A recent report from Anthropic highlights an important finding that is highly relevant for those of us working with AI tools like ChatGPT and other generative models: Good prompt = good output, bad prompt = bad output.
This may seem obvious, but the report underscores that the quality of the instructions we give AI largely determines what the AI can produce. In fact, research shows that the best results come when the user has both deep knowledge and experience in the subject being asked about. If two people with different levels of experience were to test who could create the best social media strategy using ChatGPT, the more experienced person would get the better output in 99.9% of cases. Why? Because the input from an experienced user is based on their knowledge, experience, and expertise.
The quality of prompts: A key factor in AI's effectiveness
AI has only one task: to deliver the best possible output based on the input it receives. This means the AI can only be as good as the information it is given. Experienced users understand how to formulate their prompts to get AI to generate precise and relevant output.
A less experienced user who perhaps provides an unclear or less precise prompt will not achieve the same quality in the output. This is why AI works best when it receives the right input from someone with the necessary knowledge.
Why is it important to understand prompts?
For many users who are not accustomed to working with AI, it can be a challenge to understand how to structure effective prompts. A good prompt is not just a short and precise request; it is built on background knowledge, relevant details, and an understanding of what may be important to highlight in a given task.
For example (exaggerated to clearly illustrate the difference) :
Bad prompt: “Create a social media strategy.”
Good prompt: “Create a social media strategy for an e-commerce company selling sustainable products. Focus on targeting young adults aged 20–35 in Denmark and Sweden. Use concrete examples of content types, platforms, and a timeline.”
The latter prompt gives the AI much more to work with and is far more targeted, increasing the likelihood of a relevant result.
How experienced users win with AI
Experienced professionals find it easier to leverage AI, as they understand how to formulate targeted prompts. They can quickly adapt when the output is not optimal, and they can adjust their input to better match the AI's capabilities. For less experienced users, it may take longer to achieve the desired results, as they may not provide the precise instructions the AI needs to generate quality output.
AI acts as a tool that amplifies an experienced user's abilities. A person with experience will be able to use AI to maximise their results, while a new user may need to learn how to structure their prompts in order to unlock AI's full potential.
Perspective – A good prompt is a skill
Mastering the art of crafting effective prompts takes time and experience, just like any other skill. It is about understanding that AI can only perform as well as the instructions it receives.
So the next time you use AI, remember that it is not just about asking a question – it is about asking the right question in the right way.
