There is massive excitement — and some fear — about AI. The buzz is definitely building, as more and more people start talking about it. But although there is enthusiasm about the potential of AI and its building blocks, there is much less discussion about how organisations should start to invest in AI.
Mind the gap…
Few are talking about their experience of AI, even though we know that there are plenty of organisations out there who are already dipping their toes in the water. We decided to fill this gap in our recent study on enterprise readiness for AI. We opted to carry out a series of 100 in-depth interviews. These took place this year, between SAS consultants with data and analytics knowledge, and people who had some experience of implementing AI and could offer lessons for others. Having expert interviewers meant we could probe deeper into details. We also based the questions around actual experience, rather than expectations, to ensure that there would be useful learning emerging from the interviews.
You might be wondering why we decided to do expert interviews, instead of a major questionnaire survey. After all, with a questionnaire survey, you can gather views from far more people. The answer is that questionnaire surveys do have their place, but they are not helpful for understanding wide-ranging and nuanced views, especially where opinion and experience varies so much across organisations. This is, of course, particularly likely to be the case with new and emerging technology such as AI. We therefore felt that we would be able to get closer to understanding individual organisations’ experiences — and especially the lessons that they had learned — by having the freedom of a semi-guided discussion.
Respondents and findings
We wanted respondents from a wide range of sectors and across the full EMEA area. In the end, the majority of more than three quarters of our interviewees were from five sectors: Banking, insurance, manufacturing, professional services, and media and communication. The rest were from industries like retail or travel. This therefore gave us a very reasonable spread of industries to draw on. We also managed a reasonable geographical spread across EMEA (Europe, Middle East, Africa).
We found that most organisations had started to think and talk about AI, and many had use cases either in existence or in the pipeline. Almost a quarter were already using AI and machine learning systems for customer intelligence, and between 10% and 15% were using it for big data infrastructure or retention strategies based on customer value. The most popular type of forthcoming deployment was implementing or improving analytics capacity for automation.
One of our more interesting findings was around business cases. We asked how the business case for AI was defined, and whether it was different from the business case for other investments. Answers to the first were split fairly evenly between being able to identify growth potential, keeping up with the competition, providing cost savings through efficiency and no business case (or no answer). Each of these were cited by around 20% of respondents. Fewer said that the business case was around experimentation, or improving the customer journey. In response to the second, just over 15% said that AI was no different from any other investment. Just slightly fewer said that there was a clear recognition that it was often difficult to quantify ROI from AI investments. Opinion is therefore still divided about the ‘right’ way to approach AI business cases and investment.
We also asked who was driving AI exploitation, and two main models emerged: business units, or a dedicated central team. Almost a third had opted for each of those, while others chose a hybrid of these two, an agile innovation team or AI champions in business units. We expect all these to form the main models in the future.
For additional findings, download the complete survey report.Download an IIA paper: Machine Humanity: How the Machine Learning of Today is Driving the Artificial Intelligence of Tomorrow