The Rise of Secret AI at Work: An Urgent Call for Skills Training

Whether leaders believe that their organization is ready for artificial intelligence (AI) or not, many employees are taking advantage of AI tools for themselves, sometimes in secret. In the US, an estimated 32% of workers are using AI at work without their employer’s knowledge. That should be both a warning and a call to action.

Training in fundamental AI skills is essential for safe implementation, but the cliche is true - where there is risk, there is opportunity. Change-readiness is a top concern for leaders early in their AI journey, and hidden AI adoption signals that teams are eager and prepared to embrace its potential.

In this guide, we’ll take you through the most important skills for employees: the essential skills that cover the fundamentals, the soft skills that are often overlooked, the hard skills required to drive results, and the leadership skills necessary to build an AI-driven culture.

Essential AI skills for every employee

AI literacy

Digital literacy is increasingly an assumed skill. Someone with ten years of experience in tech startups is surely “digitally literate.” But it’s a mistake to think of AI literacy in the same way, even as the tools become more commonplace. Training in the most basic AI knowledge ensures everyone has the same understanding of the fundamentals:

• The limitations of AI: Knowing not to trust outputs blindly, and when human oversight is required
• Data privacy: Understanding that any data entered into an AI tool can potentially be stored or shared, unless explicitly stated otherwise
• Transparency: Knowing when AI usage should be disclosed, internally and externally
• Bias awareness: Reviewing outputs for fairness and to avoid discriminatory decisions
• Usage and reliance: Using AI as an assistant, not a replacement

Training on these specific points builds confidence in using AI responsibly, easing adoption across the organization while preventing misuse or overreliance.

In addition to running live or recorded training sessions in these areas, consider incorporating clauses into your employee handbook and integrating these best practices into your onboarding process. Every employee, whether on the tech team or not, should be aware of your organization’s stance on AI.

Data literacy

Knowing how to interpret and question data has long been a skill that most employees can benefit from. But for many, data literacy has gone from a nice-to-have to a must-have. With AI now producing data-driven recommendations at almost every level of work, employees need to be able to validate and contextualize. Otherwise, AI is given too much authority, and human oversight becomes much more difficult to ensure.

For example, payroll teams using AI to detect anomalies must interpret flagged results correctly. Without basic data literacy, teams risk blindly acting on flawed insights. Luckily, data is a learnable skill. By offering workshops and embedding data reviews into meetings, even those who are not mathematically inclined can build up their skills.

Adaptability

AI technologies change fast, and that means that few AI-driven processes are set in stone. When a new generative AI tool lands on the market, or an existing tool launches or removes a key AI feature, processes are bound to change. Pilot programs fail to take off, or are deprioritized in favor of something new. This is why adaptability is key.

Adaptability is a trait usually inherited from experience, making it a tough skill to teach. But it can be fostered as part of company culture. A culture that rewards experimentation and is empathetic when change disrupts people’s workflows is one where employees will naturally learn to adapt. In practice, this could mean being open to revising individual KPIs when disruption impacts a colleague’s end-of-year results.

Soft skills for thriving in an AI-driven workplace

Critical Thinking

Closely tied to data literacy from our list of essential skills, employees need to be able to evaluate AI-generated content and recommendations, rather than accepting them at face value.

This is especially useful when AI outputs conflict with human intuition or expert knowledge. At best, this leads to the marketing team publishing AI-generated images of people with twelve mangled fingers. At worst, it could mean recruiters unintentionally reinforcing bias because AI screening tools overrepresent resumes from certain demographics.

To encourage critical thinking, run ‘AI vs. human’ scenarios in workshops to demonstrate when to trust machine learning and when to trust your instincts. Map out decision trees for what to do when AI goes wrong, and make it clear that raising red flags is encouraged, not punished. Nobody wants to be the only naysayer when an exciting new tool arrives, so make ownership clear and implement simple reporting processes.

Collaboration

As AI integrates into shared workflows, collaboration becomes a critical skill for better implementation. This will only increase as company-wide AI initiatives are introduced. Otherwise, you risk siloed adoption, duplicated efforts, or conflicting processes.

