Prompt engineering: freely manipulate AI as you speak and as you wish.

Even artificial intelligence that can answer questions immediately can produce completely different results depending on how it is used.
Let's learn about prompt engineering that can be applied to various AI services and explore practical, job-specific AI applications.
In addition to asking questions to effectively use generative AI, you'll also gain insights into how to apply AI to your work and life.
Let's make AI your own by learning 16 prompt engineering techniques from AI experts and harnessing the full potential of AI.

Most existing repetitive tasks will be replaced by artificial intelligence.
We will be required to perform higher-level tasks, and there may come a day when competencies different from those demanded in this era will be important.
For example, while the ability to utilize artificial intelligence is not essential for general office workers today, those who cannot utilize artificial intelligence in the future may not be able to work.
There will come a time when we will have to find and do things that artificial intelligence cannot do, and people who are good at doing those things will be in the spotlight.
--- p.10

Since this is a task of adjusting the prompt value to obtain the desired value from artificial intelligence (to improve performance), it does not necessarily have to be text.
This is because artificial intelligence can receive various values, such as structured data, in addition to images and text.
However, most generative artificial intelligence that has recently shown good results provides various types of results, such as text or images, when a value containing text is provided as an initial value.
So I want to discuss mainly text-based prompt engineering.

--- p.40

When a lot of people are talking at the same time, the conversation can't proceed properly.
I don't know what to say because I can't quite understand what the other person is saying.
The same goes for artificial intelligence.
When too much information comes into a prompt at once, it is difficult to respond properly.
Especially when a lot of useless information comes in, the response is affected accordingly.
Therefore, it is important to eliminate useless information and structure the prompt concisely, focusing on key information.

--- p.66~67

When I introduced myself to an employee of an IT company, the words I used changed.
Instead of using familiar examples, use practical examples (customer support, automated response systems, document summarization and summary information extraction, internal knowledge sharing, chatbots, etc.) to appeal to usability and explain what technologies were used to develop it.
Prompt engineering is important because the results vary depending on the role assigned.

--- p.116

OpenAI asked the machine to create synthetic images of turtles and giraffes to show non-existent images.
Then, we created an image that combined the concepts of giraffe and turtle, such as a giraffe with a turtle shell or a turtle with the colors and patterns of a giraffe.
Of course, there's no such thing as a giraffe with a turtle shell, and no turtle with giraffe colors and patterns. DALL-E2 synthesized these two disparate concepts to create a new image.
This suggests that, as discussed earlier, concepts can be combined.

--- p.182

What we wanted was a software assistant that would automatically perform tasks that the user needed to do, using generative artificial intelligence like ChatGPT.
Of course, ChatGPT can also continue the conversation and remembers what was previously discussed to some extent.
However, it does not automatically perform tasks when given a goal.
So we started developing a system that would automatically accomplish the goal, and our AI agent, AutoGPT, started to get results.
So, when you give AutoGPT a goal, it automatically performs the tasks required to produce the desired result.
--- p.211

We can assume that in the future, artificial intelligence will become easier to use and will develop in a way that makes it more natural to encounter it in our lives.
If such a society comes, what should non-experts do? They must become the best leaders who can effectively utilize AI.
To do that, we need to use artificial intelligence extensively and introduce it into our work environment.
So, this book covers the types of super-large-scale generative artificial intelligence, examples of its use, and examples of artificial intelligence use by job.
In other words, we need to use artificial intelligence a lot and get a feel for how and when to use it better.

--- p.345~346