3#18 - Erlend Aune - Bridging the Gap: Data Science Education and Industry Collaboration (Nor)

"Det var jo veldig urealistisk å tenke kanskje at en haug med folk som har matematisk eller Computer Science bakgrunn, skal komme inn og skjønne forretningen. / It was very unrealistic to think that maybe a bunch of people with a mathematical or computer science background would come in and understand the business."

Join us on Metadama as we welcome Erlend Aune, an accomplished data science expert with a rich background in both academia and industry.  Through real-world examples from the Norwegian industry, we illustrate how successful research collaborations and technology transfers can stimulate innovation and create value. Despite the promising advances, we also candidly address the cultural and operational challenges businesses encounter when integrating AI research into their workflows.

What practical steps can bridge the gap between theoretical education and real-world application? Our conversation further explores the intersection of business development and the practical application of machine learning and data science. We emphasize the need for environments that foster hands-on experience for students, such as hackathons and industry-linked thesis projects. Additionally, we discuss the importance of tailored training development within organizations, focusing on understanding trainee characteristics to achieve meaningful training outcomes. Tune in to gain valuable insights and actionable advice on nurturing the next generation of data scientists and enhancing organizational capabilities.

Here are my key takeaways:

Data Science and Business Development

• Data science needs a strong connection to business development 
• You need to embed Data Science in a cross-functional environment
• Business acumen needs to be ingrained in the work with data
• Data Science needs to start from a Business side - ensure that you work on the problems that generate value for your organization.
• Data Science works with probability, not certainty - this notion is not yet understood by everyone in business.
• Data organizations are often build on an engineering mindset, that can be contradictive to an exploratory mindset.
• Even when designing Data Warehouse, you need to understand the business impact, have a business development mindset.

Norway & AI

• Norway has a great AI and ML research community.
• The public discourse on AI portraits a quite narrow view, that doesn’t reflect the broad application and research done in the field.

Research & Business

• Responsible AI is not a one-size fits all. Different organizations have different needs, for either certainty, security, reliability of outcome, etc. So a rAI approach needs ton be tailored to the business need.
• Startups and companies that have products related to the AI research environment, have the advantage that products are improved in tact with research development.
• In addition to in-house R&D, organizations can collaborate directly with research environments at universities.
• You cannot do R&D just as a pocket of excellence, if you want to operationalize results in your organization.
• We need to shorten the distance between R&D and operations.

For the Data Science Student

• If you apply knowledge on different challenges, you will get an intuition on how to solve a broad variety of challenges.
• When selecting a task within an organization as a Master thesis, make sure the task is delimited.
• Traits to succeed as a student working in industry:
  • Interest in your discipline
  • Interest in the organization and its sector
  • Problemsolving
  • Creativity

Transcript

Industry-Academia Collaboration in Data Management

Speaker 1 0:00

This is Metadama, a holistic podcast about data management in Norway. Hi and welcome. My name is Winfried and thank you for joining me on a new episode of the Dama Norges podcast, where our vision is to give data management in Norway a boost, to show the competence and at least the level of knowledge we have within Fragfeld, and that is why I invite with me Nordic experts in both data management and information management to a talk. Welcome to Mestadama. Today we will talk about a very exciting and important topic, namely how we can connect academia and industry and the business world better together, and for this topic I have with me Erlend Haune. Welcome, erlend, thank you.

Speaker 2 0:54

It's nice to be here.

Speaker 1 0:55

It's fantastic to have you with us and it's very good to have an episode around such a topic also in Norwegian, because it meets a lot of the education around data science in Norway et slikt tema også på norsk fordi det treffer veldig mye av den utdanningen rundt datavitenskap I Norge.

Speaker 1 1:07

Men det treffer også den norske industrien Og Erlend er jeg ville kallt han en slags ekspert på tema fordi han har både jobbet I ulike selskaper men samtidig har hatt en stilling som første Amanuensis eller Associate Professor og Data Science VBI.

