Host Country Open Stage

Summary

This discussion featured three presentations showcasing innovative AI and technology applications in Norwegian public sector initiatives. Silje Sander from Lillestrøm Municipality presented their task-based labor model designed to address dual challenges of healthcare worker shortages and social exclusion. Lillestrøm, historically significant as the first internet node outside the U.S. in 1973, is leveraging its innovation heritage to tackle demographic shifts affecting Western societies. Their solution involves breaking down full-time healthcare positions into smaller, manageable tasks that can be performed by untrained community members, including young people, retirees, and volunteers. Through experiments in nursing homes and home care services, they demonstrated a 25% productivity increase while enabling social inclusion of the 18% of working-age residents currently outside the workforce.

The Norwegian police representatives, Kine Smørdal Olsen and Ragnar Thorsen, discussed their AI for interviews project focusing on interview documentation and forensic crime scene investigation. Their system uses AI transcription technology called Scriber to convert audio recordings from approximately 45,000 annual police interviews into text, significantly reducing time spent on manual transcription. Crime scene investigators are testing hands-free headsets with high-resolution cameras and voice activation capabilities, potentially saving up to 80% of paperwork time while maintaining human oversight for quality assurance.

Monica Cheng from Huawei Technologies presented their environmental protection project addressing invasive humpback salmon in Norwegian rivers. The AI-powered system uses computer vision to identify fish species within 0.01 seconds, automatically directing native salmon upstream while capturing invasive species that have grown from 6,000 in 2017 to potentially 1.5 million by 2023. These presentations collectively demonstrate how AI and technology are being strategically deployed to solve complex societal, law enforcement, and environmental challenges in Norway.

Keypoints

**Major Discussion Points:**

– **Task-based labor model for healthcare workforce challenges**: Lillestrøm Municipality presented their innovative approach to address both healthcare worker shortages and social exclusion by breaking down full-time positions into smaller, manageable tasks that can be performed by untrained community members, potentially doubling healthcare capacity without increasing employee numbers.

– **AI-powered police interview documentation and crime scene investigation**: Norwegian police demonstrated two technological solutions – AI transcription of field interviews to reduce manual documentation time, and hands-free headset technology for crime scene documentation that can save up to 80% of paperwork time while maintaining quality and legal standards.

– **AI-driven environmental protection through automated fish species management**: Huawei presented their collaboration with Norwegian communities to combat invasive humpback salmon using AI-powered automated systems that can identify and separate fish species in real-time, protecting local ecosystems while allowing native species to pass through unharmed.

– **Technology as a solution to demographic and societal challenges**: All three presentations emphasized how emerging technologies (AI, automation, digital platforms) can address complex societal problems including aging populations, workforce shortages, social exclusion, and environmental threats.

– **Human-technology collaboration and community engagement**: Each project highlighted the importance of maintaining human oversight, community involvement, and local expertise while implementing technological solutions, emphasizing that technology enhances rather than replaces human judgment and community engagement.

**Overall Purpose:**

The discussion appears to be part of a conference or symposium showcasing innovative Norwegian public sector and technology initiatives. The goal was to demonstrate how different organizations are leveraging AI and digital technologies to solve pressing societal challenges while maintaining human-centered approaches and community involvement.

**Overall Tone:**

The tone throughout the discussion was consistently optimistic, professional, and solution-oriented. All speakers presented their innovations with enthusiasm and confidence, focusing on positive outcomes and future possibilities. The tone remained collaborative and forward-thinking, with each presenter emphasizing practical applications and real-world results rather than theoretical concepts. There was no significant change in tone throughout the presentations – all maintained a professional yet passionate approach to showcasing their technological innovations for social good.

Speakers

– **Moderator**: Role – Event moderator/host (introduces speakers)

– **Silje Sande**: Title – Representative from Lillestrøm Municipality; Area of expertise – Municipal innovation, digital transformation, healthcare solutions, task-based labor models

– **Kine Smordal Olsen**: Title – Representative from the Norwegian police; Area of expertise – Police interview documentation, AI transcription technology for law enforcement

– **Ragnar Thorsen**: Title – Crime scene investigator from the Norwegian police; Area of expertise – Forensic crime scene documentation, AI technology in police investigations

– **Monica Cheng**: Title – Head of Public Affairs of Huawei Norway; Area of expertise – AI technology applications, environmental protection projects, invasive species management

Additional speakers:

None identified beyond the provided speakers names list.

