How to Work with Artificial Intelligence?

Author: Tomas Staniulis, AI Consultant

The large language model ChatGPT, created with the help of artificial intelligence (AI), seemingly celebrated its first birthday last November, and millions of inhabitants of the planet cannot imagine their daily lives without it. The creators claim that ChatGPT can answer more than 100 million questions, speaks 26 different languages fluently, and knows at least 25,000 words in each of them.

Working with ChatGPT: 7 Tips on how to work with Artificial Intelligence

Just the first five days of life brought ChatGPT 1 million users, and after 2 months, more than 100 million users were already using this app. Why did ChatGPT become a mass phenomenon? Most likely because this language model is simple, accessible, and at the same time reveals the boundless greatness of the AI world.

Today, ChatGPT is used by everyone: from schoolchildren to experienced programmers, from students who create their master’s theses using sophisticated commands, to lecturers for whom understanding artificial intelligence becomes a matter of honor.

The History of Artificial Intelligence: How Did AI Become Trendy?

Have you ever thought about a movie before turning on Netflix, and then the TV screen already shows what you would like to watch? Congratulations, AI has successfully guessed what you want! Well, if it didn’t guess, it’s likely to be trained further until it can do so.

The interesting history of AI tells about the discoveries of Alan Mathison Turing, a computer genius of the early 20th century. Already in 1935, he first described an abstract computing machine, consisting of infinite memory and a scanning device that moves back and forth through the memory decoding information and making decisions, symbol by symbol. Later, the Turing test he developed, which identifies whether we are communicating with a human or a computer program, still finds followers to this day.

The first AI conference took place in New York in 1956, and in 1964, the first prototype of a conversational program, ELIZA, was released. The program operated according to a special dictionary and a set of rules that dictated how to react to different questions posed by users. For example, if a user asked “How are you?”, ELIZA could reply: “Thank you, it’s nice that you asked.”

Over the years, AI has become more powerful. At the end of the second millennium, the development received a new impetus from faster processors, video cards, and the ability to process more and more data. Evolving programming languages like Python, SQL, and C++ created ever-larger databases, and AI could process increasingly more data. Thus began the era of big data.

Just 40 years passed, and in 1997, IBM’s chess computer Deep Blue defeated the world champion and grandmaster Garry Kasparov. Humanity experienced its first major defeat against AI in a game of 36 moves, and in 2016, the chess champion consoled himself with the fact that today, a chess app that is many times stronger than the famous Deep Blue of the last millennium can be downloaded from the Google store.

Today, AI based on deep neural network analysis is far from the few rules learned by ELIZA, but a network that mimics the activity of human brain neurons. Deep neural networks generate responses, taking into account the commands given and their context – they recognize objects, language, sound, image and are massively applied in biometrics, e-commerce, medicine, and marketing.

When I first tried ChatGPT, like many, at first I wanted to trick the system, show its imperfection, or even mock it. As if to emphasize human superiority. However, daily life showed that ChatGPT is a powerful system that can generate interesting and valuable responses, but it does not have real-life and personal experiences. Therefore, it should be seen not as a bogeyman or friend, but as an assistant who can quickly complete any task. And the quality? Try it yourself!

7 Tips for Working with AI

1. Have a clear goal. Define what result you expect – to create a business plan for the quarter or perform a market analysis? Get specific information or ideas for inspiration? Answering this question will simplify the search and help you find the first command: “My goal is to create a sales strategy for product X, suggest an action plan.”
2. Use specific prompts (commands). There are already many tips on the internet on how to formulate queries, commands, and questions for ChatGPT, but first and foremost, pay attention to specificity. Specific commands, such as: write, create, answer, invent, analyze, help the language model understand the query better and generate clearer content.
3. Provide more context. If your query is related to specific information, include it in your question. You can copy a part of the relevant text or document and let ChatGPT get acquainted with the content. Give the command: “Analyze and learn.” After analyzing the text, the language robot will help generate more accurate and context-sensitive answers.
4. Use system commands. The ChatGPT language model is designed to be able to adjust its behavior. For example, the command “You are a history expert” will direct the language model in the right direction when asking questions in the field of history, the command “You are a music expert” will allow you to find out what answers the language model can provide in this area.
5. Edit the answers. Do not forget that the language model is not perfect, so it generates answers that need editing. When editing content created by ChatGPT, always check for factual, spelling, punctuation errors, and style inconsistencies.
6. Provide feedback. Task assessment and criticism are integral parts of digital intelligence learning. By noting errors, using commands to clarify, shorten, highlight the essence, and summarize, the result can be further improved.
7. Some interact with AI as with a small child: “Hello, thank you”, others – as with a good friend: “Please write a speech for a birthday occasion.” Pleasant communication with ChatGPT reflects your social habits more than it helps to get a better result. Remember, it’s just AI. There is no need to attribute human qualities to it; the clearer the command – the better.

Author: Tomas Staniulis, AI Consultant

© Any copying, distribution, or republication of this article without the written consent of the author is prohibited.