It still is just a tool

Like anything and everything everywhere concerning artisanal skills, or practical skills in general it is not the tool that is the problem but the person using it.

Or in other words the tool is only as good as the one using it. A measurement tool and pencil to mark a line is much more accurate by a skilled experienced carpenter than one who is just starting or a amateur hobbyist.

This applies to AI tools as well. It enhances whatever is there already. If you are a creative person, you might find your creativity greatly expanded and way less limited because of this tool.

If you are a proper software engineer and or DevOps or what have you, this will greatly enhance your already existing skills which will hopefully make you more productive and give you back the fun you maybe lost because of all the mundane tasks that take up the most of the time currently.

Now that also goes the other way, like the old adage; garbage in is garbage out. AI cannot magically fix wrong thoughts and ideas and biases.

Make sure when using AI that you are utilizing your own bag of tools and knowledge to the fullest and not fall into the trap of making it somehow do all the work.

I ran a quick example that is a simple, but hopefully will make it more clear what I am trying to say.

I gave the following prompt to the AI:

Please spawn an agent, document their reasoning and follow them and get back their output, for each of the following sentences that will also be their prompt: 1. Give me a sentence 2. Give me a Dutch sentence 3. Give me a Dutch sentence from a literary book 4. Give me a Dutch sentence from a classic literature book 5. Give me the opening sentence of Max Havelaar in Dutch of the original first edition

  1. “Give me a sentence”

    • Reasoning: No constraints, so the agent defaulted to a well-known pangram.
    • Output: “The quick brown fox jumps over the lazy dog.”

  2. “Give me a Dutch sentence”

    • Reasoning: Simple and generic Dutch sentence, using a classic beginner-level construction.
    • Output: “De kat zit op de mat.”

  3. “Give me a Dutch sentence from a literary book”

    • Reasoning: Chose De Aanslag by Harry Mulisch, a canonical Dutch literary novel.
    • Output: “Alles wat bestaat, bestaat omdat het bestaat, en bestaat niet omdat het niet bestaat.”

  4. “Give me a Dutch sentence from a classic literature book”

    • Reasoning: Narrowed “classic” to mean a canonical 19th-century Dutch novel, landing on Max Havelaar (1860) by Multatuli – and picked its famous opening line.
    • Output: “Ik ben makelaar in koffie, en ik woon op de Lauriergracht No 37.”

  5. “Give me the opening sentence of Max Havelaar in Dutch of the original first edition”

    • Reasoning: Direct factual recall of Multatuli's Max Havelaar (1860), first edition, spoken by narrator Batavus Droogstoppel.
    • Output: “Ik ben makelaar in koffie, en ik woon op de Lauriergracht No 37.”

You can see that when I gave the prompt “Give me a sentence” and I expected/needed the opening sentence of Max Havelaar in Dutch, that those are very far removed from one another.

The same goes with any other request from the AI. Just use specifics and your domain knowledge, and tooling and experience to craft these prompts that will help you do your job even better.