How AI can be implemented in management

By Björn Preuß, Kathrin Kirchner

Listen on Spotify

Watch on YouTube

Welcome to our twelfth episode of AI Watch!

This episode of AI Watch features Björn Preuß, Chief Data Scientist at 2021.AI, interviewing Kathrin Kirchner, an Associate Professor at DTU, about the evolving role of AI in the workplace. They discuss a range of topics, including:

  • AI’s potential to augment decision-making and automate tasks. Björn and Kathrin explore how AI can be used to analyze data and inform decisions, as well as automate repetitive tasks.
  • The challenges of regulating AI models. They discuss the potential for bias in AI results and the need for regulations to ensure the responsible use of AI.
  • The impact of AI on workplace efficiency and potential risks. While AI can increase productivity, Kathrin highlights the dangers of biased data and the importance of using AI responsibly.
  • Strategies for successful AI adoption. Björn and Kathrin emphasize the need for organizational culture change, collaboration between data scientists and business people, and employee involvement in decision-making.

Overall, the episode provides a thought-provoking discussion about the opportunities and challenges of AI in the workplace, emphasizing the importance of responsible implementation and collaboration to maximize its benefits.

Getting started

Björn: Welcome to AI Watch, my name is Björn. I’m the Chief Data Scientist at 2021.AI and today I have Kathrin Kirchner with me, Associate Professor from DTU.

Kathrin, welcome to the interview.

Kathrin: Thank you, I’m looking forward to telling a bit more about my research about adopting AI in companies.

Björn: That looks very interesting and I think that’s something everyone is kind of wondering about, especially after ChaptGPT has overwhelmed the workspace. How do you see AI’s role in the workplace? How does it change it?

AI as decision support and automation

Kathrin: I think AI is already now a part of everyday work, and these can have a role like augmenting the people that are supporting, for instance, in decision making. So, we have a lot of data available in the companies so managers, for instance, can rely on the data, use AI, and, for instance, make decisions in sales or marketing. We already have AI that takes an automating role, that means it automatically decides without any human involvement or even can lead people. We see this for instance in work platforms like Wolt or JustEat. The algorithm is deciding who brings which food to which customer or who gets a reward and so on.

Regulation of AI models

Björn: I think that last thing you touched upon is quite interesting, right? Because if we look at regulation of AI models and things like that, it’s often said if the AI is actually taking actions, then it’s depending on the use case obviously and is perceived as high risk or even prohibited. How do you see companies handling that in terms of when is it more of a decision support approach of an AI and when is an AI really taking over making a decision, like in the cases you mentioned?

Kathrin: I would say in many cases now we have more of a supporting role so that the people still decide whether they would like to accept the results and something like this that we call algorithmic management where also the algorithm makes a decision. It’s still a high risk and also has an influence on the well being of workers because they might find that it is unfair how they are handled or that they get a tour to the other end of Copenhagen while others always get the nearby tours and maybe can earn more money.

AI’s effect on workplace efficiency

Björn: With higher automation and increasing efficiency, how do you see the effect AI has on the workplace? Is it usually increasing productivity or efficiency or could it also have the opposite effect in some cases?

Challenges of bias in AI results

Kathrin: I mean in many cases it can of course enhance our productivity if you use tools like a ChatGPT for instance to help with emails for getting better formulations and so on. Of course AI is able to analyze a lot of data in parallel and that helps to make quicker decisions. So in this way it can help us. On the other hand, there are also some dangers with using this because AI can only be as good as the data that we have and we all know that we have human bias and bias is also in the data. So we might get results that are also biased and then we cannot just use them. We also have to use the AI results responsibly and we have to be aware of such dangers that we can have.

Handling challenges and mitigating risks

Björn: I mean that’s something that from our practice we are very familiar with, right? All the kinds of biases that are potentially there, that one needs to mitigate and all the regulation we see now upcoming. How do you see it in your practice? What do you observe with companies in terms of how they handle these challenges? That does not necessarily mean it needs to be regulated, right? I mean the challenges are always there. How do they overcome that or try to minimize the risks?