Speaker 1 1:25

And in our data world we have a very high need for a good education, a good foundation, before we get out into the business world. And what has been very much criticized, especially on the data management side, is that we are still very heavy on a classic database focus education, an education as a database administrator, more than data management. At the same time we see a very exponential development in the various subjects where both data management is a basis for, but also what we see that data science, with the whole development we have seen around machine learning and AI in recent years, has taken a very big part of the importance of data science. With that as a basis, erlend and I have sat down and thought this is what we need to talk about, because this is a topic that is important for several. So before we jump into the topic, erlend, please introduce yourself.

Speaker 2 2:28

Yes, so I'm Erlend Aune. I have a long experience in data science in the Norwegian environment. I worked as chief data scientist and data scientist in a consulting company. I worked as head of analysis and insight department at Amedia, as director for data science in a startup company called Dexabel, and a couple of other companies In addition. I have these academic positions first of closely within the boundaries between what I call research and the use of research in these companies. I think it's an incredibly exciting topic how you can connect it better, how you can create systems where you can improve the use of research that is done at universities and universities, and how the?

Speaker 2 3:26

company can also see the value in doing research. That's an incredibly exciting topic for me. So a little about what I do otherwise, apart from being an extreme data nerd. So in the interest of leisure, I like very good wine and food. I like to try new places to eat and collect wine, and I also play frisbee golf. So that's my main hobby, and I also watch movies. I think that's really fun, but I also have a very strong interest in data and I'm kind of the type who reads scientific articles at night if I don't have anything else to do. It's a hobby too. Is there anything new here that I can use my own research on, or ideas that can be used for research development? That's what I'm doing.

Speaker 1 4:21

If you think back a bit, when did the interest in data start to rise? When did you start to get that interest?

Speaker 2 4:28

I think that the interest in data came back when I studied mathematics at NTNU. I was very interested in sound compression, music compression algorithms for music, for example, the MP3s, which most people probably knew and how those techniques worked. So then I started reading about how it worked and looking at what was the best way to do it and so on, and after I finished my studies I took a master's degree in pure mathematics. That's where I started with that. But when I finished I started taking a doctorate in statistics, and statistics is a data subject.

Speaker 2 5:11

I was doing it in high school and now it's like I don't think those data were the most interesting data at that time. Now I think they are much more interesting. But, for example, things like wine and beer recommendations was something I found very exciting at the time. So I talked to those who had big wine databases and beer databases at the time and asked if I could get their data sets. And then I started, for example, to download data from the wine monopoly and made some apps that kept me updated on new wines and discounts that came in. So that was maybe where I got the biggest interest and then it just went well after that, after I started working and stuff.

Speaker 1 5:56

That was really exciting and I really liked that Halloween data your introduction. I think that's super exciting and one of the most famous learning models around taxonomy is around how to categorize wine, for example. So it's very well known to me from the data management side. When we talk about data as a science, and especially about data as a science in Norway, what do you think is the status?

Speaker 2 6:24

Yes, I think that's an interesting question. When I was done with my doctorate and that was in 2012, what do you think is? And one of the types of thematics that I saw coming in was that there were several companies that were involved in the data science department and thought that they were going to make magic with data, that they were going to make something like automatic sales value, and that was very unrealistic, and I think that a meeting with people and the automatic value of the transaction that was very unrealistic. Imagine that a hub with people who have a mathematical or computer science background would come in and understand the transaction and just use the data that existed and make magic out of it. I think that was completely unrealistic at that time. I was fortunately not one of those who got to experience that, but it was something that happened in the past. But it has matured quite a lot after that time.

