Using AI and Machine Learning for Monitoring Employees

In 2022, IBM reported that about 35% of companies had adopted AI, and another 42% were "exploring" it. The word "exploring" has an important connotation here.

Often, there's a lot to consider for enterprises when it comes to overhauling their systems and processes. As such, extensive functions like employee monitoring often get left behind and continue relying on the legacy infrastructure.

But things have changed a lot in the past year or so. Much can be attributed to the growth in generative AI (gen AI) solutions. McKinsey calls 2023 as "Generative AI's breakout year." The consulting giant says that companies using gen AI for performance management and organisation design are already ahead of their competitors.

When talking about AI's influence on workforce performance management, one business function that benefits immensely is employee monitoring.

For long, employee monitoring functions have been limited to recording rather cold data that might suit certain metrics (such as the number of hours worked). This data lacks the context into the how(s) and why(s) of employee engagement and performance.

In other words, there's no qualitative aspect that could unearth the root causes for the drop in productivity or other anomalous work patterns. AI helps solve this problem by helping move from "cold data" to "insights." With AI, the employee monitoring function becomes more oriented towards deciphering work habits, productivity, strengths, weaknesses, etc.

For example, consider the case of a software development team that uses a traditional employee monitoring solution for performance optimisation. The tool provides the data for the number of code commits or the hours logged. But, this doesn't convey the complexity of the task at hand, the level of collaboration that would have been necessary, the irregularities that could have cropped up.

Now, let's assume the same scenario, but the team is using an AI-powered employee monitoring solution. The platform does everything that a traditional solution would do but also provides insight into the communication patterns, collaboration level, problem-solving efficiency, feedback mechanism, etc.

How? By noticing every little detail about work and unearthing patterns that could further help formulate best practices for improvement. For instance, an AI system can relay if a certain developer's code commits do not entail subsequent bug fixes. The developer's process can be established as a best practice.

Likewise, the system can help understand if the developer's productivity is down during a specific time of the day. This way, the team can redistribute the work and better understand the work patterns.

There's a lot of talk about security implications with employee monitoring solutions, and understandably so. We're talking about monitoring people and their day-to-day work activities. So, there are bound to be concerns — precisely why regulations such as Europe's General Data Protection Regulation (GDPR) have been set up.

Of course, there's the consensual angle to this story. But we're only concentrating on the technology front for now. To that end, AI's integration offers numerous advantages to organisations in terms of:

• Learning typical user behaviour and detecting anything that's anomalous to that
• Understanding sensitive data access patterns and flagging any activities that are seemingly unusual
• Setting up automated alerts for suspected security incidents and allowing security teams to take proactive action instead of resorting to post-mortem analysis
• Detecting any insider threat activities by ensuring that anything that's beyond the normal work pattern of behaviour gets immediately noticed
• Blocking the transfer of any sensitive information within and outside the company network

Let's quantify these capabilities to understand their tangible value. During the initial days of the COVID-19 lockdown in 2020, remote workers were constantly under attack by malicious users. Deloitte reported a fivefold increase, from 12% to 60%, in the attacks on remote workers during this time.

Elsewhere, the cost of insider threats reached $15.38 million in 2022. A US-specific report suggested that businesses are experiencing around 2,200 internal security attacks on a daily basis.

So, what would be the ideal AI-employee monitoring solution in the wake of looming cyber threats? By keeping "security" at the forefront. CleverControl's Face Recognition feature is one such AI-powered surveillance solution that makes for better data and employee security.

We surveyed several executives and understood that they were struggling to:

• Prevent people from allowing an external person to cover up for their work
• Ensure that the employees weren't cheating by opening productivity apps and were actually attending their work desk when working remotely

Hence, we brought the Face Recognition feature to our monitoring solution, which could record employees (by taking a photo every five seconds) and ensure that no unauthorised entity entertains an employee's work. This proved a viable solution for many organisations, especially those in relatively heavily regulated industries like finance.

Businesses stress that employees are their most valuable assets, but that isn't, at times, reflected on the ground. The reasons could be manifold, including those that have nothing to do with systems and technology. For the sake of this discourse, we'll avoid those. Let's focus on the operational and technology inhibitions.

Often, the operational bottlenecks could lead to employee burnout, unfair performance assessments, and more. But AI comes to the rescue with a host of benefits. For example, AI-powered monitoring can help:

• Distribute workloads according to a team's or individual's working proficiency. Teams can understand the working patterns and solicit suggestions from AI about redistributing work strategically.
• Provide personalised feedback based on employee's productivity and engagement patterns. This can be as simple as providing a nudge to improve engagement on certain channels or recommending the whole series of courses for professional growth.
• Gauge employee sentiments through their communication patterns across internal or social channels being monitored. Large language models (LLMs) that power generative AI solutions can certainly be a boon for companies wanting to go granular into comprehending what employees like and dislike. This can help alert managers to potential hiccups in productivity and take proactive actions to personally interact with employees or perhaps even tweak the working conditions for the greater good.

While this doesn't befit the usual SaaS use case but AI-driven employee monitoring is also making waves in industrial settings. Take the example of computer vision-powered visual analytics within a factory where employees could be immediately alerted about potential disasters or prevented from entering certain facilities.

It would be unfair not to outline the public sentiment attached to the increasing use of AI for employee surveillance. And there are reasons for that.

According to a Pew Research Center survey , 81% of the respondents said that employees would not feel good about being inappropriately watched or evaluated using AI.

Reuters also raises questions about excessive AI-based employee monitoring, citing that "Many employees and privacy rights advocates see this as an unreasonable incursion into an employee's home, long considered to be their private sphere."

Apart from this, there's a concern about bias and privacy with generative AI solutions. These LLMs that we talked about above are trained on data, which can be intentionally biased. The AI would not be able to filter that and might allude to biased responses and assessments. So, that's also something that organisations need to consider when making the move to more advanced AI integration.

Amid this, it's reasonable to expect that there will be employee resistance to the use of employee monitoring solutions powered by advanced AI. They will question the potential bias in the assessment data, the security of their sensitive information, and the potential misuse of the same on the organisation's part. And all these are valid concerns to say the least.

Implications of AI-driven monitoring in the context of remote work and the changing landscape of work arrangements are profound. What organisations need to do is strike a balance between monitoring for productivity and keeping employee data safe.

To establish this, they need to understand that the adoption of AI should not be driven by "desire" but the "need" to have technology in place to power performance improvements. It's easy to get caught in the hype cycle of AI and difficult to evaluate how it actually benefits the business use case.

Our recommendation would be to rigorously assess the needs of your workflow, document the discrepancies, and see if an incremental approach to AI-powered employee monitoring benefits. For example, if you're concerned about productivity drop in remote settings, choosing CleverControl with Face Recognition might prove immensely viable.

Of course, there would be some experimentation in the initial stages. But a big leap isn't required if the business doesn't demand it.