A significant portion of today’s work relies on older systems that current IT professionals do not fully understand. Due to these systems being built 20-30 years ago, the technology can be outdated and modernising can seem a daunting and expensive task. In addition, the logic behind many of these applications is difficult to interpret.
To solve this, companies have turned to leveraging AI, especially generative AI, to help understand their legacy systems better. This includes figuring out how systems work, what they connect to, and what data might be useful for analysis. Looking ahead to 2025, we can expect this trend to gain further traction but successful adoption will continue to pose business challenges.
Unstructured Data and AI
A critical area where AI is making a significant impact is in dealing with unstructured data.
It’s commonly stated that around 80-90 percent of enterprise data is unstructured, and that statistic has been around for years. The challenge with unstructured data is that it’s difficult to store in a way that allows computers to use it effectively. Searching through it manually is tedious and time-consuming.
Despite advancements in search, there isn’t a good way to analyse or make business decisions from this data without human intervention. Only about 18 percent of organisations can leverage unstructured data, creating a significant gap.
Generative AI can help bridge this gap by allowing companies to turn unstructured data into something that can be queried and analysed. For example, asking a question about a legal document without needing to be a lawyer to get a meaningful answer. This capability allows businesses to access and use data in ways that weren’t previously possible.
The power of generative AI
Generative AI can revolutionise the use of unstructured data, opening up many possibilities. It’s about making data easier to access and more actionable.
Consider this: If you wanted to know how many payments you made to a specific vendor over the past three years, you would typically have to go through your bank statements manually. With generative AI, you can ask that question directly, and the AI will provide the answer immediately.
This transformation also impacts fields like technical support. For example, technical reps in the field do not always have immediate answers to complex questions. Generative AI can help by providing answers instantly, reducing the time spent searching through manuals and increasing productivity.
Challenges in AI adoption
Despite the enormous potential of generative AI, most organisations are not fully leveraging it yet. According to a Deloitte study, many businesses are still experimenting with AI but have not put it into production. The primary reasons for this are a lack of talent to optimise models, governance concerns, and data security issues.
Protecting data is a significant challenge when working with unstructured data, particularly with sensitive information like banking statements. It is critical that businesses can protect data throughout the process, which is why focusing on bringing the model to the data, rather than bringing the data to the model is the better strategy. This keeps the data secure, while still allowing the AI to generate valuable insights.
For many companies, the focus is on managing the risks and skills gap associated with AI. By keeping data protected and lowering the skill threshold required to use AI, it will become easier for organisations to take advantage of the full potential of generative AI.
Future Predictions for Generative AI
Looking ahead, it’s likely that the most generative AI adoption will be embedded in tools. This is already happening: many products that are used today, such as conferencing software, Microsoft tools, and GitHub Copilot, are already integrating AI to improve user experience and productivity.
However, AI will increasingly move into products as companies figure out the necessary governance and privacy protections. Vendors who can help businesses overcome the challenges of skills and risk management will be the ones that succeed.
Organisations will look for vendors who can provide easy access to AI, while managing the risks around privacy and model accuracy. Ultimately, companies that can meet these challenges will be the ones that thrive in the coming years.
