As part of the digital transformation, the success factor “information” is increasingly moving into to the center of attention for many business activities. This makes it more and more essential for companies to view information in the form of data as a valuable asset and treat it accordingly. The transformation to data-driven organizations requires a completely new alignment of existing business models.
On the other hand, a recent study conducted by SAP, in collaboration with the Economist Intelligent Unit on the Introduction of Artificial Intelligence (AI), shows that the organizations that work the most with machine learning (ML) show on average 43% more growth than those who have not started using this new technology for themselves.
While artificial intelligence is now becoming the core of any business strategy, many companies are still struggling with the introduction of comprehensive data governance policies and the associated management of information. Leading organizations have discovered that the first step in making a successful journey to artificial intelligence is to improve their data management. Regardless of how intelligent algorithms may be, suboptimal training data will severely limit the effectiveness and later applicability of the generated rules. Transparency, quality, unambiguous master data, clear responsibilities and governance guidelines are crucial to the success of AI and ML. According to a survey by Accenture, 51% of the surveyed companies see the biggest obstacle to the lack of data quality. They also face difficulties with data- and cybersecurity, as well as the decision to evolve into artificial intelligence users or to work with appropriate partners.
Data management and machine learning should be considered as complementary building blocks. AI methods and algorithms help companies generate valuable insights from large amounts of data, such as can be used for a more target-group oriented customer interaction or for a more efficient equipment use.
At the same time, the question arises as to whether and, above all, how the value of the data can be quantified for the enterprise and what the most valuable information is for the respective organization. Every new technology requires new, more appropriate and sometimes better rules. Businesses need design principles that aim to gain high visibility into how AI is generating predictions. This allows a proactive approach to regulatory and ethical dilemmas which are often associated with the use of AI.
However, to ensure that the AI does not remain an empty promise, companies need data of a very high quality. While in the traditional data analysis it is possible to locate and to remove suboptimal data and start all over again , this is not readily feasible with ML procedures, since from a certain level onward it becomes impossible to reproduce on which data elements the predictions are based. KI turns into a black box technology and “unlearning” almost impossible as the removal of a single element can cause the entire model to collapse. In analogy to the human brain, the entire complex of learned knowledge, which is based in part on false assumptions or information, loses its value and the process has to be completely re-established. Bad data records can definitely be identified in advance via faulty information, outliers, distorted value distributions, redundant information and poorly explained functions.
The SAP HANA Data Management Suite solutions can be used to set up efficient measures for the management of data along business processes and to generate added value from information through integrated processes, whereby KI and ML support the user efficiently, make work easier and significantly increase the quality of data entry. Concrete examples include algorithms that intelligently pre-populate fields, individual acquisition patterns of employees being learned, and voice input that will be supported by chatbots.