Speaker 2 7:36

I think that today it's better to have business development and have it cross-examined, that you don't set a high standard with just mathematical background, for example, to think that they are going to make magic out of the data. There has been a pretty big development, I would say, at the same time as there are companies in Norway today that perhaps have the unrealistic expectation that if you create a data science department I Norge I dag, som kanskje har den urealistiske forventningen at, ja, hvis du lager en data science-avdeling, så vil du automatisk få beretningsverdi. Jeg tror jo på det her, koblingen mellom data science og business development, som vi sikkert kommer litt tilbake igjen I løpet av denne samtalen. Hvordan kan man få retningsforståelsen helt ned til der where you look at the data and then develop new values? Today I think we are starting to become quite modern in data science, despite some examples where it doesn't work so well.

Speaker 1 8:36

It's super exciting and I've seen different approaches to data science. There are two things that we still see as problematic, or at least a challenge in several companies. One is the expectation of certainty that the results you get from the data science unit is something you can bet 100%. It's a graded answer where there's perhaps a possibility of something. It's not as clear and it doesn't come as clear when we talk about data science in a setting. The one and the other thing I still know a lot about is the mindset. I think that often in companies that have built up a data team or a data delivery in a company have started with data engineering and gone a litt det løpet. Så man har et veldig ingenieur mindset som kanskje ikke alltid passer like godt overens med et veldig utforskende mindset som man har I data science.

Speaker 2 9:38

Det er et veldig godt poeng, synes jeg, Fordi hvis man begynner med engineering, så har man kanskje ikke den sterke koblingen til hvordan det selskapet.

Speaker 2 9:44

If you start with the engineering side, you may not have the strong connection to how the company will create value, but if you start with the business development side, you have a way to start with how you can use data to get value, Understand what the mechanisms are, to take it further, and that's something we've talked a lot about and I've talked a lot about with companies and up through the world. It's that okay if you start from the science part, there are always some interesting problems from a scientific perspective or from a data perspective, but it's not like okay, which of those ideas? How do those ideas connect to value development? That's probably what you need and that's where I think the link with business development is very important, and that's where I think the link with business development is very how do these ideas connect to value development? That's probably what you need and that's where I think the link with business development is very important that you see which ideas we should prioritize in order to get the value out of what is possible with the data.

Speaker 1 10:40

Exactly, and that's exactly what's so exciting, because it needs a connection to the business development side, as you mentioned. Find the good use cases in the company. This is what we want to invest in, this is what we can invest in, and it's not an easy task. It actually needs several different. Now we're talking about mindset, but it's exactly that it needs several different mindsets, several different backgrounds to find these good solutions.

Speaker 2 11:06

I agree that you might need a cross-disciplinary collaboration between different types of personalities. Sometimes you might have flux right and it happens that several of the companies that have done very well with data science they have the data science that has both the business development mindset and at the same time the data science type of mindset, and then it's easy because then it almost happens automatically. But if you, for example, have someone who has a data science perspective but perhaps not the business development mindset and you don't have a collaborative or cross-disciplinary environment, then I think it will be difficult to succeed unless you have a very specific use case that you can iterate on and know that if you do a little better here, then you get value for the company.

Speaker 1 11:56

I will take a little detour towards my field Data Management and Data Governance. In the theory, we see Data Scientist as a consumer of the data and data governance. In theory, data scientists are consumers of data. In reality, you are more about creating good data quality and the data you want to use. Our president, peter Eichen, president of Dama International, has talked a lot about that. We have to do something with education in data management and universities. We still see a very classic database focus in education. There is still a core role in the program and it needs to develop. Do you see from your perspective that something is happening? How do you think it looks in Norway?

Speaker 2 12:43

So now I'm moving us into an area that I can't talk much about, but it's still very connected and related to what I talked about earlier, and that's that if you're going to make, if we think about data management and making a data warehouse and such, then it's like, okay, if you're going to get the design of the data warehouse, then the design is how you get insight is very connected to the design of the data warehouse. I think you also need this type of business development type of thinking. What do we design these things for? What kind of insights and how should we use the data to, for example, do machine learning?

