be acted upon. All of these benefits will have a direct impact on business performance.
Operational efficiency is something that all businesses want to achieve. You can think of two categories for driving efficiency:
- OPEX efficiency: reducing operational expenditure by exposing more efficient ways of operating, reducing costs or improving scale
- CAPEX efficiency: improving impact of capital projects by exposing strategic efficiency such as number of assets required
Saving money through automation or by reducing error rates, and increasing productivity at scale are solid ways to ensure that a company survives and thrives, even through testing times. This might be a less sexy use of AI but it is the best way to use it to make the business money or, perhaps more accurately, save money.
Within manufacturing and logistics, you tend to have specialised equipment that often requires planned preemptive servicing. Where historical data is available from vibration sensors or historic maintenance records, predictive maintenance strategies can be developed to keep key machinery in a better state of health and thereby save maintenance costs.
Developing the skills in the workforce to be able to implement these sorts of processes is key to the successful levelling up of UK businesses. We are seeing an increasing number of job postings for these sorts of roles with demand for data scientists and data engineers tripling over the past five years. Upskilling existing workers and educating children in schools including in programming languages like Python will be key to filling this demand from businesses.
One less understood commercial application of AI, is that it can help protect revenue for the business. This can be achieved in a number of ways including fraud prevention, supporting compliance and churn modelling.
Although the adoption of AI and big data projects is often associated with data privacy concerns, AI can really help businesses to meet regulation and compliance requirements by enabling corporates to better manage data as well as detecting potentially fraudulent activities.
By having algorithms that can automatically clean data sets, reduce the burden of data maintenance, and detect fraud, companies can better protect customer data and other sensitive information against would-be hackers. This could save a business from ruin in a world where a data breach can result in millions of pounds worth of fines that can cripple a business as we saw with the £20 million fine given to for failing to protect the personal and financial details of more than 400,000 of its customers. AI not only prevents these sorts of breaches but also streamlines the management of the data, saving the business money as well.
We are seeing businesses begin to adopt policies that allow the deployment of AI, but educating key decision makers is still a task that the industry must continue to undertake to drive wider adoption of the technology. By acting on insights that AI can pull from big data sets, improving operational efficiencies and protecting the businesses with fraud prevention, supporting compliance and churn modelling, AI can be used to improve revenues and decrease costs rather than simply being namechecked.
By Dr Raoul-Gabriel Urma, CEO and founder of Cambridge Spark