Artificial Intelligence (AI) is the way of the future, but the right foundations have to be in place for an enterprise AI strategy to succeed.
By Chris Pallikarides, GM of ITBusiness
Machine learning (ML) and AI are certainly the way of the future, offering enterprises faster, more accurate and more efficient way of automating processes than have ever before.
Globally, Gartner reports that AI adoption in organisations has tripled over the past year, with 37% of organisations having deployed AI – or about to do so. By 2021, Gartner expects AI augmentation to create $2,9-trillion in business value and 6,2-billion hours of worker productivity globally.
In South Africa, ML and AI have been talking points for many years. However, the practical implementation and application of these technologies has not quite caught up with the rest of the world. Mid-size and large local enterprises are looking to ML and AI to streamline operations, support strategic and personnel planning and gain insights such as whether certain products are performing.
While many companies are talking the talk, they often seem to forget the fundamentals – crucially: knowing what data they have and where it is sitting. Without the data engineering piece of the puzzle in place, Machine Learning and AI cannot deliver on expectations.
Data quality is a big issue in many South African organisations, and most of them are aware of the problem. Amid exponential growth in data volumes, companies have lost control over data sources and standards; they lack effective data governance and stringent controls. On top of this, while most want to optimize their data use, possibly even monetising it, many have not formulated a clear strategy for doing so.
Therefore, while ML and AI should definitely be on the roadmap for every South African organisation, attention has to be given to the fundamentals – including strategic planning, data quality and data governance first.
Addressing underlying issues could take as little as a few weeks, or up to several years, depending on budget, the size of the business, the amount of data involved, the technologies in use and the skills available. A data governance exercise alone could take 36 months to implement. But these are necessary processes. Embarking on an AI project before addressing underlying data quality issues could result in flawed outputs, unexpected additional costs or delays in project delivery.
In addition to assuring data quality, organisations need a clear vision on how they intend to use their data in future. If, for example, they hope to monetise their data, then in early planning, they might work backwards, considering the type of data they have, its potential value, and models for monetising it, while also taking into account the regulations around data privacy.
Once organisations have clear roadmaps, all the necessary data and they know they can trust it, making strategic decisions become a much easier task; and once the right people, processes and technologies are in place, the whole discussion becomes streamlined.