The rise of AI over the past 18 months has seen huge shifts in the way people across every industry work. Sometimes, these shifts have been driven by innovative new technology being adopted by businesses or even organisations proactively developing their own AI tools. The truth is, though, that many employees have incorporated AI into their arsenal of work tools regardless of company policies, putting sensitive data at risk. With governance practices still playing catch-up, should businesses start looking to a novel solution to shadow AI in data marketplaces?
Shadow AI is, put simply, a modern disaster waiting to happen for businesses. Unsanctioned use of AI in the workplace is putting at risk not just company data and proprietary systems, but also relationships with clients and trust from consumers. Recent research suggests that 74% of ChatGPT usage at work is through noncorporate accounts, 94% of Google Gemini usage is through noncorporate accounts, and 96% for Bard - in other words, unauthorised use of these AI systems is leading to huge amounts of potentially sensitive corporate data being eaten up due to employees who are inadvertently feeding HR data, proprietary source code or any other sensitive corporate information into AI tools that the business simply doesn't know they are using.
In some ways, the rise of shadow AI could be seen as an inevitable consequence of employees recognising the productivity benefits of AI models, combined with a desire to get the most out of the data they have available to them. As data and AI becomes more mainstream, non-technical employees develop their understanding of the power of modern digital tools. Their understanding of the governance risks of using essentially ungoverned AI models, though, is still not enough to recognise the danger that shadow AI poses.
Overcoming governance challenges
Whilst corporate policies and education form a fundamental part of good data governance within an organisation, it is also true that businesses should look to nurture their employees' curiosity about data products and AI platforms. How can they best do that in a controlled and governed way?
Many frustrations around sanctioned, governed alternatives to shadow AI revolve around the difficulty faced by business users when trying to find and deploy data products in their roles. Many organisations have embraced the revolution in data-driven business by investing in data catalogs to support their data engineers as they build products with the vast quantities of data their business possesses - but whilst these are great tools for data engineers, they don't help to solve many of the challenges around discoverability and useability of data products required by business users.
Data Marketplaces
When talking about marketplaces, many data professionals tend to think of selling data externally, but internal data marketplaces are invaluable tools for a business to make the most of their data assets. These internal marketplaces are platforms to make data products more readily available to data consumers. They help to deliver data-driven initiatives and satisfy employee curiosity about data all in an accessible and, importantly, governable way.
When we talk about data marketplaces, ensure that data is not just discoverable, but that it is safely provisioned too. This data product delivery refers to a worflows which enables consumers to use data platforms or BI tools, a feature missing with traditional data catalogs. Without this proper provisioning, data catalogs can be a bit like using the App store to locate the perfect app, only to find that instead of being able to download it, you have to raise a ticket with the app creator and wait a week to use it.
Note the key difference here between a catalog and a marketplace is that they serve two distinct functions: catalogs are great tools for data engineers building your data products, whilst marketplaces are the best way to empower data consumers with those data products that they need.
Empowering and governing
Data marketplaces are great in two ways - they free up data engineers to focus on data product creation and innovation instead of having to constantly respond to challenges from business users. They also empower business users to find and use the data products they need in a controlled and governed way, helping to get the most out of a business' data, satisfying employee curiosity about new data and AI tools, and most importantly ensuring that the tools and products being used are safe and secure.
Shadow AI should be a top concern amongst data and compliance professionals at modern businesses. Whilst curiosity amongst employees about how best to use data and AI tools to improve their productivity should be encouraged, failing to do so in a controlled environment can put sensitive data at risk and cause major security and privacy challenges. By investing in a good data marketplaces, businesses can achieve the best of both worlds: protecting data and shielding against shadow AI, whilst empowering employees with secure, governed and accessible data products.