AI Sounds Right. But Isn’t Always Right

 

September: back behind our screens. And with that comes another temptation: using an LLM to make everything quicker, easier and supposedly more efficient.

To continue on my prior posts on AI, the following is a question I have been thinking about: How are we using it?

Because there is one fundamental problem with asking an LLM for an answer: It sounds right. But it isn’t always right. Which leads to "epistemic reliability" which is essentially, how trustworthy something is as a source of knowledge. An LLM can give you an answer that is beautifully expressed, logically structured and completely wrong. Fluency can create an illusion of expertise.

Previously, if we wanted to know something, we might have asked an expert, consulted a textbook, searched several sources, read a newspaper or checked an academic publication.

Now, we can ask a question and receive a confident, well-structured answer in seconds.
The problem? The quality of the presentation tells us very little about the accuracy of the information.

But another issue concerns me even more: what happens when we stop doing the thinking ourselves?

If an employee asks AI to analyse a problem and recommend a course of action, have they developed their own understanding? Can they challenge the recommendation? Can they defend it?

If a student asks AI to formulate their argument, structure their essay and find their evidence, they may have saved time. But have they actually learned how to construct an argument?

This is where the conversation about AI needs to move beyond productivity. Saving time isn’t necessarily productive if the result is wrong and we didn’t notice because we didn’t challenge it.

Perhaps this is the paradox: AI can make us better informed while making us less inclined to investigate.

So perhaps the conversation shouldn’t be whether we should use AI BUT HOW WE USE IT without outsourcing our judgement.

Photo: Pexels - Tim Witzdam 

© 2026 Brid Doherty-Appriou. "Third Culture Professionals"® is a registered trademark. All rights reserved.


Brid is a native English speaker, fluent in French, a graduate in both International Business Communication and in Instructional Design. She is also accredited (TEFL) to teach English as a foreign language. Her professional experience spans over 20 years in an international environment, working in CAC 40 communication groups and as a consultant in Executive Search.