It’s been a couple of years since AI came into our lives and since then, it hasn’t stopped surprising me. I don’t know if the same thing has happened to you, but I, being a very curious person and an obsessive about technology and digital things, couldn’t help but be subjugated by this overwhelming trend and convert to this new religion that is changing everything.
GPT 3 — Someone to write for me
When I started working as a copywriter, my mission seemed more like that of a machine chopping up text than that of a person researching a topic, drawing conclusions and writing for other people.
But that’s the reality when you decide to work 8 or 9 hours in a company as an employee and fulfil the functions of your job.
Because when quantity overrides quality, great things can be achieved. I have proof. I won’t name names, but I have seen it happen against all odds and certainly in spite of the divine laws designed by Google.
It is that moment when you realise that your job is not writing as such. It is more like the work of mass producers in an industry that has an expiry date. I realised that my job was to spend about 9 hours in the office and type as fast as I could, with enough sense for someone else to understand, the most basic SEO techniques for Google to value the content and make sure I completed 6 pieces of content a day.
So when OpenAI came out with their artificial intelligence and machine learning driven language model, I treasured that news to put my free credits to the best use I could at the time.
I had to test how this little wonder I had just discovered could help me, and it worked. My life as a slave to non-writing was saved. From that discovery, I could optimise my time and stop looking like a monkey typing, as my partner at the time and today, one of my best friends, told me.
ChatGPT — Easier and more convenient to use
It was only a short time since I used up that free credit; just long enough to wake up again: ChatGPT, the democratisation of AI in its first phase of life.
This is where I started to have a closer relationship with this new wonder and to keep exploring to get as much out of it as possible. In my case, I was trying to ask it directly for its knowledge of a topic and to enhance its quasi-human writing skills.
It wasn’t bad at all, although it lacked nuance and, of course, I had to go over what I was putting into 600 words with the SEO parameters I was introducing him to. Although I was concerned about the “excellence” of my work in revising and modelling such robotic writing, it was already being useful for me.
ChatGPT with GPT4 technology — It seems to understand and process the information better than I do.
I really learned a lot during the phase when only GPT3 could be used. Mostly I learned how to ask questions; what is now called “prompt engineering”, translated as getting the answer you are looking for by asking the right question.
By this, I don’t mean to belittle those people who have found a niche in prompt engineering as a job or to take complexity out of the question. I recognise that it is an intensely dynamic science with few certainties at the moment.
Let me give you an example with ChatGPT:
My prompt will be: “As a podcasting expert, try to analyse the market, choose a potentially successful niche, the most suitable platform to start the podcast, the most suitable title to get the maximum audience, the frequency of content it needs to have in order to monetise in the minimum possible time and the 3 best monetisation strategies, with concrete examples of tools and their practical application. I want you to present all this information in a reduced form, in 500 words and through paragraphs and bullet points that make it easy to read. Don’t forget to present truthful and accurate information that will help other people to start a profitable business, so you will have to justify your answer properly”.
Do it yourself and you will see that your answer, however similar it may seem, will not be the same in any case, but it will contain common elements. This is due to the prompt engineering we have applied, which includes the structure, a knowledge base requirement (podcasting), a requirement for efficiency (to be successful, to monetise in the shortest possible time) and the parameter of truthful and accurate information.
I show you the answer I got:
Copilot, ChatGPT and Gemini — An oracle to turn to if you know how to ask
My intention is not to offend anyone, of course, but as you will understand, without the title of my post, you wouldn’t have come in to see it, with such a pleasant surprise to find a practical example of prompts engineering.
The thing is that I want you to know that, even so, the title is still related to what I’m telling you. Why? If we go back a few lines, I was talking about my experience with AI through GPT, ChatGPT, etc., and how I modified or nuanced my use of it.
This leads me to stop for a moment and look at other people’s experience of using this technology as if it were an oracle. They ask a question and expect a “right” or at least an “adequate” answer.
Have you ever tried asking a random person on the street what the formula for relativity is? It depends on many factors whether you get the “right” answer. For example, whether that person has studied science or simply remembers that formula that he or she studied for a particular exam and then probably forgot to make way for other knowledge.
In my experience, in order to get a correct or appropriate answer from the AI, the first thing we have to do is contextualise. I am sure that if you contextualise the random person you have asked, the chances of them remembering the correct answer increase exponentially. You might make them remember that formula they thought they had forgotten. But the difference, no doubt, is in the context, and the context in this case, applies to the question or prompt, whatever you want to call it.
No, AI is not the Messiah foretold, nor is it an oracle to go to and prove how good it is and what you have to fear because it will take your job. AI is just another tool that, if you know how to use it, you can get a good summary of a book, brainstorm ideas for your next book or simply make some aspect of your life, such as work, more efficient.
I hope you find my thoughts useful and helpful.
Best regards.