T he jury is out on the role and effects of Artificial Intelligence in the workplace and in management. Many people worry about AI numbing and deskilling workers, narrowing creativity and innovation, and a wide range of unintended risks. Others focus on AI’s potential in liberating workers from tedious tasks, augmenting their abilities and capabilities, and leveraging human thought and digital assets at scale for more powerful and insightful business decisions. This dichotomy exists because AI is still mostly an unknown variable and does not fit into established management theories and frameworks. Although it masquerades as a plain and innocuous aid to human work, the reality is that agentic AI has the potential to operate and function independently or in collaboration with humans, and lead to the end of several analytical and managerial functions. We are not there yet though. In my work with Professor Malar Hirudayaraj, at the Rochester Institute of Technology; and Dr Bonnie Cheuk, at AstraZeneca, we have been following the adoption of AI at a global pharmaceutical company, studying their approach in involving the workforce in the process of learning about and integrating AI in their day- to-day work. ADOPTING AI Our research is part of an upskilling initiative to encourage employees to engage with a new internally developed and company-specific version of Gen AI, which we refer to as “GPT”. The organisation collected data in multiple stages through surveys and employee journal entries to capture concerns and to understand the support required to better integrate these technologies into everyday work. We found that the tasks delegated to GPT at this stage ranged from simple and repetitive to more complex, strategic and creative functions. All were in a supporting role to human work at this stage. We categorised these tasks by the level of cognitive effort and strategic thinking required into three distinct levels: low, moderate, and high cognitive complexity. These range from routine tasks that were mostly administrative to higher complexity tasks requiring creativity and strategic thinking, planning and innovation. LEVEL 1 At this level, workers used GPT for the summarisation of meetings and reports, organising content, and formatting and improving presentations. These were mainly routine and administrative tasks. The nature of interaction between Humans and GPT was transactional – inputting simple requests and receiving quick outputs. The tasks were repetitive, not requiring deep engagement, creativity, or strategic thought. GPT enhanced efficiency by taking on low-value activities, which are typically time and labour intensive. Employees told us: “I needed to summarise more than 1000 comments from Zoom chat during a live interactive workshop. I used prompts on GPT and got this done within a minute.” “Summarising a large report took me some time, I tried GPT and quickly summarised the key points and made the report more succinct.” LEVEL 2 At this level, the digital agent supported tasks that required more contextual understanding and a degree of analysis or creativity. Here, employees used GPT as a tool for four purposes: 1 To write and assist with professional communication such as drafting reports, formal emails, or improving language and structure. 2 For data synthesis and analysis of large datasets, identifying themes, and extracting actionable insights. 3 To support meetings in taking notes, summarising discussions, and creating follow-up items or action plans. 4 For idea generation, brainstorming and developing meeting agendas. The nature of interaction between humans and digital agents is more collaborative. AI assists with the first steps, and the user refines or evaluates the output. Employees told us: “I was writing a response to a letter that needed to have succinct and accurate arguments – I used GPT to check what I was writing and make it read better.” 8 |
RkJQdWJsaXNoZXIy NTI5NzM=