Speaker 2 13:23

It is not necessarily a normalized type of database. It's the best if you're going to do and coordinate use cases where you're going to make good machine learning models, for example, and at the same time, I also think that you need this type of business development mindset and say, okay, what is the most important insight? We need to drive it forward, then you need more than a theory around traditional database thinking. Then you need the connection to business development and how to use data to create it in the company. I'm a little unsure if it needs to be one person who knows all these things, or if it needs to be a double-faceted team, because in the company you can design a team where you're sure you have different skills. It can be good that someone is really really good at this database thing, because you need very high performance, for example, to do what you want to do, and then maybe another person takes care of this, which is about design and what are the use cases and value creation. So that's a little from my perspective, taking it out of there.

Speaker 1 14:40

Now I think that's very good, and then you get, on the one hand, the focus on value creation. On the other hand, you get a little bit like you can AI engineers in Norway and the whole world. Ai has really taken a step forward on the scene and is something that is being discussed by everyone, but has AI machine learning gained a good fo food festival in Norwegian research?

Speaker 2 15:25

I would say that we have a very good environment in research within AI and machine learning in Norway. We have good environments all over the country, from the University of Tromsø in the north to the University of Agder in the far south. There are many good groups that work with different approaches to this research. One thing I think is perhaps a bit interesting in this question you're raising is that what people think of as AI is much more than what you read about. For example, when you read about a gen-AI type of statement, then it's often large language models you're talking about, for example. But generative AI can be much more than that, and I think it's a good thing that AI is much bigger than what you see in the lenses you get through the media today. There's a lot of interesting research going on here, both basic and used research in these environments, and I hope it will be even greater in the future.

Challenges in Implementing AI Research

Speaker 1 16:26

That we may almost expect, and it is also a little reassuring to know that Norway has such a good environment in academia that there are opportunities to have our own education within these topics and to get students out into the working life afterwards with that knowledge og få studenter ut I arbeidslivet etterpå med den kunnskapen. Jeg har hatt en veldig god berat for en stund tilbake rundt Explainable AI, og det er jo også et kjempespennende fagset ikke bare Explainable AI, men også hele den der Irresponsible AI, ai Governance beaten. Jeg har en sår ventning nærmest at det kommer til å bli mye større og mye viktigere tema nå fremover. I have a feeling that it will become a much bigger and more important topic in the future.

Speaker 2 17:09

How do you see this?

Speaker 2 17:10

I agree with that.

Speaker 2 17:12

I think of Explicable AI as one part of the environment within the field of Responsible AI and Trustworthy AI, and I think one thing I think of there is that if you take a use case in a company, they will probably have different dimensions of what they need to say that a model is trustworthy or that you trust the uncertainty.

Speaker 2 17:35

For some types of use cases, it's important to understand why you get a certain type of prediction, and for other types of use cases, it's to have similar models that are robust for, let's say, complex data, for example, that you can get reasonable outputs even if you have outliers, for example. And I think that's going to be really important and I hope that, for example, and I think it will be very important, and I hope that, for example, the Research Council has put full focus on this in its new research projects within AI, so that we can do research in that direction, where people think about these things. So I'm completely happy that it will be. I hope that it will be even more important, and that's also because it's also connected to the subject of statistics. Understanding models is very important. We have a theoretical basis for the things we do. I think it is very important.

Speaker 1 18:38

I hope it is not only important at an academic level, but also to be able to get the transition into the industry and the business world, and that the models that are actually used in the Norwegian industry can be traced or we have a control over our input. And that's actually what we're working on in our main theme, which is the collaboration with industry, and the big question is does Norwegian industry have any value in the research we do in Norway?

Speaker 2 19:08

I would say that we definitely do.

