Beyond words: The power of prompting to communicate with AI

How important is non-verbal communication in ensuring mutual understanding in conversations between two people?

To be sure, I asked ChatGPT, and it provided me with many good arguments that reinforce the idea that communication isn’t just about words. Our gestures have significant value because they complement and reinforce the verbal message, convey emotions and feelings, facilitate message interpretation, identify contradictions and dishonesty, and foster connection and empathy.

As you can see, these are very logical reasons that any of us could also come up with. The difference is that Artificial Intelligence provided these answers in a matter of seconds, while it would have taken me quite a while to think of all these things unless I were a communication expert.

Experts were the ones who designed the systems on which AI relied until the invention of transformers in 2017, a technology developed by Google that enabled the rise of Generative AI. Before, if we wanted to have a conversation with a chatbot, it would have been necessary for a linguistics expert to pre-program or pre-prepare the most common conversations that people might have with the machine.

When it came to making cars drive themselves or detecting suspicious elements in an X-ray, the situation was the same: despite having machine learning and deep learning, expert intervention was still necessary, usually software developers who could speak the language of machines. Natural language processing and image recognition techniques were significant advancements, but they were only accessible to a few.

blog_ia_gestos

The big change

Since 2022, the emergence of tools like DALL·E, Midjourney, ChatGPT, Gemini, Claude, Runaway, or Magnific has driven a major shift. We’ve moved from being passive observers of what AI experts did and made available to us through recommendation systems or spam filters to becoming active participants who can use these tools in a personalized way.

For example, we’ve spent years being observers of what AI does when Spotify recommends music or Netflix suggests movies. But we couldn’t change anything; we couldn’t tell them we didn’t like a certain genre or wanted more of another. Only our behavior afterward could somewhat influence the recommendations.

Prompting: How we xommunicate with these tools makes all the difference

Now, things have changed significantly, and that’s why everyone is so excited. Anyone can access GenAI tools to request all kinds of content to be created according to their preferences, whether it’s text, images, audio, video, 3D, and obtain results that were unimaginable just a year ago. And what has made the difference is the concept of Prompt, which is how we communicate with these tools.

We’ve replaced computer code with natural language. Now, we can give instructions to the machine using our own words, just as we would to another person.

But beware, this can be tricky: we often think communication works like magic, that the people we talk to perfectly understand us, and that we understand others, but that’s not always the case. In fact, this is one of the reasons why nature has deemed non-verbal communication necessary, sometimes, words aren’t enough.

The problem lies in thinking that just because we speak, others understand us, and vice versa, without ensuring that we’re using the right words and giving the correct instructions. That’s why so many methodologies have been developed over time, like those for project management, to ensure that work dynamics can avoid some of the problems caused by poor communication.

Expresing in detail

When writing source code, this can’t happen. A programmer who doesn’t correctly express instructions in the software they’re developing won’t get it to work and will have to make a greater effort to write the code. In our case, when we ask GenAI for any task, we need to do good preparatory work when writing the Prompt. The machine understands us but doesn’t know us; it can’t get inside our heads, so the results will be disappointing if the instructions aren’t properly expressed.

We know ourselves and know what we have in mind. That’s why, when we request something in any old way, we might be disappointed with the response. A small effort is needed on our part to clearly express what we’re thinking so that the machine can understand us and give us the best results.

Think about when we ask it to create an image. There are countless things we can express in the prompt: from the design style to the type of lighting, the texture of objects, the characteristics of the setting, the historical era… The best prompt creators can do incredible things; those who go with the first thing that comes to mind get mediocre results and think this technology isn’t all that impressive.

When we apply this situation to our work, it becomes even more apparent how essential it is to write good prompts. It’s what can make the difference if we want this technology to help us make a leap in productivity and creativity.

blog_ia

Tips for interacting with AI

To start, it’s advisable to tell the text-based GenAI tool you’re using to assume a specific role. For example, ask it to be an expert in the field you’re working on to ensure more accurate results. Then, provide Context: the more information the machine has, the more personalized the result will be. In fact, as it gets to know you, it will offer more context and better results, especially when it can access your documents. Finally, provide the instruction itself, what you want it to do, write, summarize, design…

It’s also essential how we relate to this technology. For starters, it’s best to consider it a Co-pilot. It should accompany us in our daily work to help us do things faster and better, but initially, we shouldn’t delegate part of our work to it because it will lack much of the context. Then, it’s interesting to start learning new things using this technology, as if it were a kind of Mentor guiding us in the learning process. Finally, with acquired experience, we can think about which tasks can be automated in Auto-pilot mode because we know it can perform them perfectly. Although this process currently requires supervision, major tech companies are working on developing intelligent agents capable of chaining tasks without human oversight.

Prompting: The art of communication

To understand the value these tools can offer us, we need to experiment with them, and when it comes to GenAI, the quality of the Prompt is what makes the difference. Start by assigning the role you want it to play: for example, if you want it to be an expert for more accurate results or a teacher for more educational outcomes. Next, provide context—any information that will be useful to it in providing more personalized or tailored responses to your needs. Finally, be as specific as possible in the instruction of the task you want it to perform (write, translate, summarize, create, prepare…).

Whenever we want to use GenAI to increase our productivity and creativity, we need to understand that the better we express and communicate with the machine, the better the results will be. Mastering this communication is an art, and many prompting techniques already exist. At Tecnofor, we’re aware of this, which is why we offer training to improve your expressiveness and, therefore, your productivity:

Javier Martín - August 20, 2024 / Share it: