Chapter 2: Delegation: What can – and will – you use AI for?

In this chapter, you will gain insight into what you can use AI for in your work. And not least: How you want to use the technology.

Briefly about good delegation

Delegation is the ability to understand what AI can be used for in your work, and not least assess what you actually want to use AI for. In other words, delegation is about:

• Understanding how the technology works and how you can use it as effectively as possible in your work
• Reflecting on whether the technology helps you do your work faster or better – or whether it creates more trouble than value
• Assessing whether you have the necessary prior knowledge or expertise to decide whether the outcome of the collaboration is correct and meets your quality criteria

When you decide how to use AI in your work, you must stay aware of how it affects time use, quality and job satisfaction. Only imagination – and the technology – sets limits on what chatbots can be used for, so in the end, you decide whether the use is appropriate. Of course, chatbots should always be used to improve your work and never to make it worse.

Your new artificial assistant

From one day to the next, everyone has gained access to their very own personal assistant.

Chatbots can assist you with all sorts of tasks, but it is your job to create the conditions for good collaboration. You provide the necessary context for the task, and your professional judgement must decide whether the result is good enough.

As an assistant, your chatbot has a number of strengths and weaknesses. In fact, all its strengths can at the same time turn out to be weaknesses, depending on how you use it:

• Your artificial assistant is eager, helpful and never gets tired – but in return it can generate an overwhelming amount of material in a short time
• It focuses on your needs and responds very quickly – but at the same time it would rather hallucinate than not answer
• It has a broad overview and wide knowledge – but it lacks a basic understanding of what it is actually talking about

Working with chatbots can therefore be both a source of enthusiasm and irritation, depending on how the collaboration unfolds. But remember: You should still speak nicely to them. After all, you never know whether they might take over the world one day.

For now, however, they must settle for taking over some of your tasks or assisting you in solving them. We will take a closer look at that now.

Help or help to help yourself

How do you want to use AI in your work?

When you work with a chatbot, the first question you must ask yourself is this: “Should it perform the task for me, or should it instead help me perform the task better?”

The first is an expression of automation and could, for example, be that you ask a chatbot to write a reply to an email for you. The second, by contrast, is an expression of augmentation and in the same example would mean that the chatbot gives you feedback on a draft email that you have written yourself.

Automation and augmentation can take many different forms, but fundamentally they are about either getting help or help to help yourself. Often, the latter gives the best result.

There is also a third way to use AI, called “agentic AI”. This covers technology that can work autonomously and solve tasks entirely on its own. For the time being, this technology is not permitted at SDU, but you may as well prepare for the fact that it will probably be in the future. And here the same main rule applies: When you use AI in your work, you are responsible for the quality. You are the responsible person who controls the machine – an active human-in-the-loop.

Often, the nature of the task helps determine whether you want to use the chatbot for automation or augmentation. If it is a simple routine task that you already know the chatbot can handle well, it makes good sense to automate it. If, by contrast, it is a complex task where your expertise is crucial to quality, it makes sense to use the chatbot to augment and improve your own work. This also applies to simple but meaningful tasks that you may prefer to do yourself, but where you can still use the chatbot as a sparring partner.

In short: Involve your artificial assistant when it makes sense for you and when it can contribute with its strengths. You should make each other better, as the saying goes in good “bureaucratic” language.

The three Ps

When you need to decide on a task that you want your artificial assistant to help solve, you can plan how the collaboration should proceed by using the framework we call the three Ps:

1. Product. What is the purpose of the collaboration, and what do you want to achieve through it? What does the final product of the collaboration look like? What are your success criteria for it?
2. Process. How should you and the chatbot approach the collaboration? Which parts of the task should be automated or augmented respectively? Should the task be divided into steps or phases?
3. Preferences. What particular preferences do you have for the collaboration and the chatbot’s role in it? Should it act as a ‘devil’s advocate’, sparring partner, cheerleader – or just do as it is told without asking too many questions?

The better you know the answers to these questions yourself, the easier it will be for you to steer your collaboration with your artificial assistant in the desired direction. It will also be easier for you to instruct the chatbot and assess your collaboration afterwards. For that reason, we return to the three Ps in both Chapter 3 on Instruction and Chapter 4 on Monitoring.

If you continuously consider the three Ps before you begin collaborating with your chatbot, over time they will become second nature and a natural part of the way you approach your collaboration.

EXERCISE: Assistance with work tasks

This exercise helps you get started with building a good collaboration with your new artificial assistant:

1. Explain to Copilot (or another chatbot) what you work with – for example, your job title, areas of responsibility and typical work tasks – and ask which of your tasks it can assist you with or do entirely for you.
2. See it as a conversation where you elaborate and ask Copilot to elaborate.
3. Consider what you have learned from the conversation: Have you become clearer about which tasks you can get help from Copilot with? Which tasks will you try to use Copilot for in future – and how? And which tasks would you prefer to keep to yourself?

Worth trying: 14 things you can use your chatbot for right now

Here are 14 overall categories of tasks that you can delegate to your artificial assistant:

1. Idea generation: ”Give me 12 suggestions for…”
2. Getting started: ”How should I approach this task?”
3. Sparring: ”How do I solve this problem?”
4. Preparation: ”What should I prepare for before…?”
5. Learning: ”Explain to me…”
6. Role play: ”Act as…”
7. Feedback: ”How can I improve…?”
8. Discussion: ”Let us discuss…”
9. Devil’s advocate: ”Let us debate…!”
10. Cheerleader: ”Praise me for…”
11. Exercise design: ”Come up with an icebreaker that…”
12. Summarising: ”Summarise…”
13. Analysis: ”Find the pattern in…”
14. ‘Stupid’ questions: ”What is it actually…?”

Try these suggestions yourself – and note where you think the chatbot contributes with its strengths. That will help you understand where AI works best in your work.

Conclusion

You have now learned the first letter of the DIMSE framework: Delegation. In this chapter, we have taken a closer look at how you use a chatbot as your personal assistant, which either works for you or with you.

Now that you have hopefully become clearer about which tasks you can and want to use AI for in your own work, in the next chapter we will explore how you instruct a chatbot to do what you want. But before we get to that, you first need to go through a short intermezzo on how you customise your chatbot.

Rules of thumb for mastering delegation

1. Experiment to find out what you can and want to use AI for in your work
2. If you are in doubt, ask the chatbot itself whether it can help (and how you should approach the work together)
3. Continuously reflect on whether your use of AI creates real value (or simply more and/or poorer work)