Some entrepreneurs and managers think of using AI for data analysis to reduce the cost of employees. But this practice has hidden costs that arise from an exquisitely psychological phenomenon, which can go hand in hand with others.
There are various cognitive distortions or prejudices but in my opinion there is one that makes the expert very angry when they hear someone who does not come from their field speak about their subject: the Dunning-Kruger effect, where the foreign individual from the subject, perhaps informed with a few Google searches, tends to overestimate his own preparation. There are many examples on Twitter that can provide second-hand embarrassment. During COVID we have seen many of them but we see many more of them in politicized scientific topics (e.g. nuclear energy, public health, anthropogenic climate change).
In this case the effect leads the entrepreneur or manager on duty to not to hire, fire or demote the data analyst or scientist because they think, after having heard, read or attended some events on generative AI, that it can replace the figure mentioned in part or completely. Or even developers.
It certainly increases productivity, but this depends a lot on WHO and HOW they use it. The generative part is writing code that performs the analysis. For example, a SQL query with the generation of some graphs that come from a python or R library. The convenience lies in transforming a request in natural language (Italian, English, etc.) into at least one programming language. Both the request and the result can be automated (e.g. once a week), resulting in the savings mentioned.
The hidden cost lies in using it in an elementary way, for example to make useless graphs of a single variable, something that even a secretary who deals with other things on Excel can do. Or the hidden cost lies in reinventing the wheel even though pre-packaged solutions already exist that best solve a specific problem.
Possibly the manager on duty receives the bonus because they have effectively reduced costs but has also removed functionality, in the first precedent case. And only a worrying ignorance on the topic makes us blind to this evidence. And some supplier companies cheerfully invoice thanks to this ignorance, even when they receive requests via tenders.
If you have an idea of using generative artificial intelligence for your company or the one where you work, Let’s discuss it first in a free call. It can save you from nasty surprises or toy tools with no business use.