Talks on AI

A talk on AI should be about more than ChatGPT

For many, ChatGPT was the first real encounter with modern artificial intelligence. Suddenly you could hold a conversation with a computer, have it write text, analyse material, and come up with ideas in seconds. That was — and still is — impressive. But ChatGPT is only a small part of what is interesting about the development we are in the middle of.

AI can work with text, images, audio, video, documents, data, and software. It can be used as a sparring partner, analyst, research assistant, and creative collaborator. And the development keeps moving: from systems that mainly answer to systems that can increasingly act.

AI in the working day

Where can artificial intelligence actually make a difference between nine and five? That is perhaps the most relevant question for most businesses. We can talk about big technological revolutions, but AI only gets really interesting when you can relate it to everyday tasks: creating an overview of large amounts of information, analysing documents, preparing meetings, drafting first versions, rewriting communication, brainstorming, doing research, working with data, or helping with technical tasks.

The point is not that AI should take over the work. The point is to understand where the collaboration between people and technology makes sense. For some tasks AI is fantastic. For others it is mediocre. And for some it should not be used at all. That difference is the interesting part.

ChatGPT and the new AI tools

“How do you actually use ChatGPT properly?” is still a relevant question — but the answer is rarely just a list of 25 clever prompts. A good result depends on the context the AI gets, how the task is described, the material it works with, and how we assess the result afterwards. Depending on the audience, the talk can touch on tools like ChatGPT, Claude, Gemini, and other platforms — not to crown a winner, but to show what the category can do and how to think about it. The tools will change. The ability to understand the possibilities is more durable.

From prompts to AI that knows the context

The first years of generative AI became very much about prompts. Prompts are important, but AI becomes far more interesting when we give it the right context. If it is to answer a customer enquiry, it is relevant to know the company’s products, tone, and guidelines. If it is to analyse a document, it needs the document. It sounds obvious, but it moves AI from being a very intelligent guessing game to becoming a real tool. That is often an important point for organisations that have already tried ChatGPT but find the results vary a lot.

AI agents — when AI starts performing tasks

What happens when AI no longer just produces an answer, but can work through several steps of a task? Picture the difference:

The latter is suddenly not just about generating text. It is about workflows. This is where AI moves closer to automation, software, and the company’s other systems. AI agents are therefore an obvious topic for businesses that want to understand what comes after the ordinary chatbot. We can look at what an AI agent actually is, the difference between a chatbot, an automation, and an agent, when a human still needs to be involved, and what new risks arise when AI is allowed to act. It is not science fiction — but it is not magic either. And that distinction matters.

Can we get rid of some of the boring work?

A surprisingly large part of the working day still consists of tasks that follow almost exactly the same pattern every time: an email comes in, information is copied, a system is opened, a task is created, a colleague is notified, a document is saved, a confirmation is sent. It works — but it is not necessarily a good use of people’s time.

Here it is important to distinguish between traditional automation and AI. If a process follows fixed rules, ordinary automation is often cheaper, more stable, and more predictable. AI becomes interesting when the task contains something that used to be hard to automate: text, interpretation, categorisation, summarisation, or decision support. The combination can be powerful — and it is an area I find especially interesting, because it sits at the intersection of AI, systems, integrations and real work processes. A talk with this angle is less about “look what ChatGPT can write” and more about: why do we still do this task manually?

Vibe coding — when building software gets easier

AI is also changing the way software is created. Increasingly, you can describe a problem or a desired tool in plain language and use AI as an active part of building the solution. This is often called vibe coding. It does not mean that technical understanding has become unnecessary — but the threshold for creating small internal tools, prototypes, and specialised solutions is getting lower. The idea “it would be smart if we had a little system that could …” need not necessarily end in a six-month IT project. It is a relevant angle if the audience is interested in where AI is heading after ChatGPT.

When AI becomes a colleague — what happens to our work?

“Will AI take our jobs?” almost always comes up, and that is understandable. Generative AI can already handle parts of tasks that until recently required a human. But “will AI replace my job?” may not always be the most useful question. A better one can be: which parts of my work will change?

