The practice of outsourcing simple processes and repetitive tasks to third parties or offshoring them to take advantage of lower costs in other locations in nothing new. Neither is the fact that many of these tasks are now being performed by AI agents.
What is new, according to Andreas Hahn, the employee representative on the SAP Supervisory Board, is that now, even the low-cost offshored workforce has to be upskilled in order to be on and in the loop when using AI tools.
Hahn was discussing the impact of AI agents on the workforce on a webinar sponsored by SSON, the Shared Services Outsourcing Network, with Uli Erxleben, CEO of Hypatos.ai, a certified SAP partner that excels in automating back office tasks.
Leveraging employee knowledge
“At SAP, with the rise of AI, our workforce is doing more complex work, so we need to continuously upskill employees. We’re seeing that the deep knowledge of the existing workforce in all locations, even low-cost ones, must be leveraged to increase productivity and quality,” said Hahn.
Hahn believes it’s a big mistake to think that AI will eventually lead to layoffs. Reducing the workforce leads to less productivity, whereas AI tools help humans deliver higher quality work. Companies should be exploring ways to successfully integrate AI tools in their current workforce. SAP’s approach is to reinforce the role of management in delivering training at all levels of the organization, beginning with the Executive Board.
“The new managerial role is focused on coaching humans while training AI agents to learn and increase productivity together,” he said. “AI does not have all the necessary context, so human supervision, called human on the loop, is necessary to provide feedback to the AI system to improve its performance over time.”
Hahn explained that AI is extremely beneficial to the SAP salesforce for analytical purposes, particularly during quarter end and year end when deals and pipelines are being finalized. The company’s AI tool SAP Joule is used to support the process but cannot replace the human in the loop element, which is the collaborative approach that integrates human input and expertise into the lifecycle of machine learning (ML) and AI systems.
“AI cannot deliver the necessary ‘trusted relationship’ or make judgement calls that depend on deep insights, or demonstrate empathy, all qualities that help create customer satisfaction,” Hahn said.
Harnessing the human element
AI must be highly trained to achieve maximum benefits, and training and coaching AI agents remains a human task. Hahn pointed out that AI may come across as extremely confident, but that doesn’t necessarily mean it is always right.
“Software developers and other employees must be trained to question results because AI delivers statistics based on analyzing vast volumes of data. As with any statistical model, there may be deviances that require human knowledge and expertise for interpretation,” he explained.
Hahn also explained that SAP was an early contributor by setting up an AI ethics policy, that includes rules similar to the new EU AI Act which will shape the global standards for the ethical use of AI.
SAP Joule was developed to provide out-of-the-box integration with SAP backend systems ensuring compliance with AI ethics, GDPR, and privacy controls while maintaining regulatory compliance. SAP staff are being trained to answer customer questions on this topic and to help customers train their own staff to review results delivered by AI.
As an expert in labor relations and software development, Hahn has some key advice for companies in the process of implementing or scaling up the use of AI tools in their workforce. Because AI is changing the way humans work, its implementation can cause fear and insecurity amongst employees.
“It’s important to involve social partners at an early stage and remain transparent throughout. Nothing is more toxic than the suspicion of having a hidden agenda, so building trust is essential in diffusing tense situations,” said Hahn, who was a contributor to a research project by the Hans Böckler Foundation on adopting AI in the workforce in a humane way.
Beyond productivity
For Uli Erxleben, the discussion with Andreas Hahn simply reinforced his view that two years from now, AI agents will not just be considered a productivity tool. They will be performing all back office operational tasks better than any human being at a much higher quality, so it would be senseless not to use them.
Hypatos’ market-leading solutions were developed to liberate people from manual document processing. They ensure extraction accuracy and quality in document processing while improving over time with the right human supervision. In addition, its end-to-end, automation enable document coding that processes documents in downstream systems and processes, resulting in a 20x productivity uplift.
“We provide a human-in-the-loop platform that allows no-code annotation, validation, completion and routing of documents – all while providing powerful business analytics,” Erxleben explained.
Erxleben concluded that managing the transition to a hybrid workforce is clearly the task of human beings. It requires strategic engagement with educational institutions, trade unions, workers councils and HR departments. The task at hand is to reskill the workforce now, keeping in mind that much of the current workforce performing repetitive tasks may be eliminated in the future.
Both experts agree that employees who are empowered to act on the values, ethics and culture of the company will be key to creating successful workforces in the future. Managerial jobs will focus on supervision of AI agents, and involve human capabilities like making judgment calls, taking risks and accepting responsibility. The role of managers will be to ensure accountability and compliance.