# Summary: AI and Technology Solutions for Norwegian Public Sector Challenges

## Introduction

This session featured three presentations showcasing artificial intelligence and technology applications addressing challenges within the Norwegian public sector. Representatives from municipal government, law enforcement, and technology sectors demonstrated how emerging technologies are being deployed to solve operational and societal challenges across Norway.

## Speaker Presentations

### Municipal Innovation: Task-Based Labour Model for Healthcare

**Silje Sande, Lillestrøm Municipality**

Silje Sande presented Lillestrøm Municipality’s task-based labour model designed to address healthcare worker shortages and social exclusion. Lillestrøm, notable as the location of the first internet node outside the United States in 1973, is developing an innovation district called Challenge Innovation Hub, comparing their initiative to Oslo Science City, 22 at Barcelona, and White City in London.

The municipality’s approach involves restructuring healthcare work by breaking down traditional positions into smaller tasks that can be performed by community members including young people, retirees, and volunteers. Through experiments in nursing homes and home care services, they achieved a 25% productivity increase while enabling social inclusion of the 18% of working-age residents currently outside the workforce.

Sande demonstrated their micro-learning approach with a detailed training module example featuring interaction with Anna, a 72-year-old client with mild dementia. The module shows how untrained individuals can be educated for specific healthcare tasks without requiring formal education. Initial experiments confirmed that over 50% of healthcare tasks can be performed by untrained individuals with appropriate guidance.

The project utilizes technology partners NIBI (platform provider) and Copano (micro-learning platform) to deliver this innovative approach. Sande emphasized that current job market requirements create exclusionary barriers that society cannot afford given demographic pressures from aging populations.

### Law Enforcement Innovation: AI-Enhanced Police Documentation

**Kine Smørdal Olsen and Ragnar Thorsen, Norwegian Police**

Kine Smørdal Olsen detailed their AI transcription project using Scriber technology to convert audio recordings from approximately 45,000 annual police interviews into text format. The system reduces time officers spend on manual transcription while maintaining strict human oversight protocols. Officers must read through, verify, and ensure transcribed text accurately reflects interview content, retaining full responsibility for report content.

The system faces technical challenges including background noise, regional dialects, intoxication effects, and stress-related speech patterns. Despite these challenges, pilot testing continues with 20 patrol officers in Oslo and West police districts.

Ragnar Thorsen presented crime scene documentation innovations featuring hands-free headsets with IP66 certification, Wi-Fi, Bluetooth, GPS, 48 megapixel cameras, 7p resolution screens, and noise cancellation microphones. This technology enables investigators to document evidence while maintaining full use of their hands. The system includes real-time video conferencing capabilities with forensic experts in Denmark, Australia, USA, and England for immediate consultation during investigations.

Thorsen also described development of AI applications using the Capture app that transform unstructured text into formal reports, potentially saving up to 80% of time currently spent on paperwork. The technology remains in development with ongoing testing required.

Interested parties can visit booth 9, located straight behind the stage, for more information.

### Environmental Protection: AI-Driven Salmon Management

**Monica Cheng, Head of Public Affairs, Huawei Norway**

Monica Cheng presented their environmental protection project addressing invasive humpback salmon in Norwegian rivers, supported by the Norwegian Government. The AI-powered system uses computer vision algorithms to identify fish species within 0.01 seconds, automatically directing native salmon upstream while capturing invasive species.

The system addresses the exponential growth of humpback salmon populations: 6,000 individuals in 2017, 20,000 in 2019, 200,000 in 2021, and almost 600,000 in 2023. Their system has captured 60,000 humpback salmon to date. The technology operates continuously using solar panels for electricity supply and provides detailed fishing statistics while keeping rivers accessible for community engagement.

Captured invasive fish are utilized economically through local markets and dog food production. The system enables 24/7 river monitoring with minimal wildlife disturbance. Implementation focuses on Berlevåg, Finnmark, and Badevog locations, with plans to install systems in northern Norway rivers during the current salmon season.

