Watch now: AI between hype and pressure to act – a reality check for 2026

Artificial intelligence put to the test: Why data curation is more important than the next model

AI is everywhere: in presentations, in strategy papers – and seemingly in every other software program. But while the hype grows, many fundamental questions remain unanswered:

• Can I trust the answers from my AI?
• How do I keep my employees professionally fit when "the machine" is taking over more and more tasks?
• And: How do I prepare company knowledge in such a way that AI can actually deliver meaningful results?

That was precisely the focus of the expert webinar – with a clear emphasis on knowledge management, data structuring and knowledge graphs as a basis for explainable AI.

1. Between hype and reality: Where AI really stands today

Initially, the question arose: Is the current AI hype justified – or are we sitting on a bubble?

Prof. Dr. Ute Schmid, who has been researching artificial intelligence since the 1990s, assessed the situation: The bubble will not burst completely, "but it will in some areas."
While the “battle of the big models” and generative AI currently dominate the headlines, the real innovation is happening more quietly – for example, in… agentic models and in the vicinity of Retrieval Augmented Generation (RAG).

What is crucial here is less, how large a model, but how well the knowledge behind it is structured.

2. The real problem: Information overload, data clutter, and hallucinations

A common theme of the webinar: trust in information.

Hallucinations are often a symptom – not the cause.

Hallucinations – that is, false or fabricated content – ​​are not an exotic, fringe problem, but rather an everyday occurrence in many business scenarios. Interestingly, they are often not (just) a problem limited to a model, but a Context problem.

If the input is vague, contradictory, or imprecise, even the best model will deliver poor answers. Or, as it was more bluntly put:

"If you ask bad questions, you get bad answers."

Big Data without curation: More is not always better.

Hope still stems from the Big Data age: We'll just save everything – AI will sort it out. This is precisely what is causing massive friction losses today:

• Companies archive almost everything for decades instead of deciding what is truly important.
• For both humans and AI, this blurs the line between relevant knowledge and "data noise".
• The AI ​​has to fish out the few high-quality pieces of information from an unstructured, overflowing data pool – a game with bad cards.

The result: Models struggle to provide reliable answers because the input is neither curated nor structured.

“More content” is not a strategy

Around 40% of companies are already actively using AI, primarily for marketing, content creation, and chatbots. However, if AI is only used to generate even more content even faster, we will continue to clutter our digital landscape – without solving the underlying structural problems.

3. Humans and AI as a team: Human-in-the-Loop instead of autopilot

A central theme of the webinar was the question: How do we prevent employees from becoming "rubber stamps for AI spending"?

Skill skipping: When skills erode unnoticed

In large companies, it's already being observed that employees, after intensive use of ChatGPT and similar tools, hardly feel capable of writing a text "from scratch" anymore. Instead, they only respond to prompts, corrections, and approvals. This poses risks:

• Professional and linguistic skills are gradually diminished.
• Employees are increasingly relying on systems that they neither control nor truly understand.
• In extreme cases, a “WALL-E scenario” arises: people relinquish responsibility and degrade themselves to mere commentators on AI results.

High performers use AI as a sparring partner

The webinar also made one thing clear: it's not about demonizing AI – quite the opposite. Those who view AI as sparring partner uses it, benefits enormously:

• Better ideas through rapid variation and iteration
• Higher quality through structured cross-checking
• More output with the same or even better content depth.

The clear prediction: Employees and companies that learn to use AI thoughtfully and consciously develop their own skills will overtake those who resist the technology. and Those who surrender to it uncritically.

4. Knowledge Graphs & Neuro-Symbolic AI: From Data Pile to Corporate Knowledge

The most exciting question in the second part of the webinar was: How do we create the foundation for AI to become reliable, explainable, and controllable in companies? The answer: Knowledge graphs and neuro-symbolic AI.

"Simply dumping documents from the last 20 years into a vector database is not a strategy."

Instead, conscious action is needed. Curating data:

• Which documents are technically relevant and up-to-date?
• Where are the binding truths (policies, contracts, architectural standards, approvals)?
• What information is even allowed to be included in AI contexts (compliance, data protection)?

The recommended procedure: Cherry picking Only selected, high-quality information is transferred to the knowledge graph. This reduces costs, lowers environmental impact (less computational effort), and – most importantly – improves the quality of the answers.

What makes a knowledge graph so powerful?

A knowledge graph breaks down unstructured information – e.g., documents, emails, or tickets – into Entities (Objects, people, systems, processes) and Relations (who is connected to whom or what, and how). Instead of simply storing files, a network of:

• technical terms
• Systems and interfaces
• Responsibilities
• Rules, guidelines and exceptions

This linked information generates a “connected energy”:
The AI ​​no longer accesses disjointed text fragments, but rather a business model with context and structure.

Neuro-symbolic AI: The best of both worlds

Knowledge graphs are not a new idea – they originate from a time when AI was still largely knowledge-based. Today, they are combined with modern language models:

• Neural components (LLMs) provide language comprehension, generation, and interaction.
• Symbolic components Knowledge graphs provide structure, rules, relationships, and verifiable facts.

This neuro-symbolic approach AI makes:

• more explainable (answers can be traced back to specific nodes and edges in the graph),
• more controllable (rules and governance can be mapped),
• sustainable (knowledge survives generational changes of tools and models).

The webinar demonstrated, using an example, how a knowledge graph framework such as aikux.Brain or that the underlying graph technology can serve as the foundation for solutions like migRaven.MAX – thus intelligently connecting corporate knowledge, IT data, and AI.Migraines)

5. Technological sovereignty: Why Europe needs its own answers

Finally, the discussion focused on a strategic perspective: Do we want to make our corporate intelligence permanently dependent on just a few US or Chinese providers? The risks are obvious:

• Strategic dependence on individual platforms and clouds
• Uncertainty regarding data flows and usage rights
• Difficult starting point for regulation, auditability and compliance

At the same time, it is true that if employees reduce their own thinking and increasingly outsource decisions to (inexplicable) systems, an “extreme danger to the company” arises – both professionally and organizationally.

The webinar's message:
Europe must invest – into proprietary platforms, into explainable AI, into data and knowledge architectures that are not only efficient, but also sovereign and auditable.

Conclusion: Without data curation, there can be no responsible AI.

The webinar made it clear:

• The problem is not AI – it is how we handle knowledge and data.
• Those who do not curate their information landscape receive unreliable, difficult-to-explain answers.
• Companies that fail to empower their employees to work reflectively with AI risk losing important skills.
• Companies that structure their corporate knowledge – e.g., in a knowledge graph – create the basis for explainable, secure, and sovereign AI applications.

For companies, this means specifically:

1. Clean up and curate data instead of "simply save everything".
2. Establishing human-in-the-loop structures, so that expertise and AI can mutually reinforce each other.
3. Examine knowledge graph approaches, in order to make corporate knowledge modelable, linkable and auditable in the long term.
4. Securing one's own sovereignty – technical, organizational and strategic.