The Frankfurter Allgemeine Zeitung has examined the opportunities and challenges that Artificial Intelligence presents for companies. The article focuses on the question of how data needs to be structured so that AI can process it at all.
Data quality as the key factor
It is true that many decision-makers have come to realize that AI language models such as ChatGPT can offer enormous added value for companies. However, “it does not yet work comprehensively”, the authors conclude. “But it won’t work properly in the future either if companies don’t have an adequate database.” Companies therefore not only need to invest in the actual AI models, “but also set up good data management in order to create and maintain high-quality data for AI use”.
AI is only as good as its data
There are simple tasks that AI can perform even without the right data. For example, according to the FAZ, a chatbot is easily able to respond to a customer’s complaint. “However, the support provided by generative AI only becomes really helpful if the prompt is provided with a lot more information in addition to the specific complaint,” write the authors. “For example, with information about the customer’s history […] which products they use, how important they are to the company or how often they have complained in the past.” But although all this data is available in the company, it is usually not accessible to the AI.
Consistency is the name of the game
In addition, there is another challenge: most companies use different applications that do not communicate with each other. As a result, the data is inconsistent and the information is often outdated or contains errors. There is “no ‘golden record’, i.e. no consolidated, complete and reliable version” of the data, as the FAZ authors put it. “But even if the creation of such a golden record is successful, it must be ensured that changes to the master data are recognized.”
Generative AI alone is not the answer
All of these factors contribute to the fact that the introduction of Generative AI systems usually fails to meet expectations. “On the contrary: if the Generative AI can only see that a customer responds to marketing measures with many orders, but cannot see that these orders are all returned, then the Generative AI will recommend even more marketing measures,” the authors conclude. Productivity in the company will not increase as a result and could even suffer in the worst case scenario.
About Brain4Data
Brain4Data GmbH & Co KG develops solutions in the fields of Artificial Intelligence (AI), Robotic Process Automation (RPA) and Augmented Intelligence (AuI) in Saarwellingen. We specialize in processing existing company data in such a way that it can be understood by all Generative AI models – without complex IT projects or complicated interfaces. To achieve this, Brain4Data bundles cross-departmental information from various IT applications (e.g. BI and ERP systems, SAP applications, Excel, etc.) and processes it in a fully automated manner to create Gen-AI-capable content, personalized recommendations for action or proactive messages on urgent business transactions. We also offer our own chatbot that uses Retrieval Augmented Generation (RAG), which gives it numerous advantages over conventional Gen-AI models. Brain4Data was awarded the Seal of Excellence by the European Commission in 2024.
Your contact
David Woirgardt-Seel
Chief Knowledge Officer Brain4Data
Phone: +49 6838 50209-63
Email: david.seel(at)brain4data.de