Leveraging Generative AI for Job Augmentation and Workforce Productivity 2024

Many organizations interviewed for this report do not know exactly what percentage of their workforce uses GenAI for work. Reported figures vary from 20% to 80%, while some stated that “almost everyone” was using GenAI, or that at least everyone could because they gave the entire organization access to a GPT-powered chatbot or Copilot type of solution. The accessibility of GenAI tools across the case study sample varies widely. While some respondents grant all employees access to all tools, others limit this to one or two departments, and in some cases, licenses can only be obtained on request. This decision is usually strongly related to the organization’s risk appetite and past experiences. A few organizations have had “accidents” such as data leaks while others saw that GenAI tools were used irresponsibly, such as composing sensitive letters or emails. This led to delays in further implementation, causing frustration among employees eager to proceed. Such risk appetite determines the pace of implementation and whether GenAI is only deployed internally or also in client-facing services. “Once you take the leap of faith, you can’t go back,” said one organization. “Meanwhile, those people have also started doing something else or something new in the organization. GenAI really does change jobs. If you want to go back, you have to put everyone back in their old position.” The importance of “humans-in- the-loop” Past scandals involving discriminatory algorithms, the black box nature of GenAI, and the introduction of government regulations (such as the European AI act) have increased respondents’ risk awareness when it comes to GenAI workforce deployment, with a notable example being the tendency of GenAI to generate illustrations of meetings that predominantly feature white, older men. This highlights the importance of validation, verification and human intervention in the process. One interviewee emphasized that “the biggest mistake you can make is to remove humans from your processes.”GenAI councils to safeguard quality and ethics To monitor the risks, quality and responsible use of GenAI, most organizations interviewed work with internal committees or councils. These councils establish internal rules, standards and frameworks and assess use cases. These councils are often organization-wide, with representatives not only from Risk and Compliance and Legal functions but also from Strategy, Marketing, IT, and Business. The reason for such an approach is to safeguard that all perspectives, risks and opportunities are considered in decision-making. However, opinions differ among respondents on whether this leads to acceleration or delay in the workforce deployment of GenAI. Some complain about the slowness of this alignment process, while others argue that it helps achieve alignment sooner and safer. Nearly all organizations interviewed report that they have developed training in responsible use; however, not all organizations make this training mandatory. Sustainability considerations LLMs, central to GenAI systems, are energy- intensive compared to smaller, task-specific AI models. Each prompt from large models requires calculations that consume a significant amount of energy. The higher the adoption of GenAI, the more energy consumption this entails. This contradicts the mission of most organizations to reduce their environmental footprint. While this is a problem that most organizations acknowledge, few have developed a strategy for acting on it, yet. Some argued that it should be addressed in the supply chain, by the cloud providers of GenAI tools who have promised to strive for climate neutrality, while others expected policy and regulators to weigh in eventually. For the moment, based on the interviews conducted for this report, one conclusion is that environmental considerations do not seem to be central to GenAI workforce deployment decisions.Insights on GenAI workforce deployment: Risk governance3.4 20