Cheng noted that humpback salmon follow a two-year lifecycle, only coming to Norway in odd-numbered years. She emphasized that successful implementation requires collaboration between technology experts and local communities, acknowledging that AI cannot replace local knowledge and community engagement.

## Key Technical Challenges

Several technical challenges emerged across presentations:

– **Police transcription**: Background noise, regional dialects, intoxication effects, and stress-related speech patterns impact AI transcription quality

– **System integration**: Incorporating AI tools into existing criminal case management systems requires careful planning

– **Environmental scaling**: Addressing multiple invasive species beyond humpback salmon across Norwegian rivers

– **Municipal scaling**: Expanding task-based labour models beyond pilot programs to address broader social exclusion

## Implementation Status and Next Steps

**Municipal Healthcare**: Continuing to expand task-based labour model testing beyond initial nursing home and home service experiments, focusing on developing effective micro-learning modules.

**Police Documentation**: Ongoing pilot testing of AI transcription tools with 20 patrol officers while developing proof-of-concept systems for AI-based report generation from unstructured text.

**Environmental Protection**: Installing fish sorting systems in northern Norway rivers during the current salmon season, with plans to scale automated systems for comprehensive river management.

## Conclusion

These presentations demonstrate Norwegian public sector organizations’ strategic use of AI and digital technologies to address complex challenges while maintaining human oversight and community involvement. Each solution focuses on augmenting human capabilities rather than replacing workers, whether in healthcare task distribution, police documentation efficiency, or environmental protection automation.

The initiatives show how relatively straightforward AI applications can deliver significant practical benefits when applied thoughtfully to specific problems. From municipal healthcare innovation to police efficiency improvements and environmental protection systems, these projects illustrate practical approaches to technology implementation that balance efficiency gains with human values and community needs.

Moderator: Please welcome to the stage from Lillestrøm Municipality, Silje Sander.

