Nothing works without artificial intelligence (AI) anymore. At least, that’s the impression you get when you browse the websites of software manufacturers, relevant IT publications, corporate financial earnings reports, or social networks. And this impression is not entirely wrong: The “KI Monitor” of the Bundesverband Digitale Wirtschaft (BVDW) establishes an AI index that attests to an increased relevance for AI over the past few years. Furthermore, the area of application of AI also extends across the most diverse areas of private and professional life: assistance systems in cars, medical diagnostics, biometric recognition, predictive maintenance in industry – all of these would be almost inconceivable today without AI with its subtopics of machine learning, deep learning, and neural networks. Even the software used by plagiarism hunters, which has cost some German politicians their doctorate, is of course supported by AI.
Some might object that AI has already existed for years. Chess computers have been around for more than 30 years, and data mining became a big topic in the late 1990s. Nevertheless, it cannot be denied that AI processes have been further developed, refined, and made far more effective in the past ten years in particular, which has to do with big data, the expansion of IT infrastructures, and even cloud models.
AI is supposed to increase efficiency in companies. A nice (side) effect: It relieves people in their work routines so that they can concentrate on essential activities that add value. However, one question remains: Does AI also promote job cuts when algorithms increasingly take over the activities of humans? That is certainly a valid concern. Nevertheless, AI will never be able to completely replace humans. As a rule, AI processes work retrospectively: They analyze processes and procedures of the past in order to derive suggestions for actions to be taken in the present. Although humans do nothing fundamentally different, they also use empathy, networked thinking, creativity, and cognitive flexibility. And this is where AI reaches its limits.
An example from finance is the accounts payable (the invoice receipt) process. AI has been used in this area for some time to increase the read rates when using OCR for paper and PDF invoices. The idea is to save the accounting department the hassle of entering invoices into the SAP system. Instead, accounting can attend to more complex operations such as fixed asset accounting, depreciation, or periodic consolidations and add even more value to their organization.
Master of the process
However, AI doesn’t stop at OCR operations. AI is now also used for automated account assignments for financial documents or for managing discrepancies between invoices, purchase orders, and goods receipts. Still, companies need (human) accountants to regularly check tax compliance, intervene when exceptions occur, and ultimately be the master of the process. Accounting on autopilot – that’s impossible for tax compliance reasons alone. AI may well be responsible for cutting some accounting jobs, but these are mostly jobs that are already difficult to fill due to the state of the labor market. Already, finance managers complain that employees leaving due to age are almost impossible to replace. AI can alleviate some of these issues.