A job rarely consists of a single task. It consists of many — some routine, some requiring experience, relationships, professional judgement, creativity, or responsibility. Some should perhaps have been automated ten years ago. So a talk on AI in the workplace can also be about how our roles change when AI becomes a natural part of the toolbox: Which skills become more important? How do we avoid losing expertise? How do we check a result if we can no longer do the task ourselves? And when does greater efficiency turn into worse quality?

There is no single answer. But these are questions more and more workplaces have to grapple with — and they are often more interesting than yet another demonstration of what the technology can do.

Can you trust AI?

An AI system can deliver an answer in seconds, write it professionally, structure the argument, and sound confident — and still be wrong. That is one of the most important things to understand about generative artificial intelligence. So a talk can focus on hallucinations, source criticism, and human control: How do we assess an answer? When should information be verified? And what happens when people start trusting a system because it sounds competent? AI can be a fantastic tool. But it is a poor place to park your responsibility.

Deepfakes and a reality that is harder to verify

Generative AI is no longer only about text. We can create images of people who never existed, alter photographs, generate voices, and create video — and the quality keeps improving. That opens enormous creative possibilities, but at the same time makes source criticism and digital judgement more important. A talk can therefore also take a more society-oriented angle and address AI-generated images, synthetic audio, deepfakes, misinformation, fraud, and trust in digital material. It is an area that can both fascinate and worry — and one we all need to learn to navigate.

Data, confidentiality, and responsibility

Just because you can paste something into an AI tool does not mean you should. When AI becomes part of the working day, very practical questions arise: What information may employees share? Which tools does the company use? What do we do with customer data and internal documents? When should an employee stop and ask someone responsible? It is not the flashiest part of AI, but it is important. I try to handle these questions without turning the talk into a ninety-minute compliance course. The goal is understanding — that participants leave with both the desire to use the possibilities and a healthy sense of when to be careful.

The talk starts with you

A talk for a management team should not necessarily have the same focus as a talk for 150 employees. A business network has different questions than an association, and an organisation where almost no one uses AI yet needs something different than one where employees already use ChatGPT every day. So before the talk we have a short conversation about the event, the participants, and what you want to get out of it. That way the talk does not become just “something about AI” — it gets a direction.

You do not need to know exactly what you want to hear about. An enquiry can be as loose as: “We are holding a theme day for about 80 employees. The level varies a lot, and we would like something about AI that both inspires and makes us a bit wiser about what we can actually use it for.” That is plenty to start with. Then together we find an angle that fits the participants and the rest of the event.

Who is a talk on AI relevant for?

I give the talk in language where you do not need to be a programmer to follow along, and it can be adapted to very different audiences:

What should participants take home?

A good talk should do more than make people think “wow, technology is wild.” It may fascinate, but there should also be something left the day after: a better understanding of what modern AI can and cannot do, concrete ideas for tasks and areas where the technology could be relevant, a more nuanced view (AI is neither fairy dust nor the end of the world), a healthy critical sense — and a good starting point for the internal conversation about how you want to use AI.

I would rather show than tell

AI is an obvious topic to demonstrate live. Instead of only showing slides, relevant parts of the talk can include small, concrete demonstrations of current AI tools: How does the result change when the AI gets better context? How can it work with a document? And what happens when the AI gets it wrong? Live examples make the technology less abstract — it is easier to form an opinion about something you have just seen happen in front of you.

My starting point is practice

I work as a freelance digital consultant at the intersection of technology, business, and people — day to day with AI, automation, integrations, web, and software, among other things. That shapes the way I talk about AI: I do not just look at the technology from the outside, but care about how it actually connects to a task, a process, or a system. Because there is a big difference between an impressive demo and a solution that still makes sense after three months. You can read more about me.

Format, length, and location

A talk on AI is typically 45–90 minutes, depending on the event and the desired depth: a concentrated inspiration talk of 45 minutes, a classic talk with room for perspective, cases, and demonstrations at 60 minutes, or a more in-depth talk of 75–90 minutes with more time for questions and dialogue. If you have a different frame, we will work it out. I am based on Funen and am happy to give talks in Odense and the rest of Funen — but of course also travel to events elsewhere in Denmark. What matters is not the postcode, but finding the right format for the participants.

What does a talk on AI cost?

The price depends on the format, length, location, and degree of tailoring. Send me a few details about the event — date, location, approximate number of participants, and a bit about the purpose — and I can give you a specific price.