Speaker 2 19:11

There are many examples of companies that have come from research environments. For example, it's one thing that you have a research environment that does some research and then you find out that you can use your research to create a company there. You have, for example, technology Transfer at NTNU forskning og så finner man ut at man kan bruke den forskningen til å lage en bedrift. Der har du, for eksempel, technology Transfer på, for eksempel, ntnu, som har masse spin-offs som har laget verdi. Der man sier at hvis du lager en bedrift så har du en use case som potensielt kan ha verdi. Men det er også sånn at de fleste selskaper som har som også får verdi. There are several knowledge sets that also have a value for research within AI in Norway, for example, startups. Within startups it is perhaps easy if you see that what we do in the startup, if they do something that is very connected to AI or machine learning, then it is potentially easy to get that value in because the progress in research there will make the product better. There are several start-ups here in Kærslo that are very machine learning driven, for example sound and sensor technology and industrial AI, cognite and Intelessi and Sound Sensing, for example, companies that work a lot with machine learning, time and sound, which get the value of the research by themselves, but also because they run their own research. For other companies that come from another domain, it may require other types of mechanisms to get that value. If it's a large company that comes from more traditional domains, then you may have another mindset and a certain type of culture where you decide that, okay, we see that this research can be important for our future value creation. And then the question arises how do you do that? Do you create your own research department that deals with research or do you encourage collaboration with academia? Personally, I have had a lot of collaboration with, for example, telenor, which has its own research department called Telenor Research, and they get value out of basic research where we do, for example, anomaly detection on time zones and will use those models towards monitoring of their networks, for example. But they have their own research department which has the purpose of researching to drive the business forward.

Speaker 2 21:45

I think, even though there are many companies where you have the value of research, either by having research that automatically leads to value because it affects the direct business area, or that it is established as a form of mechanism to use the research. But you also have many companies where you take advantage of the research, but you also have a lot of companies where you don't get it right, that there is no such thing as the culture or mechanism where you might meet with skepticism. If you go and talk to, if a person goes and talks to a research environment, then the company may happen both more as a service and in collaboration with a research environment. An example is that, okay, there are many explanations at the research board that require that you have, let's say, 10% of the budget that should come from industrial partners. There are several companies that can find out that they are just doing it as a service and don't get anything back, and that's a challenge, right.

Speaker 2 22:39

And then we come back to this with business development, that you might not, that you might not see the connection, that, okay, if they, if it's a person who doesn't see the connection to the research that needs to be done to their own business, then it's difficult to defend them and, on the other hand, it can and on the other hand, those from the research environment might say, okay, this use case, I don't understand how I'm going to fit in. You have a dynamic there that can be quite challenging if you haven't established an environment or a culture for this, and there is the double-facetedness, where you have people who understand both AI or machine learning, and business development. I think that is a very important component here, and we have perhaps too few of the key leaders in Norway. It's a challenge that I've thought a lot about.

Speaker 1 23:34

This is super exciting. I don't know exactly where I'm going to start. I had Jeba Martin-Kneite on the podcast in an earlier episode where we talked about Responsible AI. She is the director of Research Innovation for a whole year, as you mentioned, as a research department.

Speaker 1 23:54

I think there are several companies that have made R&D units in Norway, so there has been a lot of research and what you're saying now is a very challenging problem that you get research in companies like Pockets of Excellence and you don't reach the entire company with it, and there are some side effects, some disconnections and use cases that may come from operative environments, but the results of the research are not as usable and we have seen this in several places. You say the connection is lacking a bit, but I feel that it is also. How do you call it operationalizing the results of the research? That's where the problem lies. You can find good use cases. You can connect the use cases to a research work. You manage to research on them, but as soon as you take it into operations again there will be a challenge.

Speaker 2 24:46

There will be a way, a little too far away from the research to implement it in the product. So it's probably much easier to get it there, while for large organizations with a lot of complex what should you call it control systems, complex matrices and all that, it's difficult to disseminate the research through the whole company. I'm coming back to the point that you need people who understand both domains, that you understand the business development case and the AI part, the OML bit, so that you can connect it and see that, yes, here we have a, here we have a research, here we have a project that we can run to get it out to the business. Yes, here we have a new project that is based on the research that we can use elsewhere in the organization and so on. I think it's very and maybe even no one who has something like that up, but I think it's a very interesting thing how you can have both research which you know is a good Jewish, as you mentioned, you know and this structure in the company where you can push it out and use it and get scale advantages from it.

