It seems like we can hardly go more than a few days at the moment without a new generative AI assistant, often dubbed 'copilots' for their inherently collaborative nature, hitting the market or rolling out as an augmentation to an existing application.
How chief AI officers can streamline strategy from the boardroom down
With so much focus placed on generative AI and product after product aiming to fill new use cases, it’s more important than ever to slow down and ask: “Do I really need a copilot for this?”.
The tech industry’s answer, at least for the most part, seems to be “why not?”. If you’ve got a task that can be automated or a question that can be answered, the chances are a firm is already working on the perfect copilot to fill this role, from startups to hyperscalers.
One can’t deny that AI copilots offer much in the way of enhanced productivity and efficiency, with the added benefit of being embedded within cloud platforms or office suites. But there is a limit beyond which the volume and specificity of these tools become ridiculous.
Most recently, Microsoft introduced its brand-new ‘Copilot for Finance’ which, of course, will provide chatbot assistance specifically to employees in finance roles. Currently, ‘Copilot for Finance’ can run variance analysis, reconcile data in Excel, speed up the collections process in Outlook, as well as draw on information stored in SAP and Microsoft Dynamics 365.
In and of itself, ‘Copilot for Finance’ seems a perfectly reasonable offering from Microsoft, an AI solution to a productivity problem. Put it against the backdrop of every other copilot release and you start to see things differently. Within Microsoft’s ‘Copilot’ umbrella there’s a Copilot for Microsoft 365, a Copilot for Windows, a Copilot for security, a Copilot for sales, a Copilot for services, a Copilot for Github. The list goes on.
And that’s just Microsoft. Enterprise software company Sage has developed its own copilot, Siemens has created an industrial AI copilot, and other major tech companies have their own copilots in everything but name.
“We want every one of the departments to be enabled and enriched with a Copilot,” says Charles Lamanna, a Microsoft corporate vice president, in an interview with CNBC.
We could soon reach a point of copilot fatigue as a form of digital transformation burnout. With so many copilots to choose from in so many increasingly niche areas, more and more of the time saved by using AI assistants could be wasted in the process of copilot augmentation.
We’ve seen a similar thing happen in the security space, with multi-factor authentication (MFA) fatigue. This has come about as a result of employees facing an increasing number of MFA requirements, to the point of tiredness and annoyance that puts them at risk of hitting ‘approve’ on login attempts without even checking that they were the ones attempting to gain authentication.
Though the advent of the copilot may not bring with it the same sort of security risks, there’s potential for a similar issue. Unless the dozens of proposed copilots are all serving clear and definite functions, then they’ll quickly contribute to a level of organizational chaos in staff workflows.
Staff could become inundated with copilots, causing issues in which they don't know which to use or begin to ignore them altogether. This effect, often known as ‘tool sprawl’, refers to a situation in which organizations fill their stack with so many tools that things become overly complex for staff and efficiency plummets.
Think of reams of niche copilots that need to be uninstalled or unlinked from company systems once a better one shows up. The effect on teams will be a sense of listlessness and a lack of transferable skill bases. Copilots also feed into underlying disorganization within business, as employees field siloed responses and summaries with their myriad AI helpers rather than collaborating with coworkers or putting information in shared channels.
How far will this go? A copilot for each role within a firm? A copilot for each customer a business interacts with? As the tools become ever-more specialized, summoning them will feel less intuitive and more like trawling an endless contact list for the one person in the company who knows how to do a specific task.
The promise of AI has always been to automate and simplify the tasks that we, as humans, would prefer spending less time on. Copilots are the most immediate manifestation of that aim. But if tech companies continue to eschew generalized assistants and focus instead on a product for every problem, tech teams may become overwhelmed. Copilots need to strike the right balance, or risk being dropped down the line.
