Issue link: https://maltatoday.uberflip.com/i/1532146
10 OPINION maltatoday | WEDNESDAY • 5 FEBRUARY 2025 IF you've ever felt that ChatGPT isn't delivering the results you expected, you're not alone. Whether it's providing overly generic answers or missing the mark entirely, many users walk away disappointed, dismissing the tool as overhyped. But here's a question worth asking: what if the issue isn't with the tool itself but how we use it? Artificial intelli- gence, particularly tools like ChatGPT, fundamentally dif- fers from the search engines we've grown up with. And while it's easy to treat ChatGPT like a Google replacement, this ap- proach will almost always lead to frustration. Instead, using AI effectively requires preci- sion, patience, and a thought- ful approach—more similar to fishing with a specific line in a precise spot than casting a wide net and hoping for the best. This distinction is critical. Search engines like Google fil- ter through extensive informa- tion to present the most rele- vant results. You type in a few keywords—"best restaurants in London"—and Google will re- turn a list of curated answers. It casts a wide net, pulling infor- mation from countless sourc- es, and lets you review the op- tions. ChatGPT, on the other hand, works differently. Rather than searching the internet, it generates responses based on the data it was trained on. Its ability to provide the correct answer hinges entirely on how you frame your request. In oth- er words, with ChatGPT, your results are only as good as the prompts you give. Unfortunately, this is where many users go wrong. One of the most common mistakes is being too vague or unclear in prompts. Asking ChatGPT something broad like, "What should I do for my business?" is unlikely to yield actionable advice because it lacks con- text. The tool cannot under- stand what kind of business you're running, what chal- lenges you're facing, or what goals you're trying to achieve unless you tell it. To use the fishing analogy, this is like tossing a line into the ocean without knowing what kind of fish you're hoping to catch. In- stead, a better approach might be, "I run a small bakery that specialises in artisanal bread. How can I increase foot traffic to my store during the winter months?" By being specific, you give ChatGPT the information it needs to generate tailored, valuable suggestions. Another mistake is treating ChatGPT like a one-shot tool. Many people stop after the first response, assuming that the tool is flawed if the initial an- swer isn't perfect. But ChatGPT thrives on iteration. If the first response doesn't quite hit the mark, you can refine your re- quest, ask follow-up questions, or add new details to the con- versation. For instance, if you ask it to draft an email and the tone doesn't feel right, you can say, "Can you make it more for- mal?" or "Could you add some specific examples to support the points?" This iterative pro- cess sets ChatGPT apart from static tools like Google—it's a conversational partner de- signed to adapt and improve based on your feedback. Another crucial difference lies in the tool's versatility. While search engines are pri- marily used to find existing information, ChatGPT can cre- ate entirely new content. It can draft essays, generate creative ideas, write poems, or debug code. But tapping into this ver- satility requires experimenta- tion. Many users stick to basic queries, like asking it to sum- marise an article or answer a straightforward question, with- out exploring its full range of capabilities. This is like owning a Swiss Army knife but only ev- er using the bottle opener. To get the most out of ChatGPT, you must push its boundaries, test its limits, and see what it can do. However, none of this is to say that ChatGPT is perfect. The tool has limitations, and under- standing them is key to effec- tively using them. For one, its knowledge is frozen at a certain point—it doesn't have access to the latest news or trends. Addi- tionally, it can sometimes pro- duce factually incorrect or sub- tly biased results, as its training data reflects the strengths and weaknesses of human-generat- ed content. This is why it's crit- ical to approach ChatGPT's re- sponses critically. Cross-check facts, refine prompts and don't rely on them for tasks requir- ing absolute accuracy. Like any tool, it's most effective when used with human judgment. So, how can you ensure you're using ChatGPT effectively? Start by treating it like a col- laborative partner rather than a static tool. Be specific in your prompts, providing clear con- text and desired outcomes. Engage in a back-and-forth dialogue, refining your queries and building on their respons- es. Experiment with its capa- bilities, from generating crea- tive ideas to solving technical If Artificial Intelligence isn't delivering, it may be your fault! Prof. Alexiei Dingli is Professor of Artificial Intelligence Alexiei Dingli