If 2023 was a year of experimentation for generative AI, 2024 is the year the technology has become a firm fixture in employees’ lives.
Today, 75% of knowledge workers already use AI, which is set to rise to 90% in the near future. The surprising thing is that more than 50% of this group are using personal or otherwise non-company issued tools. More surprising still is that half of these employees are so attached to such tools that, even if their company banned their use, they would still continue using them.
This is the Shadow AI phenomenon, where employees source and use AI tools or applications to complete tasks for their employer without formal approval or IT oversight. Battling burnout, skills shortages, and reduced head counts, staff have turned to GenAI as an invaluable tool to boost productivity and tick off their day-to-day responsibilities. However, Shadow AI means that businesses have little visibility or control of the AI applications used in-house, leading to cyber-security risks, skills gaps, and inaccurate work.
So, if Shadow AI isn’t going away, what can companies do about it?
We surveyed 6,000 knowledge workers in the US, UK, and Germany to bring attention to Shadow AI and understand the root of this trend. Let’s dive into how Shadow AI presents itself in the workplace, what staff need from employers to use AI more responsibly, and how businesses can adapt their GenAI strategies to suit.
A New Defiant Mindset
AI isn’t just emerging, it has arrived and is appealing directly to the masses. With the genie completely out of the bottle, Shadow AI presents an evolving challenge for businesses. On the one hand, they want every employee to be as productive as possible. On the other, they face a myriad of risks associated with colleagues using such tools. So, why are knowledge workers continuing to use AI in day-to-day work, whether it is sanctioned or not?
Mainly, it’s a productivity aid. The majority of knowledge workers state that they use Shadow AI to save time (83%), make their job easier (81%), and get more done (71%). Day-to-day efficiency gains aside, almost half of these employees believe that AI tools will help them to be promoted faster (47%). Meanwhile, two-fifths think FOMO will be a factor, saying those not using AI are missing out (40%). All of this points to a future where AI use is more common, so businesses need a plan for how they’re going to channel it.
The divergence of business and employee vision
Generative AI is seen as a new bolster of productivity; it’s marketed directly to consumers; and more people in the workforce are tech-savvy enough that they’re willing to give it a try. It’s little wonder that Shadow AI is a burgeoning trend that organizations have little means of stopping.
For one thing, 53% of people say they prefer their own choice and independence in deciding which tools to use. Perhaps it can be reduced, though, with the one third of employees (33%) who go out on their own because their business doesn’t offer them the tools they need.
Whatever the background or motivation, the risks are the same. At a strategic level the divergence in approaches can lead to a fragmentation of key processes. This hurts efficiency, compliance and consistency of key services, as well as potentially denting digital transformation objectives.
The danger of the occasional user
There are also more tangible and immediate risks, although the good news is that most knowledge workers are wise to them.. Most recognise the risks around cyber-security (72%), data governance (70%), and inaccurate information (67%).
The bad news is that only 27% run security scans, and just 29% check data usage policies, before taking the AI plunge. Meanwhile, one-third use AI to automate tasks (34%), running the risk of producing inaccurate information.
Within these numbers there are some correlations between use and responsibility: those who use AI the most take more precautions. This should be re-assuring, but it’s the group of employees who use AI tools only occasionally whopose the real danger. These people are generally aware of the risks, but less likely to do something about it, and equally as likely to go rogue and use the AI tools that they want to use.
Everyday users are 50% more likely to check data usage policies and 65% more likely to run security scans than occasional users. While everyday users can be more trusted to pick their own tools and not pose too great of a threat, the occasional users are where businesses could be opening themselves up to the biggest risk.
Under the right conditions, including better training and more suitable tools, 90% of knowledge workers will start to use AI more. It’s likely that this increase will largely be occasional users, only worsening the risks that companies face. The die is cast for businesses with Shadow AI…the only thing to think about it what to do next.
Empowering the individual and de-risking the occasional user
Shadow AI needs to be embraced and channelled. As more people use AI, organizations need to create processes that allow for its inclusion, without exerting excessive control over specific tools that results in more off-the-books usage.
Full transparency of processes is a key building block in this regard. Even if a tool isn’t officially sanctioned by the organization, when its use can be integrated into key processes, then data use, compliance and efficiency can all be monitored and optimized.
Training is the other key element. If people are going to use their own AI, they need the skillset to use it responsibly. However, 35% of knowledge workers say their employers offer no AI training , while a further 39% say there is training, but it is very basic and/or does not cover big risk issues. Better training would make 46% of employees use AI more, but crucially, they would use it effectively and responsibly.
Better choice could also help to mitigate some Shadow AI risks. One third of those surveyed say their employer does not offer the tools they want. If those organizations consult more with individuals, perhaps more helpful tools can be officially sanctioned, thus reducing some of the shadows in the Shadow AI landscape. It could help boost overall productivity as well, because 44% would use AI more if they had the right tools.
Harnessing turbocharged ‘operational chaos’
‘Operational chaos’ affects the majority of organisations, as surging volumes of new technology create new ways of working . Unless companies can establish a transparent framework for processes, they can lose control of what’s going on in their organization day-to-day. Shadow AI will supercharge this operational chaos if the right steps aren’t taken.
When leaders boost the visibility of all processes, are flexible to incorporate alternative ways of working, and can effectively integrate relevant AI solutions into employee digital toolkits, they can harness this chaos for good. Not only can this power a better flow of process intelligence, but also unlock insights into efficiency gains.
The actions that companies take now will determine if the unmanageable can be brought under control. AI can be a blocker or an enabler and it will certainly play a huge role in the day-to-day operations of a business. The question is, will that impact be positive?
6,000 knowledge workers (defined as those who primarily work at a desk or computer) in the US (2,000), UK (2,000), and Germany (2,000) were surveyed between September 13-25, 2024. All respondents were 18 or older, and the sample was census-balanced by age. The survey respondents were independently sourced from RepData and the research was conducted by TEAM LEWIS.
AI
Artificial intelligence (AI) is the simulation of human intelligence in machines built to perform human-like tasks such as language translation, visual perception, logic-based techniques and decision making. AI technologies include machine learning, natural language processing, machine vision and speech recognition.
GenAI
Generative AI (GenAI) is a form of artificial intelligence technology. It automatically generates content including text, imagery, video and audio in response to written prompts. GenAI utilizes Large Language Models (LLMs), which are trained on vast amounts of online content including social media channels, webpages and academic papers. The technology has surged as a result of new user-friendly interfaces.
Shadow AI
Shadow AI is a term used to describe the use of unsanctioned AI and GenAI technology outside of the control or ownership of an organization’s IT governance.
