Organisations continue to struggle to find practical applications for artificial intelligence, even as the hype around the technology continues at pace. However, one left-field suggestion to getting the most out of AI has come from experts at Airwalk Reply, with the firm recommending in a recent thought-piece on its website that companies might try “treating AI systems like employees”.

While treating an AI like an employee might vary wildly between different employers – especially among those who don’t respect the rights of their human staff either – what the piece means broadly is that rather than treating AI as a tool or piece of software, businesses should define specific roles for ‘AI staff’, monitor performance, ensure accountability, and foster continuous improvement.

Principal AI Consultant Sanjay Dandeker, who collaborated with an AI to produce the article, suggested that by conceptualising AI as a ‘digital employee’, organisations might “better align AI-driven processes with business goals, enhance transparency, and mitigate risks associated with autonomous decision-making”.

Of course, there are challenges to this mode of operation. Most notably, the non-deterministic nature of AI outputs mean it could be tough to predict how an AI system will respond to new input. While this might be similar to human employees, whose actions can be challenging to fully anticipate, the speed and scope of AI work also means tracking the quality of its output following that new input means wide-spread validation, testing, and monitoring measures will be needed “to ensure consistent and reliable performance”.

However, according to Dandeker and company, by “aligning AI management practices with those used for human employees”, organisations can still integrate AI more effectively into their operations. By linking it with human staff, employers can ensure that AI systems contribute with consistency to reaching key business objectives.

Making a plan

The key is to set out clear goals from the get-go. According to the Airwalk Reply article, just as with onboarding a new employee, employers need to define the specific roles and tasks that AI systems are expected to perform.

Dandeker adds, “This includes outlining the functions the AI system will perform, identifying the scope of its decision-making and operational boundaries; aligning AI roles into the broader business strategy by aligning AI outputs with key performance indicators and business goals; and creating detailed documentation that specifies the AI’s purpose, operational limits, and expected outcomes – much like a job description for a human employee. Regularly update this documentation as roles and responsibilities evolve.”

Meanwhile, when it comes to performance monitoring and evaluation, just as employees undergo regular performance reviews, “AI systems require continuous monitoring to ensure they function as intended”. This could include defining specific, quantifiable AI metrics such as accuracy rates, processing speeds, error rates, and contributions to business outcomes; conducting periodic assessments of the AI system, using the findings to adjust and improve AI models; and comparing the AI system’s performance against industry standards or benchmarks to ensure it operates at a competitive level.

What does differ from human staff, is the accountability for the performance of the AI is unlikely to stop with the AI itself. While if a human worker who underperforms, even if they have continuous training and support from their surrounding team, may end up being issued a warning, or even losing their job, an AI doesn’t have to worry about rent, so losing its job is hardly a threat. And should a business ‘fire’ an AI, replacing it with a new ‘digital employee’ is likely to be a very expensive process.

A human manager is always likely to carry the can for the mis-firing of an AI then. However, as AI adoption matures across an organisation, this will require “specialised AI supervisory roles can be introduced to manage and provide oversight for more advanced ‘Agentic AI’ solutions – review our previous blog post for more insight.”

Continuous improvement

According to Dandeker, AI systems benefit from constant input and training. So, organisations will need to invest in “updating AI models, incorporating new data, and refining algorithms to keep pace with changing business and technology needs” continuously.

Key practices include regularly furnishing AI models with new data and feedback. At the same time, firms should establish feedback loops where human employees can provide input on AI performance and outputs, which can be used to make iterative improvements to the AI system. By fostering a culture of continuous improvement, Dandeker argues that “organisations can ensure their AI systems evolve alongside their business needs and the rapidly changing technology landscape.”

Dandeker concludes, “Ultimately, viewing AI as a remote employee allows organisations to manage AI with the same rigour and oversight as their human counterparts. This creates a cohesive and efficient environment where AI can thrive alongside human employees, empowering AI to contribute effectively to advancing the business. It also prepares organisations to navigate the complexities of an increasingly AI-driven world with confidence and clarity. Managed effectively, it can help to drive innovation, efficiency, and competitive advantage.”