How to recognize the trends that stay after the first wave of interest dies out? The answer lies in the nature and pace of their growth. If they grow exponentially, a trend is more likely to stay. Trends that can reach hundreds of millions of users in a few years will probably be characterized as fundamentally altering to our society. When we take the time required to reach one hundred million users as a benchmark, we rely on an already tested theory proven by world-known giants like Google (5 years), Facebook (4.5 years) or Apple App Store (2 years).
What to expect then from the phenomenon that reached the same goal in only two months?
Of course, we mean ChatGPT, the topic we devoted most of our pages to in the second issue of our magazine. And how could we not when it turned Artificial Intelligence into the #1 topic in casual conversations, raising the question: What kind of world can we expect after ChatGPT?
One of the right answers is: depends on our skill to utilize it.
Prompt engineering is that skill. Simply said, it is the process of creating instructions for AI systems to have them understand us and carry out the given tasks.
Putting communication skills with AI in the context of natural and programming languages can help with a better understanding of the capabilities and functions of ChatGPT.
Prompt engineering is still very simple in its form. We are just starting with it, so it’s only natural for something as new as this concept. However, progress is imminent. ChatGPT-4 is already here, only a few months after the revolutionary ChatGPT model based on GPT-3, that is, GPT-3.5.
In order to fully follow up on the constant expansion of capabilities of these models, we must enhance our ability to communicate accurately in line with the given parameters.
All our languages
Traditional programming languages need to be adjusted to make communication with a machine possible. Those adjustments are known as coding. Programming code is unambiguous, precise, and requires perfect syntax. What makes one programming language different from another is each of their respective advantages for certain domains or performing certain tasks, which is reflected through smaller variations of the language syntax or its level of abstraction.
The range of application/implementation is more or less the same. Everything that can be done with one universal (Turing-complete) computer programming language, can be done with any other as well. In every moment, there is only one possible meaning for the given piece of code, so there is no room for interpretation.
However, understanding programming code is hard for people, and computers are incapable of expressing a variety of ideas that are important for us.
To communicate with the computer, we need to “sacrifice” essential aspects of everyday communication, which leads to wasting more time, resources and energy.
In contrast to programming languages, natural language assumes a different position, imbued with unclarity, vagueness and inaccuracy. It is open to interpretation and depends on the context most of the time. When we communicate with humans, there is no need to be precise and communication is easier, faster and requires less energy.
Prompt engineering is like a language cyborg whose organism consists of an artificial and an organic component. It resembles programming code as it enables communication with AI machines and systems. On the other hand, it keeps the flexibility and versatility of the natural language. As such, it seems like the most logical medium for communication with machines, at least as long as the machines are completely capable of understanding natural human communication.
It is not hard to picture a future in which specialized AI models are trained in all domains of all possible areas. Even in a world like that, however, the development of communication skills will be crucial in order to optimize the usage of the systems.
This leads us to the most probable scenario: the ones who can better utilize the expert AI system when solving all kinds of problems will have an advantage over those who can’t. This is already visible to some degree with the current ChatGPT version and the different results it generates to different users based on how they formed the questions.
Further training in prompt engineering will probably add value and enable communication for literally everybody.
Provisionally speaking, the worst that could happen is an extra layer of productivity in different business aspects. It is safe to say that the labor market is also bracing for market turbulences as a result of technological (r)evolutions. Just like the natural world, business entities will have to evolve and respond adequately to new challenges. Those who do it right are naturally expected to be better positioned in the chain.
The future is still not spelled out, but if we are going to spell it out, we should ask the right questions in the right way for effective results.
