In the business world, everyone is currently acting as if they are absolute AI experts. LinkedIn is teeming with "Prompt Engineers," and referencing ChatGPT has now become standard practice in meetings.
Yet the reality looks completely different
Behind the scenes, there is deep uncertainty about how we should actually approach Artificial Intelligence. A recent study by the University of Hohenheim now puts its finger deep into a psychological wound. The researchers led by communication expert Prof. Dr. Holger Schramm uncover two extreme, completely opposite behavioral patterns:
Laypeople trust AI far too much.
Experts trust AI far too little.
Both sides are fatally wrong. While some run the risk of being manipulated, others are missing out on the greatest wave of efficiency of our time.
Extreme 1: The "Overtrust Trap" of Laypeople
The empirical data from the University of Hohenheim shows a fascinating phenomenon: In private contexts and among people with little technological background, trust in AI systems tips almost into the mythical.
Laypeople tend to view Large Language Models (such as ChatGPT, Claude, or Gemini) not as software, but as an omniscient authority. They delegate existential questions to the algorithm:
They ask the AI for deep life advice.
They consult systems on highly complex moral and ethical dilemmas.
The problem: They do this blindly and without any critical distance. Because the answers from language models are always formulated eloquently, confidently, and grammatically without error, the technological facade is mistaken for genuine wisdom. Anyone who hands over control of moral or strategic questions to the AI makes themselves extremely vulnerable to misinformation and subtle influence.
Extreme 2: The "Ignorance Dogma" of Experts
On the other side of the spectrum are the specialists – and this is where the scientific study delivers perhaps the most surprising finding. One would think that people with high technical expertise would use AI systems particularly efficiently. The opposite is often the case.
In professional decision-making processes, experts regularly block advice from AI systems. Even when the data-based recommendation of the machine is demonstrably correct and superior to human gut feeling.
Why do experts react so skeptically?
Professional pride and ego: Accepting that a software program can recognize patterns in fractions of a second that take a human decades of experience to master scratches at one's own status.
The principle of reactance: Experts subconsciously resist the perceived patronizing behavior of a system.
The result in companies? Processes stagnate, data-driven market opportunities are ignored out of pure principle, and valuable resources fizzle out because "we've always done it this way here."
The truth lies in the middle: AI is a tool
Artificial Intelligence is neither an infallible digital healer nor a useless toy for tech geeks. AI is a tool. No more, but also not a single bit less.
Anyone operating a chainsaw does not stand in awe of it, but utilizes its power – while being extremely careful not to injure themselves. Exactly this healthy, pragmatic distance is missing in the current debate.
The winners of the digital transformation are characterized by a completely new core competence: Calibrated trust.
They know exactly where the AI fails due to hallucinations and lack of context (and examine results meticulously).
But they also know very well where the AI is miles ahead of the human brain in terms of scaling, data processing, and pattern recognition – and consistently use this leverage.
Conclusion: Time for a reality check
The scientific findings are a wake-up call for the economy. They show that the success of AI projects is not primarily a matter of IT infrastructure. It is a matter of psychology and education.
We must stop mystifying AI and start understanding it rationally. Only those who can objectively assess the functioning, the limits, and the real potential of these systems will ultimately maintain control.
The question is not whether the AI is right. The question is whether you are competent enough to properly contextualize its answer.
Where do you catch yourself? Do you tend to trust AI too blindly in some areas – or do you still sometimes block useful tools out of pure principle?
Scientific source for the article:
University of Hohenheim (Department of Media and Communication Studies, esp. Media Psychology / Prof. Dr. Holger Schramm). The official study results and details of the investigation can be viewed via the University of Hohenheim press portal.