ChatGPT: One tool to rule them all

Learning the basics of ChatGPT is crucial for deciding when to use it, and more importantly, when not to.

"Futuristic city" from DALL-E

Our Linkedin feeds are now full of our friends' experiments with AI tools.

Those who boarded the early train are already offering training, workshops and commercial presentations. The rest are sharing their learning journeys, which can be either interesting or quite dull.

Midjourney or DALL-E generated images are becoming more common. Nothing too impressive yet, but they carry the charm of novelty.

Most of these technologies are now being integrated by the software providers themself and are (or will be) included in their tools for our use - think of platforms like Canva, Adobe, Wix, Google Search, Bing and personal assistants like Siri. However, understanding the underlying technology, and more importantly, the origins of text, data and images is not a bad idea.

"R2D2, you know better than to trust a strange computer"

With that said, this is a super fast, new kind of assistant. I recently asked him to challenge my website and improve a baseline for one of my clients.

I also inquired about when to use, and not to use, ChatGPT. ChatGPT (which I use in French) gave the following response (I'll spare you the details of each point):

"Quand utiliser ChatGPT :

1. Recherche et Synthèse d'Informations

2. Aide à la Rédaction

3. Apprentissage des Langues

4. Divertissement et Créativité

Quand ne pas utiliser ChatGPT:

1. Informations Sensibles et Personnelles

2. Décisions Médicales ou Juridiques

3. Situations d'Urgence (santé)

4. Analyse Approfondie et Spécialisée

En résumé, ChatGPT est un outil polyvalent et utile dans de nombreux contextes, mais il doit être utilisé avec discernement, en tenant compte de ses limites et en évitant les situations qui nécessitent une expertise humaine spécifique ou la gestion de données sensibles."

The Ring in Die Walküre, soon in La Monnaie in Brussels.

I also asked if all articles might end up being very similar. ChatGPT agreed, much like a friend would at a cocktail party: "Indeed", it said, 'there is a risk of producing similar styles and tones, with limited creativity and a lack of personal perspective. ChatGPT learns from existing data but cannot create information beyond what it has learned."

The conclusion is always the same, let's use it as an assistant for research, to summarize topics, or generate a first draft. But afterwards, we should do the work ourselves, including the sources.

The war of personal assistants

A question remains: how long did it actually take? Between optimizing the prompts, rectifying, checking, correcting and adding, was it really a time saver?

One thing is certain, this will lead to a boom of enhanced personal assistants and skilled in-application helpers.