Speaker 2 26:21

And then we're perhaps a little back to what we talked about earlier where, okay, is this something you learn in education. I don't think you learn this in education. To be completely honest, I don't think there's anyone who works so closely between the business development case and understanding of the possibilities in research and data slash, machine learning, slash data science today. And that's a question is there something you should learn in education or is there something you should learn when you start working? And the other question is do you have good environments, do you have good mechanisms in your companies to teach them to the conditions? I'm a little unsure. I think there's little of it.

Speaker 1 27:04

I had to think back to what you said, hedvigstad, in 2012, when you started, and that was a time when the Lean Startup Guide, erik Reis, had just come out. I think everyone has said about him at that time and he has a letter in the book for innovative environments or maybe a startup in large companies. How can we get that startup culture into a large company? I think there are many, also Norwegian companies who have tried this to set up innovative units in companies and again, there is the way between the innovative startup or R&D department in the company to operations has become quite long, and I think it's a bit similar that you have some mistakes and expectations to what you can get in a company at a certain size.

Speaker 1 27:57

There was also another thing I thought about and that is perhaps a problem, especially in data governance, but also enterprise architecture, which we talked about in the last episode. It is some of the areas that think very structurally, that think very theoretically around data. They also get that disconnection from what actually happens and then the company actually starts to see what we are doing around data. They get that disconnection from what actually happens and then the company actually starts to see what it is about. Why should I have an enterprise architecture department that just has high-level thoughts about how things should be, instead of adapting to the reality of how things actually are?

Speaker 2 28:34

Yes, that's good points there as well. I have to admit that I don't know much about business architecture, but it's important to ask these questions and say, if we're going to do innovation within, for example, machine learning, how are we going to do it? How are we going to define structures that make it able to get it out? It's a lot of personal dynamics out here as well, that if you manage to get the right people with your ideas, then it's often quite good to get a breakthrough, to do this research and create value with the research that they do.

Speaker 1 29:21

To take it further, and that was one of the things that we talked about in preparation, which I think was very exciting and very valuable from the conversation we had. And it's up to the student out there who studies in data science what do I have to remember when I want to succeed in the job market? What are the principles and what are the important things for a student in machine learning and data science to succeed?

Speaker 2 29:56

I think that's a very good, very theoretically heavy and dive into it and either go the research way or take very technical roles in the company.

Speaker 2 30:10

I think we will always need that.

Speaker 2 30:13

But if I think a little more broadly that we might need more of and that we might not have so much of today is the concept of trying to solve many different tasks, such as machine learning, data science and statistical methods and so on.

Speaker 2 30:31

There are many different types of problems you can try to solve and I think that if you try to solve many different types of problems and challenges with these methods, you also get an understanding of what they can do, a deeper understanding of how they can be used.

Speaker 2 30:47

And that doesn't mean that you get a very necessarily much stronger theoretical understanding of the method, because it can be your own word right that you get this theoretical, mathematical understanding of a method. But the other type of understanding, if you use the method on many different problems, then you get an intuition for a problem solution that you can use in your work life. And if you do that, if you use these methods that you learn in courses and studies on many different programs, then you are very well equipped, I think to solve many different tasks. I also think that maybe to a certain extent it is linked back to if you understand that this method or this type of thought process works on several things, then you can also see more links to the business you are starting to work in, the. I think it's a big trust. Just go out and solve a high level task, take part in hackathons, go on kagging and try the things you learn in the course on something completely different than what you thesis in the industry or to connect a company.

Speaker 1 32:23

You get the practical part. You can solve a problem for a company while you write your thesis. But what is it you need to aim for when you want to write your thesis in an industrial company?

