If asked why organisations are hyped about Generative AI (GenAI), it’s sometimes easy to answer, “who wouldn’t be?” The attraction of a technology that can potentially answer any query, completely naturally, is clear to organisations that want to boost user experience. And this in turn is leading to an average $6.7 million investment in GenAI in 2023-24.
Yet while GenAI attracts the headlines, Machine Learning (ML) is quietly doing a huge amount of less glamorous, but equally important, work. Whether acting as the bedrock for GenAI or generating predictive insights that support informed, strategic decisions, ML is a vital part of the enterprise toolkit. With this in mind, it’s no wonder that organisations are still investing heavily in AI in general, to the tune of $21.1 million.
The closest thing to a time machine
At its core, machine learning is currently the nearest technology we have to a time machine. By learning from the past to predict the future, it can drive actionable insights that the business can act on with confidence. However, to realise these benefits, organisations need the right approach.
First, they need to be able to measure, monitor and understand any impact on performance, efficiency and competitiveness. To do this, they need to integrate ML into operations and decision-making processes. It also needs to be fed the right data. Data sets must be extensive, so the AI can recognize and learn from patterns, and make accurate predictions. And data needs to be real-time, so that the AI is learning from and acting on the most up-to-date information possible. After all, as most of us know, what we thought was true yesterday, or even five minutes ago, isn’t always true now. It’s this combination of large volumes of real-time data that will give ML the analytical horsepower it needs to forecast demand; predict market trends; give customers unique experiences; or ensure supply chains are as optimised as possible.
For ML to create these contextualised, hyper-personalised insights that inform strategic decisions, the organisation needs the right data strategy in place.
One data strategy to rule them all
A successful strategy is one that combines historical data – with its rich backdrop of information that highlights long-term trends, patterns and outcomes – with real-time data that gives the most up-to-the-minute information. Without this, AI producing inaccurate insights could send enterprises a wild goose chase. At best, they will lose many of the efficiency benefits of AI through having to constantly double-check its conclusions: an issue already affecting 23% of development teams that use GenAI.
What does this strategy look like? It needs to include complete control over where data is stored, who has access and how it is used to minimise the risk of inappropriate use. Also, it needs to enable accessing, sharing and using data with minimal latency so AI can operate in real time. It needs to prevent proprietary data from being shared outside the organisation. And as much as possible it should consolidate database architecture so there is no risk of AI applications accessing – and becoming confused by – multiple versions of data.
This consolidation is key not only to reduce AI hallucinations, but to ensure the underlying architecture is as simple – and so easy to manage and protect – as possible. One way of reducing this complexity and overhead is with a unified data platform that can manage colossal amounts of both structured and unstructured data, and process them at scale.
This isn’t only a matter of eliminating data silos and multiple data stores. The more streamlined the architecture, the more the organisation can concentrate on creating a holistic view of operations, customer behaviours and market opportunities. Much like human employees, the AI can then concentrate its energies on the data itself, becoming more agile and precise.
Forging ahead with machine learning in the GenAI age
A consolidated, unified approach isn’t only a case of improved performance. As the compute and infrastructure demands of AI grow, and commitments to Corporate Social Responsibility and environmental initiatives drive organisations towards greater efficiency, it will be essential to ensuring enterprises can meet their goals.
While GenAI is at the centre of much AI hype, organisations still need to recognise the importance and potential of predictive AI based on machine learning. At its heart, the principles are the same.
Organisations need both in-depth historical information and real-time data to create a strategic asset that aids insightful decision making. Underpinning all of these is a data strategy and platform that helps enterprises adopt AI efficiently, effectively and safely.
Rahul Pradhan, is Vice President of Product and Strategy for database-as-a-service provider Couchbase.
- Data & AI
