AI at Work: “Workslop” and Pitfalls to Avoid

AI at work: An enterprise-wide focus

Leveraging AI in the workplace remains a top strategic priority, driven by both top-down leadership initiatives and rising employee expectations. At the leadership level, AI has become central to corporate planning, with Gartner reporting that 68% of companies intend to develop strategies for integrating human employees and AI agents. Many executives are also setting bolder ambitions, with one in three companies planning to achieve fully automated operations in functions such as logistics, product design, and contract management [1].

As AI adoption moves beyond the experimental phase, organizations are shifting their focus from simple usage metrics to measurable performance outcomes. Developing ROI measurement has become standard practice: 72% of companies now report formal ROI tracking, and half of them go a step further by continuously monitoring, optimizing, and embedding these metrics into corporate strategy and planning [2].

The surge in AI adoption at work, however, is not driven by leadership alone; it is increasingly powered by employees themselves. According to research by McKinsey & Company, AI is attracting attention from almost everyone, including those most skeptical of the technology. The study finds that 71% of the most AI-skeptical employees are familiar with GenAI tools, and around half of them feel comfortable using these tools at work [3].

The same research also reveals a notable perception gap between leaders and employees regarding actual AI usage. While leaders estimate that only 4% of their employees use GenAI for at least 30% of their daily work, self-reported data from employees puts the figure at 13%—more than three times higher than leaders expect.

AI is now being actively leveraged across industries and business functions, generating tangible productivity gains in specific areas. Content generation and fraud detection stand out as leading generative AI use cases in marketing, sales, and operations, respectively [2], with real-world companies already reaping substantial benefits.

One example is a leading international insurance group that implemented FPT’s iSuite, a comprehensive AI solution package for insurance. As part of this initiative, the company deployed an AI assistant on its digital platform for insurance agents. The assistant supports agents with key daily operations such as task management, lead engagement scripting, and follow-up recommendations for lead conversion.

These AI capabilities translated into measurable business impact for the insurer. The key performance improvements are summarized below:

Furthermore, iSuite’s AI-driven fraud detection capabilities analyze claims patterns and submitted documents to flag potentially fraudulent activities. This application has helped the insurance company both strengthen its fraud detection rate and significantly cut fraud-related losses.

“Workslop” and key AI pitfalls to avoid

AI-generated workslop and its impact on productivity

“AI-generated workslop” is a term coined by researchers at the Harvard Business Review to describe content that masquerades as good work but lacks the substance to meaningfully advance a given task [4]. In other words, it looks polished on the surface but fails to move the work forward in any meaningful way.

According to the research, up to 40% of full-time employees in the US have already encountered this kind of low-quality, AI-generated output. They then have to spend extra time cleaning up, revising, and interpreting the content instead of focusing on higher-value tasks.

As a result, “AI-generated workslop” does not just fail to boost productivity; it is actually destroying productivity, as the study puts it. For a company with 10,000 employees, the lost productivity is estimated at around $9 million annually if each employee spends 1 hour and 56 minutes fixing such workslop [5].

The consequences are not merely theoretical. In practice, Deloitte recently agreed to partially refund AUD 290,000 to the Australian government after errors were discovered in an AI-assisted report, including a fabricated quotation and references to non-existent research sources [6].

AI bias and unfair outcomes in the workplace

Workslop is not the only risk organizations and employees face when adopting AI in the workplace. Another critical concern is AI bias, which refers to biased outputs generated by AI systems that can result in unfair treatment of certain groups of people.

Such bias can arise from several sources, including imbalanced or unrepresentative training data, flawed algorithmic design, or the unconscious biases of human developers that are embedded into the models. When these issues go unchecked, AI systems can reinforce and even amplify existing inequalities.

A widely cited real-world example emerged in 2018, when Amazon shut down an experimental AI tool for screening job applications over fears it was discriminating against female candidates. The model was trained on resumes submitted to Amazon over the previous 10 years, which were predominantly from men.

Because of this skewed training data, the algorithm learned to favor applications from men and gave a lower rating to resumes that included the word “woman” [7]. This case illustrates how quickly AI bias can translate into real, harmful impacts on people’s careers and opportunities.

Productivity is about more than speed and volume

AI unquestionably boosts productivity by streamlining workflows, automating repetitive tasks, and generating content at unprecedented speed. However, this output is not always fully accurate. Most current productivity metrics emphasize how fast and how much AI can produce, but these indicators alone are insufficient. Organizations need to assess AI performance beyond usage rates by setting clear measures for the quality and accuracy of AI-generated outputs. A well-defined AI usage policy is therefore essential, with explicit guidelines on which platforms are allowed, what types of data may be shared, and how AI-generated content must be reviewed. Without such governance, AI adoption can quickly become uncontrolled. With as many as 63% of surveyed software developers reportedly using unauthorized tools, the associated security and ethical risks are increasingly evident [8].

Training is another critical enabler. Despite the widespread attention AI receives, employee training remains limited: only 29% of U.S. workers report receiving adequate AI training from their employers [3]. Increasing the amount of training is important, but enhancing its quality is even more crucial. Effective AI training should move beyond basic prompt creation to help employees understand AI's limitations, recognize potential risks, and learn how to mitigate errors and biases in real-world use.

Finally, companies should clearly define the role of AI within their operations. However advanced or autonomous it may become, AI remains a tool. The responsibility to use it accurately and ethically rests with humans. Organizations should therefore establish robust frameworks for governing AI usage, clarifying when human judgment must intervene and how accountability is allocated. Maintaining strong human oversight is essential to ensure that AI-generated outputs align with corporate vision, ethical standards, and regulatory requirements.

Conclusion

AI is rapidly becoming the backbone of modern enterprises, powering impressive gains in productivity, automation, and decision-making across functions. Yet, as “AI-generated workslop,” bias, and high-profile errors have shown, speed without substance can quietly erode value, waste time, and damage trust. True productivity in the age of AI demands more than volume: it requires robust metrics for quality and accuracy, clear policies and governance, and meaningful training that equips employees to recognize limitations and intervene wisely. Ultimately, the organizations that benefit most from AI will be those that treat it as a powerful tool under thoughtful human stewardship, and the real question now is how deliberately you are designing that partnership before it designs your workflows for you.