Power to the people: supporting the generative AI revolution

Why businesses should embrace, not restrict, workers’ “shadow adoption” of generative AI.

Ready or not, generative AI is already seeping into the modern business, and the basic contours of the next few years are already clear. The executives who’ll emerge victorious are those who embrace and support this shift: putting up guardrails to protect data, offering training to upskill employees, and developing forward-looking processes that enhance outputs. Those who resist the coming changes, conversely, will get left behind, and find themselves playing a game of “catch-up” that they’re unlikely to win, if history is any guide.

Executives are not the only players here, however. Gen AI’s abilities are so impressive, and its barriers to entry so low, that many employees are already making use of generative AI’s potent toolkit, with or without the knowledge and approval of their employers. This so-called “shadow adoption” presents real risks to most companies. Given the high stakes of the present moment, many leaders are feeling an urgent and understandable need to restrict and crack down on the unsanctioned use of these powerful new tools.

But this impulse should be resisted. The reality is that your employees are going to find a way to use gen AI, and the risks will be more easily monitored and managed if workers feel not only encouraged but enabled in their exploration of this new technology.

Not a new problem

Comparisons to generative AI in business history are hard to come by; the technology’s adoption trajectory is even steeper than that of the Internet in the 1990s. But battle-scared CIOs and compliance officers may have some shadow-adoption war stories from that era. The basic problem with shadow adoption is that when employees introduce diverse hardware and software into the workplace, it poses no end of challenges to IT support structures, while undermining the value of existing learning and development investments.

The shadow-adoption of public storage solutions like Dropbox is a good example. Initially, businesses imposed strict restrictions due to data loss and security concerns. But as a deeper understanding of the risks emerged, business leaders adopted a more mature approach, focusing on enhancing system security and providing comprehensive training on the risks and proper usage of these tools. A blanket ban proved ineffective, while a more collaborative model balanced risk mitigation with employee demand.

Data confirms the wisdom of this approach. A recent study by Gartner revealed that in 2022, 41% of employees engaged with technology outside the IT department's oversight, with this figure expected to surge to 75% by 2027. The futility of trying to plug this firehose of enthusiasm and curiosity should be obvious. The better course, for almost every business, is to encourage this organic adoption of technology, implementing points of oversight and control where possible to mitigate negative impacts, ensuring that as adoption increases and the technology evolves, jobs families evolve with them.

A roadmap for rapid adoption

What does this mean in practice? What concrete actions can leaders take to help support the growing people-powered adoption of generative AI, while balancing business concerns?

1.    Training and support

Sometimes, the adoption of new tools and technologies is a slow and reluctant process. Teams and frameworks praise the virtues of new tools, hoping they encourage employees to embrace the latest business investments. The cost of failing to do so is laid bare in recent research conducted by Cognizant in partnership with The Economist, which highlights a growing ROI gap across a wide range of business technologies, primarily due to sluggish adoption rates by employees.

With generative AI, as we’ve seen the picture is reversed. Businesses are proceeding cautiously as they assess the technology's risk profiles, while employees charge ahead. That enthusiasm for this new technology comes with a hunger for knowledge and training that employers can use to exercise oversight and control. Not to mention the benefits of the training itself. As highlighted in Forrester's Future of Work Survey more than half of the workforce currently lacks the confidence to question the outputs of Large Language Models such as OpenAI’s Chat GPT. Equipping employees with the knowledge and skills to identify issues is paramount. As a starting point, leaders can:

• Develop relevant, compelling training content. Here, it’s crucial to remember that different people have different “learning styles.” Some prefer tutor-led instruction, for instance, while others engage better with video content. To maximize effectiveness, the training options should blend “traditional” self-paced e-learning with live forums, coaching sessions, and hands-on practice.
  
  
• Create “communities of practice,” in which employees can self-manage within parameters established by central teams such as HR, IT, and Learning and Development. This approach enables ongoing support and organic upskilling of teams and employees, while staying roughly aligned with overall corporate goals and strategy.
  
  
• Take a multi-prong approach of quick hit sessions along with formal trainings to not only provide awareness and enable expertise, but also continue to hone skills to raise the capabilities and digital IQ of your organization. This flexibility enables leaders to supplement more structured learning with quick updates that match the pace of new technological capabilities and system updates.

2.    Establishing clear risk guidelines

Given the speed of generative AI’s adoption, some risk is inevitable.

In every case, the key question is: where and how can we build safeguards to mitigate these risks? In many cases the answer will involve human oversight. In a new generative AI initiative based around email, for instance, one (human) individual could be assigned to scrutinize the template of the email before its integration with customer data, and another to conduct a final review of the output.

For certain activities, naturally, the stringency of these protective measures will be amplified or reduced, but the overarching goal is to build a comprehensive risk tolerance framework that can encompass the full spectrum of business activities and develop the organizational agility to adjust rules and guardrails as the generative AI landscape evolves.

3.    Building strong data foundations

One of the more prosaic predictions of how generative AI will transform the modern business is the way it will democratize access to corporate data, empowering employees at every level to make use of data without in-depth understanding or technical expertise. For leaders looking to mitigate risk while encouraging shadow adoption, ensuring the quality of that data must be a top priority. The higher the quality of your data, the less important it is that employees only use sanctioned, pre-approved gen AI tools to access it.

With this basic safeguard in place, employee access to corporate data can become a virtuous circle, as employees are empowered to identify problems in the core data set during their daily work. For instance, a worker noticing outdated payroll codes in the data can flag them for removal from the database, thereby “cleaning” the data set and further reducing the risk of mistakes. It’s even now possible to use gen AI itself to improve data quality, by gathering and consolidating data from disparate sources and in different formats, and it will soon be commonplace for human workers, in partnership with gen AI models, to create intuitive data-cleansing cycles that ensure valuable information and insights are always easy to access, by humans and machines alike.

Empower, rather than hinder, your people

Faced with insatiable appetite for generative AI from employees, time is a precious commodity. The adoption cycle has accelerated to the point where decades spent on proof of concepts are no longer an affordable luxury. Business leaders who want to harness the power of shadow adoption must act now, and start building the infrastructure to supply training, set risk guidelines, and maintain robust data foundations. Those that do will be richly rewarded, as they discover that what their employees are craving is not just access to this exciting new technology, but what workers have been craving since work began: a chance to belong to something greater themselves, and to enhance it with positive change and impact.

To learn more, visit the Generative AI, Intelligent Process Automation and Performance Imperatives for AI sections of our website or contact us.