Asking the right questions is an art. Since the time of Socrates, asking for something in the right way allows you to get valuable answers full of information. And now, thanks to artificial intelligences such as ChatGPT (banned in our country), it allows you to do much more. But what, exactly? And how do you learn?
In the United States they are coming out job announcements by companies for what is called prompt engineer, a role that pays between $250,000 and $335,000 a year and is attracting the attention of many. And there are those who wonder what exactly a prompt engineer does and what kind of skills and qualifications are required for this role.
Artificial intelligence
The AI between racism and discrimination: two years later, we are worse off than before
by Emanuele Capone
Meanwhile, the meaning of the term: prompt engineering or prompt engineering, is the description of the task that the AI has to perform that is given in the form of a question, instead of being implicitly provided with traditional programming. It is not a random question or a wish to express to the genie out of the lamp: requires analytical skills and expertise both IT specifications and within which one intends to operate.
”Prompts are the way artificial intelligences are programmed – he told us Mattia Monga, computer science teacher of the State University of Milan and among the founders of Aladdininformation technology dissemination and teaching laboratory – It is a complex skill, because the person who builds a prompt must not only know what he wants, but also what characteristics it must have”.
Prompt is actually a term that has been around for quite some time: in English it means to suggest or incite, and in computer science since the seventies it indicates the invitation to type a command within a text interface of the computer. It was the flashing cursor with a series of symbols for writing commands on Unix or in MS-DOS (which, not surprisingly, was called “MS-DOS Prompt” for a long time within Windows). After forty years of graphical interfaces with icons and windows, ChatGPT, Midjourney and all the others conversational artificial intelligence they have brought back the importance of text (and in perspective of voice commands). Now knowing how to ask the right questions, i.e. the prompts, is back in the news. Indeed, it has become a real way to program the behavior of the computer without having to write code. And it requires skills that cannot be improvised.
The researchers saw that, for example if you ask ChatGPT to solve a math problem, the answer is systematically wrong. But if the question is asked to “analyze the problem one step at a time”, the answer changes and becomes right. Or, if you’re asking to create an image to achieve a specific result, you need to have both knowledge of the limitations of AI models (what they can and can’t do) and illustration knowledge to design the image: perspective, framing, subject, colors, style, and so on.
“There is nothing magical about artificial intelligence – Monga told us again – Instead, there is the challenge of specifying what we want in a structured and coherent way. ChatGPT and the other systems are still primordial than what can be done in the future, but they require the ability to analyze a problem and learn to express yourself in the right way to ask the machine to do something. You need to know how to think correctly and have skills in the area in which you intend to operate”.
In short, we don’t all automatically become writers, designers, programmers without already knowing how to do it, without having specific traditional skills. Just as the introduction of the calculator first and then of Excel hasn’t transformed anyone into an accountant: ChatGPT won’t make the difference on its own but, among people who are experts in a given sector, who knows how to use it and who doesn’t”.
Artificial intelligence
The novelty in the world of AI is called AutoGPT: what autonomous agents are and how to use them
by Francesco Marino
In the US, companies specify the skills required for the job as a prompt engineer: Bachelor’s degree in Computer Science or a related field, Structured writing and analysis skills, Knowledge of AI models, Previous experience in Natural Language Processing. And knowledge of the specific type of work they are going to do.
This changes perspectives on many things, including the skills needed by those who go to school tomorrow. Coding (i.e. the ability to program) is only a part of the cyber-skills of information technology, which now also includes the relationship with artificial intelligence. Aladdin’s laboratory in which Monga works is among the first in the world that they are researching ways to rethink computer science teaching in such a way that it also includes cyber-skills for artificial intelligence. It is a field of study on the border between computer science and pedagogy, whose immediate objective is to train the teachers of the future, those capable of giving the right cyber-skills to the students of tomorrow.
“With artificial intelligence it will be possible to do many things – Monga told us again – Gli archeologists eh philology they will be able to reconstruct the missing fragments of the ancient oi texts magistrates they will be able to summarize the complexity of the laws and sentences to arrive at more in-depth judgments. But, in addition to being experts in their disciplines, they need to learn how to construct prompts analytically and logically, knowing the capabilities and limitations of the systems they will operate on.”