Over the summer, I had the pleasure of being involved in a major SAS study on enterprise readiness for artificial intelligence (AI). The study involved in-depth interviews between SAS consultants and 100 C-level senior executives from organisations across the EMEA region. It was designed to explore their understanding of AI, its implications and issues, the challenges faced by their organisations, and their platform and team readiness to adopt AI.

Having conducted several interviews myself, it is fascinating to see the aggregated results in the report and how the people I have spoken to relate to the whole population. As part of the analytics platform architecture team, I was particularly interested to see the findings on readiness.

Discussing platform readiness

Creating architecture to support AI means having a platform in place to support all steps of the analytics lifecycle. In other words, being able to take full advantage of the adoption of AI will require organisations to have the right platform in place, as well as staff with the necessary skills to handle that platform. The questions about readiness in the study explored this issue further, and the responses were revealing.

We found that there was a wide gap between those who felt that their platform was either ready for AI, or well on the way to being prepared, and those who were not ready at all. Only a small number, just 12%, felt that they were absolutely ready to go. The largest group of respondents, 29%, said unreservedly that their platform capability was not ready, or that they had no specific tool or platform for AI. In between were two groups, each of around one quarter of respondents. The first felt that they were handling data management and algorithm processing well, and the second felt that they needed to adapt their current platform to make it more fit for purpose.

Some of those who were not ready made the point that they did not, at present, have any need for this type of platform. They had therefore, quite rightly, not made any decisions about what they would need, to avoid being tied to something unsuitable. Others, however, were already thinking about it, and many were looking at cloud-based solutions. The financial organisations in particular were avoiding cloud because of security issues, but many recognised that they might need to move to the cloud in future. Quite a number of respondents cited the additional scalability and flexibility of cloud-based solutions, which they thought would be important as they started using AI.

One of the companies I spoke to had been running several data experiments in a separate lab environment and these helped them a lot to get a better idea of what they really need in their production platform, both from a functional capability as from a capacity sizing and scalability perspective.

Strategies for team readiness

This emphasis on flexibility and scalability—and therefore cloud solutions—was echoed in some of the responses on team readiness. Over 20% of our respondents were considering the use of consultants, or drawing on staff in partner organisations, rather than recruiting or training up their own data science teams. They saw this as a way to build or expand their data science skills more flexibly than would be possible through recruitment.

This is, of course, dwarfed by the number of organisations that were planning to either train up their teams to acquire the necessary skills, or recruit new and skilled staff. But one fifth is nonetheless a significant number. Flexibility as an organisational strategy is linked to agility—being able to scale activity up and down at speed is crucial to responding effectively to rapid changes in the environment, and perhaps, to keeping up with the competition. It has the huge advantage that it does not require a huge upfront investment, although it can often end up costing more in the longer term.

Small organisations often opt for this kind of flexibility, for two main reasons. The first is upfront cost, but the second is recognition of their own limitations. Data scientists are a scarce commodity. They can command high wages, and choose where they work. Smaller organisations, with fewer opportunities, are less attractive employers than large consultancies, with their teams and communities of analysts and exciting opportunities for short- or longer-term projects. Using consultants is, therefore, a pragmatic response to a real-world problem that may be impossible to solve in any other way.

Among the respondents I interviewed myself there were people both from a large corporation and from a quite small very much business focused organisation, and it is interesting to see how their responses were very much in line with these results. The smaller organisation told me that the consultants were there to help them getting up and running quicker but they prefer taking over with own resources where possible on the longer term.

In the same way, cloud solutions allow organisations to buy a ready-made solution, without having either upfront investment costs, or the need to develop the expertise in building. One of the organisations I spoke to additionally saw the advantage of being up and running almost instantaneously as an additional benefit. The rise of options like analytics-as-a-service shows that many organisations see this as an important element in their strategy. That being so, it should probably not come as a surprise to see it being applied to the preparations to take advantage of AI, too.