C-level AI skepticism
The most common (counter) arguments about AI and why companies should still start now
Artificial intelligence has left the phase of excessive expectations behind it. We are no longer at the peak of the hype cycle, but in a phase in which companies are questioning more critically where AI actually delivers added value. Right now, many organizations need guidance. In this article, we clarify the most important concerns and show why they are understandable, but do not necessarily speak against a timely start.
“AI doesn't bring any real added value and without a clear ROI, we shouldn't start anything.”
This criticism often comes about because initial initiatives fall short of expectations and positive financial effects are difficult to measure in the short term.
That still speaks for AI:
AI can already generate added value today, even if it is not always immediately visible. This includes faster response times, relieved employees, fewer errors and higher service quality. Although these effects do not always pay directly into the income statement, they improve operating performance. It makes sense to first check which processes are causing specific problems, instead of immediately calculating a comprehensive ROI. Small, clearly defined quick wins offer a realistic opportunity to test the benefits without taking major risks.
“AI costs jobs and endangers our culture.”
Many companies fear that automation will lead to job cuts or put a strain on existing structures.
That still speaks for AI:
Historical developments show that new technologies rarely completely replace people. Instead, they change activities, work processes and priorities. Many organizations are seeing that AI allows more tasks to be completed in the same amount of time, while increasing quality at the same time. As a result, roles shift instead of disappearing. It is important to get employees involved at an early stage, to create transparency and to show how activities can develop further.
“The models are getting worse again.”
There is discussion that large models are increasingly being trained on AI-generated data and could lose quality as a result.
That still speaks for AI:
Current models are already sufficiently powerful for most business applications. The central challenges today lie less in model quality than in meaningful integration into existing systems such as CRM, ERP or internal work processes. The practical benefits come from the interplay of model and environment, not from maximum intelligence in a chat window.
“The legal framework is too uncertain.”
The EU AI Act creates uncertainty for many companies, particularly with regard to future requirements.
That still speaks for AI:
Most business applications do not fall under the high-risk category of the EU AI Act. The legislator aims to create transparency and control and not to ban common AI tools. Organizations can continue to use AI securely, particularly when using hosted, compliant solutions such as Azure OpenAI. Transparency and documentation requirements apply in particular to sensitive applications such as automated application checks. Many requirements can also be introduced gradually.
“We're not a tech company. We can't do that.”
Many organizations feel overwhelmed because they believe that AI requires deep technical expertise.
That still speaks for AI:
Modern generative AI is comparatively low-threshold. Especially in areas such as customer service, communication or back office, ready-made models can be used without having to develop complex AI yourself. Clear project responsibility, internally (e.g. CTO or product owner) or externally via a service provider, is important for this. In the medium to long term, it is worthwhile to build up your own know-how in order to become more independent and to be able to better assess requirements. Corporate culture is just as important. When the C-level exemplifies openness to AI, actively communicates and encourages employees to try out new ways of working, there is acceptance across the company. Only this combination of technical competence, cultural support and visible leadership creates the basis for AI to have a lasting effect.
“Our data isn't good enough for AI.”
Unstructured, incomplete, or scattered data is often seen as an obstacle.
That still speaks for AI:
Generative AI can also handle unstructured data, for example through document processing or vector databases. Although clean, structured data is beneficial for analytical or automated decisions, it is not necessarily a prerequisite for getting started. A step-by-step approach makes sense: start small, work with a defined data section, improve it and then build up further use cases.
“I'm still skeptical. Shouldn't we just wait and see? ”
Many companies want to see how technology develops before taking their own steps.
That still speaks for AI:
Skepticism is legitimate, but waiting long term involves risks. Companies can already identify where challenges exist and carry out initial experiments. Controlled testing, feedback, and iterative improvements create practical knowledge that cannot be quickly made up for later. AI will not disappear, and organizations that build competencies early on will benefit from future developments much more quickly.
The right time for AI is now
The hype has subsided, and it is precisely as a result that the view of the actual benefits has become clearer. Modern models are powerful, reliable and versatile. However, the greatest added value comes less from the technology itself than from its targeted application in everyday life. Companies that are now trying out and starting to implement pilot projects and actively involve their employees are creating a solid basis for sustainable efficiency, better quality and modern work processes.
The article summarizes the most important findings from FOlge 5 of the podcast AI & Why — Business Talks together.