To encourage collaboration surrounding AI, launch cross-functional pilot programs between teams. This could be an AI-powered customer support chatbot that answers simple queries. Customer support teams provide the knowledge base, product teams ensure accuracy, and engineering handles implementation. For more advice on AI pilot programs, check out our complete guide.

Experimentation

Experimentation as a mindset is a valuable skill in the era of AI, one that takes employees from reactive to proactive. Tools like ChatGPT, Gemini, and Perplexity are user-friendly and endlessly adaptable. When employees find new and exciting ways to free up valuable time and resources with AI by themselves, your organization wins.

To foster experimentation, create spaces where teams can share ideas and recommendations among themselves, like an open forum during weekly meetings, or a dedicated Slack channel. Encourage managers to test new AI tools and processes for their teams, and highlight easy wins.

Hard skills that drive results

Prompt engineering

Prompting is an unusual skill in that it looks easier than it is. If you can string a sentence together, you can use ChatGPT. But if you’re looking for accurate and reliable results from basic chatbots, let alone more complicated tasks like automating workflows, prompts need to be more specific. Bad prompting leads to inefficient AI use, low-quality outputs, and wasted time.

To fix this, share prompt libraries and run training on the basics of how prompting works. If you create channels for employees to share ideas, encourage them to share the prompts they used for others to copy and paste, or to suggest improvements.

Data analysis and visualization

This is a step up from the data literacy previously mentioned. Data literacy is knowing how and when to question data. Data analysis and visualization are more proactive. It involves being able to interpret data and communicate insights through clear visuals. This makes AI-driven data actionable, making it more impactful to present to leaders or clients.

AI is helpful here, with tools like Tableau and Power BI needing only natural language queries to turn data into visible insights. But teammates in non-data-focused roles may still need extra training to get the most out of them.

Process automation

Automating routine tasks is often heralded as the easiest and quickest way to start adopting AI. It removes the rote work that no one enjoys and frees up time for more important tasks. However, without internal guidelines, processes may be automated in inconsistent or risky ways, leading to duplicated efforts, overlooked compliance requirements, or errors at scale.

One way around this is to use AI agents, like Deel AI Workforce. With a library of pre-built AI agents, you have control over what gets automated within your organization and how. And when your teams are ready for something more complex, they can build their own agents for specific workflow needs in minutes.

Leadership skills for future-ready AI

Change management

We’ve mentioned already that employees need to be adaptable, but it's not a one-way street. Leaders must guide teams through uncertainty and resistance to AI, or else they risk employee pushback, low adoption rates, and stalled initiatives.

Effective change management means taking a strategic approach to transitioning teams from their current way of working to an AI-driven future. Leaders involved in introducing new tools or workflows need to develop a clear communication strategy with clearly defined roles and responsibilities. Managers are the most important players in change management. Aligning them around one strategy and providing resources will help them manage friction if it arises.

Ethical AI stewardship

Trust is the cornerstone of scalable AI adoption. When teams trust the technology they’re using and leadership’s vision for the future, everyone can get to work faster and with clear direction. Equipping managers and decision-makers with knowledge of ethics and compliance in relation to AI ensures they can guide their teams in the right direction.

Ethics and compliance training, not just for managers but across the organization, builds trust and reduces risks. This training might include scenario-based workshops, bias awareness labs, and ethics playbooks that teams can apply in their day-to-day work.

The future workforce is hybrid: Human + AI

AI is already in the workplace, whether formally rolled out or not. By proactively equipping teams with the basic skills they need to navigate the new technological landscape, leaders can both reap the benefits and mitigate risks.

At Deel, we believe the future of work is AI-powered. A hybrid of human intelligence, intuition, expertise, and the speed and efficiency of AI. That’s why we built Deel AI Workforce, a library of pre-made and customizable AI agents designed to join your team and proactively handle routine work. They work alongside your teams to handle everything from PTO requests and tax compliance research to payroll checks and device requisition. They’re easy to use and have compliance and data security built in.

Sign up today, and be among the first to welcome Deel AI Workforce to your team.