To err is human... and artificial: let's put ChatGPT back in its place!
How to use ChatGPT wisely? With such a powerful tool, we may be tempted to delegate more and more tasks to Artificial Intelligence... especially the ones we find boring. By doing so, we run the risk of not using our brain enough. Not to mention the errors made by ChatGPT: because AI is far from infallible. You will see that, for the moment, we can only rely on it in moderation.
Ease has a sweet taste. Succeeding in accomplishing a large quantity of tasks in a minimum amount of time and with a minimum amount of effort... a human philosophy that constantly drives us towards progress?
Smartphones, and now AI, lead us to use certain areas of our brain less. Memory, for example, since we note everything on our phones instead of making an effort to remember the information. Who still remembers their parents' and friends' phone numbers? So when we lose our "second brain," it's panic.
AI reinforces this mechanism of minimal effort, as it is capable of performing tasks for us that seem uninteresting.
Let's take a classic example: writing an email to apply for a job offer. A simple operation for an AI. From a human perspective:
- Either we copy-paste a template that we use all the time.
- Or we write a specific email, spending a lot of time trying to find the best turn of phrase.
In any case, writing a job application email is (very likely) a waste of time and effort. The risk of getting a negative response - and having wasted time for nothing - is high. Nevertheless, should we eliminate this task from our daily lives if the opportunity arises? Is it beneficial for our brain?
Let's compare all these tasks that seem useless - or unproductive - to sports:
- In itself, running on a treadmill in a gym is useless and wastes energy.
- Running for 1 hour a day around a lake seems just as useless.
- However, pedaling on a bike to generate electricity is a little better.
Yet sports activities, despite their apparent uselessness, are essential to the proper functioning of our bodies. They keep us in shape. Well, it's the same thing for the brain: it needs varied activities, productive or not, and sometimes even boring ones!
How to get the most out of ChatGPT?
First of all, use this AI with moderation. If a task only takes you a few minutes, do it yourself. This technology consumes far too many resources to be used for such small tasks.
Thus, I suggest considering ChatGPT as a writing tool, rather than a big brain filled with knowledge.
Feed ChatGPT with your own data (interesting and sourced), and ask it to transform them into something useful (concise summary, LinkedIn post...). This way, ChatGPT won't retrieve any erroneous data present in its corpus. The AI only puts its writing capacity at your service.
After obtaining a satisfactory result, you can still ask the AI to improve it according to your needs (more synthesis, humorous tone...). But use this tool wisely.
If your goal is to use an AI to replace a provider or an employee, think twice. You are about to waste a lot of time. Everyone has their own job, and AI is clearly not ready to replace humans.
When ChatGPT messes up
In a few months, the AI has been put to the test in many fields. With more or less success, factually speaking. Here's a little anthology, for fun:
If only cat eggs existed...
In the above case, the error is intentional: the question asked is deliberately wrong. However, the AI searches for potential answer elements and outputs a text. The artificial intelligence looks for the identified keywords in its data corpus. Then it plays with statistics to obtain the best information to output. Finally, it uses its writing model to produce a well-formulated text. Too bad that cat eggs do not exist, we almost believed it.
Mr. Mayor, where are you?
Lost: Gérard Collomb is no longer the mayor of Lyon since July 2020. Moreover, if asked directly, ChatGPT knows very well who the current mayor of Lyon is:
This error is easily explained: ChatGPT specifies that it does not have a sufficient amount of information about the world and post-2021 events.

Therefore, the AI takes its references from a set of contents where Gérard Collomb is clearly more mentioned than Grégory Doucet as the mayor of Lyon.
Even better than exam bloopers! That makes us want to use ChatGPT with moderation. Share your best blunders in the comments, we will discover them with undisguised pleasure 😉
[Cover photo: Lysander Yuen]
Tell us about yourself
Do you use ChatGPT?