Intellectual property and data privacy

Kathrin: One thing is, for instance, what is a risk with everyone using ChatGPT, which is a tool that is based on the internet and we are not aware if ChatGPT invented or hallucinated it or if it is real? So we always have to be able to get new skills to be able to understand how this is produced and on which basis, or do I have intellectual property rights for instance if I would invent something based on already existing knowledge from other people, can I use it? And claim that is my knowledge now or my invention? And also there are already companies that introduced tools, can we use ChatGPT? Because when I copy my portrait report or my patent description into ChatGPT then of course ChatGPT, and the underlying database also knows. So there are companies that for instance forbid this or there are other companies who develop their own large language models based on the company data so that we have a better usage and not leak information to the public.

Björn: I agree I think that that’s definitely a suitable approach if resources and money are available, obviously.

Kathrin: Yeah, and that’s a trade-off.

Adoption of AI within the workforce

Björn: If we look a little beyond the technical requirements or regulation requirements, and say generally from using AI in a workplace, I always see that one hurdle besides the technical one is also the adoption of that within the workforce. How do you see that? Is it more something managers like to use AI and employees don’t like it, or how is that?

Kathrin: I think it depends because there are managers who are very open to this and they heard, of course, that AI is cool and it can optimize a lot of processes, so they would be very open to that. And there might also be employees that are open to that and others that might be afraid that somehow they can be replaced by AI. That’s also the discussion right now, will AI take away our jobs? Or what will change here?

Culture and process changes

Björn: And I fully agree, we also need to have an organizational adoption. We cannot just use AI somehow, or the manager can give the data scientist some data and say okay, please analyze and find something with AI. There needs to be a business value behind it, so we need to produce something that the company needs and can really use. So where they see the value of it, and of course, we also need a kind of organizational culture for adapting it, no? Then we now suddenly want to rely more on the AI or use a lot of data we base our decisions on. There are also changes in how we optimize work processes, that we have to take the people with us, the employees and also the managers of course.

Strategies for adopting AI in the workplace

Björn: Now we touched upon two major hurdles or challenges, being the culture, the how to work and processes, as well as regulation and what is actually possible and allowed to do. So what kind of strategies do you see on how to work with AI, how to adopt AI in a workplace? The best way, are there any cooking recipes? Or is it really case by case basically?

Kathrin: I think there are some things that we need to be aware of, because AI is still like a fashion, right? And a lot of companies would like to do it, but they are not yet aware of what it can do. And do I have the right data for that? When we define such a project in the company, then it’s always good to say okay, what would I like to achieve with using AI, not just I would like to use it to become more efficient, but we are completely what I really expect. And then we also need to have both data scientists and the business people on board, because it’s often the case that the managers or employees don’t fully understand what AI can do. It’s like a password, so what is behind us? On the other hand, it’s also the data scientists who don’t know the business and cannot fully understand what is the meaning of the data, so there should always be a collaboration between both sides.

We need to bring the people on board, that could be like with training, that they state now we use AI for this, this is the result, how can it be used? And, could it be possible that the AI is visible? Like for instance, we come into the company, we see some kind of monitor where we see the business results and the prediction of other results, so that we are more exposed to what AI is doing, so we are not afraid that AI will take our jobs, but we may see more of what AI can do for us, for our company.

Embracing AI opportunities

Björn: I think that’s a very good closing remark, being not afraid of using AI, but seeing the opportunities and working towards that to get the whole workforce backing the decision, not just management making it. With that being said, I’ll thank you very much, Kathrin, to be here, to have the discussion with us. Check out AI Watch on our digital platforms and see you next time.

Björn Preuß

Björn Preuß

Chief Data Scientist, 2021.AI

Björn is an Assistant Professor at CBS and the Chief Data Scientist at 2021.AI. He is the company’s industry leader in accounting and legal processes and works closely with financial clients.

Kathrin Kirchner

Kathrin Kirchner

Associate Professor, DTU

Kathrin is an Associate Professor at DTU. Her research focuses on the impact of innovative digital technologies in the workplace. She is Co-Editor in Chief of the International Journal of Workplace Health Management and a member of the Editorial Board of Electronic Markets – The International Journal on Networked Businesses.

Watch the previous episodes

AI Watch Newsletter: Episode 11
AI Watch Newsletter: Episode 10

AI Watch Video Newsletter

Get the latest know-how from those in the know. Sign up for our AI Watch Newsletter and receive the latest insights from AI experts.