Speaker 2 32:37

Yes, this has been held as a proposal at a time, but it was held for those who were supposed to give the tasks. Originally, I had made the proposal for the students, but then I discovered that there were no students in the room, so then I held it for those industries or people who were supposed to give tasks to the students instead. But one thing I think is super important here is that there is a big mix som skulle gi oppgaver til studentene I stedet for Men. En ting som jeg synes er superviktig her det er at det finnes en stor miks når det kommer industrioppgaver in AI-maskinlæring. Det er noen oppgaver som er veldig åpne og det kan være veldig utfordrende for studenter. For noen studenter kan det funke kjempegodt å ha en åpen oppgave.

Speaker 2 33:19

For some students, it can work really well to have an open task where you explore a problem, try to find out methods where you can analyze data and maybe come up with something new, maybe create something that the company itself didn't think of, because they didn't have a precise task that they wanted to solve. But it's a very data science, business development type of task which is very difficult for most students to do. There are a few who can do it and when it happens, it's really fun, because then you get a lot of new insights, you learn a lot of things, you learn a lot of things that you might not have known before. But it's also like this if you're not that type of person, then you should look for a precise, defined task and if you have control over the data the data quality Because it's very often that you think, okay, we have some data that we don't have time to analyze ourselves, so we think we can try to check what we can get from a student, and then it turns out that it's high with data quality problems. So the student has to spend a lot of time on fixing the data quality to potentially do some forecasting or do some classification or do some machine learning tasks on it.

Speaker 2 34:33

The point is that it takes a lot of time to do this and then you can get quite frustrated if you don't get it, for example. So that's one of the things that I, if I talk to industry partners and stuff, then I use to be quite hard on myself to learn if they have a precise program setting and if you have control over the data quality, because if not, then it can be quite challenging. So I think that, as a student, if you are not interested in the data or the company and what they do, that's maybe the first question you have to ask yourself. If the answer is yes, then you have to go in and check if they have the exact defined tasks. Maybe I'm the type of person who can go out and be creative and do a little more open task, but it's really difficult to do it right.

Speaker 1 35:20

I feel that you already have some of the characteristics that are quite central to Meo Lucas, and that is interest interest for the people you work with, but also interest for the domain you work in, and that the company is a problem solver, and, not least, creativity. Are there any more things you think are important to have with you?

Speaker 2 35:41

Yes, one of the things that might be important is that you have a compatible supervisor. So let's say that you write a task for the industry, then you should have a supervisor who is also interested in the problems that come from the industry. I think that's pretty important and you probably find that out. He's talking to the person.

Speaker 1 36:01

And the best thing is to get that combination of an academic leader on the one hand and a leader on the business side who can actually talk together.

Speaker 2 36:11

Yes, completely agree, and if it works well, then it will be really fun. Then it will be fun for both leaders, our co-leaders.

Speaker 1 36:18

We are already at the end of the conversation, which has been fantastic. Thank you so much for everything you have shared. Before we end, do you have any key takeaways or call to action?

Exploring Training Development in Organizations

Speaker 2 36:35

What we have talked about is the connection between business development, ai and machine learning. How can you ensure that you have the right level of satisfaction when you start doing projects or get projects out and scale the benefits with a research project, skalere fordelene med et forskningsprosjekt? Det er de tingene som jeg synes er et viktig tema og jeg synes det er noe vi bør gjøre mer av. Det bør kanskje være noen folk som prater om det her og prøver å finne ut litt mer om hvordan det her kan gjøres på tvers av organisasjoner, og da tenker jeg også på offentlige organisasjoner, hvordan det kan gjøres der. I also think about how public organizations can be done there, not just in the private sector. Learning development what kind of personal characteristics does a trainee have and what kind of double-faceted trainee is there to get the damage effects and create the right research projects to ensure that you get the right results? It's a form of call to action. Slash my takeaway.

Speaker 1 37:40

Fantastic. Thank you very much for a very good conversation. Thank you as well.