AI is rapidly redefining how we work, and IT teams are at the forefront of these changes. Whether it’s providing a first draft of basic code, flagging incidents that need to be responded to or answering simple client questions, AI can make both these teams and the wider business more productive.
For IT professionals, this is creating new responsibilities. The vast majority – 9 in 10 – say AI is forcing them to reevaluate their technology strategy, while 86% say it’s making their job more important.
Go deeper with GlobalData
However, AI can only help us if we’re confident that we can trust it. If the code it is drafting is riddled with bugs, if it fails to monitor the right incidents or provides incorrect answers, that trust will evaporate.
For many IT leaders, this is the core issue that’s holding them back from unlocking the potential of AI. In fact, almost 9 in 10 global leaders working on implementing AI in their organisation say that it’s not possible to know if their company’s AI output is accurate.
These are valid concerns. Hallucinations – which involve an AI making false assumptions – are an inherent challenge to AI, particularly in the advanced large language models (LLMs) that are driving the current wave of innovation. It’s important that these challenges aren’t ignored.
However, it is possible to minimise and mitigate the risks associated with AI, empowering our IT teams to achieve greater productivity while simultaneously prioritising accuracy and safety. The key is making sure you’ve got the right data to feed the AI to begin with.
How well do you really know your competitors?
Access the most comprehensive Company Profiles on the market, powered by GlobalData. Save hours of research. Gain competitive edge.
Thank you!
Your download email will arrive shortly
Not ready to buy yet? Download a free sample
We are confident about the unique quality of our Company Profiles. However, we want you to make the most beneficial decision for your business, so we offer a free sample that you can download by submitting the below form
By GlobalDataIdentifying the challenges with training AI
The internet is a big place. In 2020, the amount of data across our online world was 64 zettabytes, with each zettabyte being made up of one trillion gigabytes. In other words, the internet contains a mind-boggling amount of information – and it’s only growing.
For LLMs that require huge pools of data to learn from, the internet is an essential resource. However, the problem is that the internet isn’t just big – it’s also messy. The data that makes up the internet is riddled with inconsistencies, misinformation, bias and other issues. This is why training AI on it can lead to some significant problems.
While strides are being made by AI providers to mitigate these issues, businesses and their IT teams already have a solution available that can help boost AI accuracy.
Untapping the value of structured and unstructured data
IT teams are the guardians of every company’s most important asset when it comes to AI: their data. With the dawn of generative AI, this information is becoming more valuable than ever.
Traditionally, IT teams have focused on maximising the value of structured data. This is well-organised information with concrete data points. For example, structured data could be a sales history found in a customer relationship management (CRM) platform. Or simply a HR spreadsheet covering employee details.
Whether by integrating apps, upgrading solutions or training teams, IT leaders have been able to help mine this data for value for some time. However, most of the information businesses share isn’t structured – it’s unstructured. And it’s this data that can finally be tapped for value today.
Unstructured data includes all the conversations, communications and content a workplace creates. It’s the daily conversations teams have, the loose notes from a creative brainstorm and the action points following a meeting. It’s unorganised but essential context that gets shared every day.
By implementing AI that draws on both structured and unstructured data, IT teams can help their businesses achieve more. And because this AI draws on existing data from the business, it provides accurate and actionable insights that teams can trust.
Whether it’s recapping updates for a team member that was on holiday, providing intelligent answers to a client or pulling information from other integrated apps, AI in this environment can provide relevant, up-to-the-minute results based on the organisation’s collective knowledge.
All that’s needed to achieve this is the right platform for capturing both unstructured and structured data. The team at digital bank, Inter, is a great example of how this can be done.
Inter’s team wanted to unlock the benefits of automation and AI, while maintaining a strict focus on data safety. By moving to Slack, the IT team has been able to use their unstructured data to build custom apps, including a new AI tool called InterGPT.
Drawing on their existing data, InterGPT can help develop code, calculate ROI for customers, draft marketing copy and much more. With the efficiencies gained, the company saves over £10,000 every month. It’s a clear example of how internal data from a messaging app can be combined with AI to rapidly create new value.
Trusting in AI’s future
How IT teams implement AI will be a critical factor in their success over the years to come. A loose approach that leaves teams relying on public, web-based AI apps, risks corner cutting, mistakes and harm to both productivity and customer experiences.
However, instead of looking outward to those big, broad AI platforms, by looking inward – at the data that already exists in their organisation – IT leaders can hone a sharper AI strategy. Implementing AI that draws on those rich digital insights that make up our workplaces – and which are growing every day – offers a path to a genuine AI revolution that will make work simpler and smarter for everyone.
