Ignoring artificial intelligence is no longer an option for many businesses, and yet many leaders are struggling to find their footing with AI. A quarter of UK business leaders cannot explain AI-generated outputs to stakeholders, according to a new survey by Startup.co.uk. What’s more, only 22% said they could explain their AI-generated outputs easily, and 12% of them stated that they could “but with a lot of difficulty.” Not only is AI changing how we work, but also impacting what can be achieved with the work being done right now. Ignoring these AI adoption challenges is certain to complicate business prospects and the future of the organization overall.
It isn’t an exaggeration to say that AI is changing how we operate today, but the full scope of the challenges that accompany workplace AI implementation isn’t frequently discussed. Startups.co.uk’s latest Startups AI Paradox Report surveyed 404 UK businesses that have been in operation for 10 years or less to understand the relationship between AI and leadership today and bring to light some of the challenges introduced by the technology. While the data gives us a closer look at struggles among leaders across startups, the results introduce a new angle to the universal problems that require resolving.
New data shows that 25% of business leaders across EU startups struggle to explain the results of AI-generated outputs to stakeholders. (Image: Pexels)
The Problems Start From the Top: A Quarter of Business Leaders Struggle to Explain AI Output to Stakeholders
Once upon a time, complete familiarity with data was a mandatory requirement for operating and making business decisions with that information. In the modern world, getting the work done quickly often takes precedence over getting the work done well. Around 25% of business leaders struggle with explaining the logic behind AI-generated outputs to stakeholders, and this likely only includes the share that is willing to admit to such a faux pas.
We have already seen evidence of AI-generated errors making it into official documentation and workplace submissions. Much of it is a result of users becoming complacent and failing to fact-check information, either due to the additional work that it requires or because AI implementation demands that work move faster. Separately, there is also an apparent lack of understanding of the nature of AI usage. Tools are haphazardly employed to seek results, and without the step-by-step process of getting to those results, users miss out on learning what the data has to say.
Organizations now rely on AI to write the code for their websites, but don’t maintain someone with the expertise to understand what has been constructed. Many AI chatbots handle sales calls and customer interactions, and they make bold offers that don’t exist. Fixing these errors becomes difficult because the groundwork is done by AI and then reviewed by AI. Similar errors eventually continue to circulate within the AI systems, causing problems to unwittingly expand. This is one of the biggest problems with how organizations are adapting the tech today, rushing towards capitalizing on AI without a thorough understanding of its limitations.
AI Adoption Challenges Extend to a Lack of Trust in These AI Tools
Most organizations that are now working with AI are doing so with the understanding that they just need to take the leap and the rest can be sorted out later. There is a very real fear of AI that is circulating in the world today, but many continue to experiment with the tech regardless of their concerns. About 37% of customers don’t trust companies that use AI for processing their personal and financial data. As it turns out, business leaders struggle with trusting AI as well. The survey found that 90% of founders who use AI are fearful of their data being ingested for training public AI models, and yet 50% have still fed their data into free, publicly available tools over the last 30 days.
Some leaders who are particularly cautious about their AI implementation strategy do have some measures in place. About 56% admit to manually scrubbing data before running a prompt, which wastes a considerable amount of their time. As the report explains, “If your data cleaning time matches or exceeds the AI output time (as 23% of founders report), your net productivity gain is exactly zero.”
There are workarounds to this, and it all starts with governing AI usage strictly and investing in closed-loop AI systems that allow for a locked ecosystem of AI interactions that don’t exit into the public domain. This isn’t a foolproof solution either, as it requires a degree of trust in the AI provider and the security measures set in place, but it does beat relying on publicly available AI tools if the use of these tools is non-negotiable.
AI Implementation Requires Planning and Greater Ownership of Its Implications
Using AI for data evaluation or decision-making does not absolve an organization or its business leaders of the responsibility for that data or decision. While the tool might furnish incorrect or incomprehensible results, the quality and accuracy still reflect on the organization’s reputation as a whole. Beyond that, employers could also be held legally accountable for the content that is generated, even if they use a third-party AI platform to do so. One of the most important business challenges with AI highlighted in the Startups.co.uk report is the emphasis on regulations.
The EU Artificial Intelligence Act applies to any business with even a single client in the EU, and it requires employers, not the ones that developed the AI tool, to take legal responsibility for the safety and proper use of AI systems. Similarly, the UK General Data Protection Regulation allows individuals the freedom to enquire about how their data is being used, and automated decision-making systems are no exception. Without internal understanding of how data is being used and protected, AI transparency can be hard to achieve.
While AI regulation and governance in the U.S. has been more lenient in recent years, businesses are being forced to recognize that they maintain responsibility for what happens with the tools they use. Waiting for cases to set legal precedents and laws to formalize safety and compliance around AI usage isn’t a sustainable strategy, and organizations need to get ahead of AI implementation by seeking training, awareness, and expertise on how best to navigate the technology before it is put to use. From auditing internal AI tools and practices to monitoring their application from start to end, there is no time like the present to get started with better control of workplace AI usage.
Subscribe to The HR Digest for more insights on workplace trends, layoffs, and what to expect with the advent of AI.