Silje Sande: Good morning. First of all, I want to welcome you to Lillestrøm Municipalities, one of Norway’s most innovative municipalities, as demonstrated both by our history and current innovation areas. We are honored to be here today to represent the spirit and the capacity of the innovation that exists in Norway’s municipalities. As most of you probably know by now, Lillestrøm was in fact the first internet node outside of the U.S. back in 1973, and this was not by accident, as there is a long history of digital transformation and governance in precisely this area, as you can see from all these historic moments. Today we are leveraging this history as we are scaling up our effort to meet new challenges in new ways. We are currently building an innovation district here in Lillestrøm, centered around the Challenge Innovation Hub just down the road. Inspired by the Oslo Science City and aligned with global districts like 22 at Barcelona and White City in London, it brings together academia, business, and public institutions to accelerate science-based innovation. The goal is to create a new gravity center for future industries in our region. However, the topic of today is not to show you this, but to show how we use our innovative muscles in new ways. The western world is currently facing a demographic shift characterized by an aging population and declining birth rates. This leads to a shrinking working age population and an increasing number of older individuals, creating especially serious challenges for healthcare systems and the overall economy. Due to this, there are concerns about the sustainability of western economics, and in the face of these challenges, some even claim we head for bankruptcy. The current healthcare models are indeed not an alternative for the future, and unless we take radical actions, it can destroy our welfare state as we know it, and this calls for action. Five years ago, we started working on a solution to this and to the lack of healthcare personnel. Our goal is to double the capacity without increasing the number of employees. Our hypothesis was that aggressive use of technology and radically changing the way we use our resources will solve at least part of the problem, but as we delved into this topic, it didn’t take long before we realized that our society is facing yet another massive and somewhat opposite problem. Social exclusion with an increasing number of people working of working age outside work and education system. In Lilleström, this number is 18 percent and it is rising every year. This is not a problem unique to Lilleström or even Norway. People involuntarily outside the labor force is a problem everywhere, and the irony is not lost on us that we lack people in healthcare at the same time as the exclusion rates grows. So we need to address both problems at the same time, the theory being that they can partly cancel each other out, but how do we mobilize people to enter the workforce? When we started digging into this social exclusion problem, we became aware that the current job market has very strict framework. It is mostly suited for people working in full-time positions with a high degree and a predictable life situation. As the requirements for higher education rises, the threshold for entering the job market also becomes increasingly high. Our job market is not suited for those who fail to deliver on these elements, and it is becoming a very excluding factor that we cannot afford, neither for human nor economic reasons. We had to extend our mission and add social inclusion to our solution, and we therefore created the task-based labor model. By dividing full-time positions into small limited tasks, we can make them available to community resources. With a high volume, we can standardize many of these tasks and present them in a platform delivered by our technology partner, NIBI. Community resources can pick and execute on tasks how and when their situation allows it. Creating such a marketplace for tasks allows us to free our mind to think of other solutions on education and skills development as well. We are currently exploring how we can use micro-learning to educate and qualify resources on task level, removing the need for more complex education and training. Using tasks as the entry for education reduces the competence issue and makes it easier for resources to enter the workforce. We can teach them enough for that specific task and continue to offer micro-learning modules for new tasks, giving them the chance to grow and to develop. We believe we can solve both the labor shortage, reduce social exclusion, and create a work life with more flexibility and equality. The endless possibilities of types of tasks and micro-learning modules makes a way for new and easier career entries and career paths, separating us from gig economy and reducing the risk for an underclass. Our task-based labor model has been a somewhat theoretic exercise and for that reason we created the future experiments where we simulated 10 years into the future with an up and running task-based labor model. The goal for these exercises was to test whether such a market would give real capacity increase in the health service as well as document positive effects and identify inhibitors. In the experiments we mapped all the tasks currently performed by health care personnel and in both cases we found that more than 50 percent could be performed by untrained people. In the first experiment, lasting one week in a small nursing home in Lilleström, we used 19 untrained people to do work related to kitchen and food, room service, stock room, activities, and simple health tasks. They represented work inclusion companies, community service after convictions, both very young and retired people, volunteers, and more. The experiment showed a 25 percent increase in productivity throughout the week, which is huge. The experiment taught us a lot, for instance the need to automate processes, but also that we need a digital system to facilitate micro-learning. The next experiment was performed in our LAI municipality, Norderfollo, this time in the home service. They had 23 community members joining the service covering a hundred users. The training was done digital through a micro-learning platform with interactive training modules created by Copano on we are learning technology. It was a huge success and the experiment also confirmed the results from the first experiment. Here is a short clip from one of our training modules. You are on your first shift in home health care and have been assigned a new client on your rote, Anna, 72 years old. The report states that she has mild dementia and lives alone. Today you will go by to help her with personal care and check how she is is doing. You know that people with dementia may react differently than others and that the first meeting is important for creating a sense of security. Hi Anna. Who, who are you? Do we have an agreement? Relax, I’m just going in to do my job. I don’t know you. Go away. I’m locking the door. Anna is clearly nervous and anxious when she opens the door. Even if you know Anna, she might not recognize you. Try a different option. Hi Anna. My name is Kayhan and I come from home care in Nordrefalla. I’m here to assist you with your care today. Yes. Is it you who is coming today? I can’t remember all of you, but you can come in then. Great. A clear and friendly introduction helps Anna understand who you are. It provides comfort and respect for her situation. This was a short clip. The labor-based market model has given us a lot of attention in Norway. However, we believe this is relevant for other countries as well. So if you want to learn more or you want to test our model, you can see us down at the stand, right down there. Thank you so much.

Moderator: Please welcome to the stage from the Norwegian police, Kine Smørdal Olsen and Ragnar Thorsen.

Ragnar Thorsen: Good day. Good morning everyone. I’m Ragnar. I got with me Kine and we are from the Norwegian police. We are going to talk a little about a project we are working in. The project AI for interviews is using innovation and technology to make solutions for the Norwegian police force. Today we will focus on two parts of the project. Interview documentation in the field and forensic crime scene documentation. I will start off by showing a short clip. of an interview with Kine Smørdal Olsen and Ragnar Thorsen, and then I will show you a short clip of the project AI for interviews with Kine Smørdal Olsen and Ragnar Thorsen.

Kine Smordal Olsen: As said in the video, in Norway about 45,000 police interviews are done by patrol officers on the scene each year. And since 2016, all police interviews must be recorded, including those done conducted on the scene. And by recording all interviews, the material can later be used in court, often eliminating the need for a second interview at the police station. When conducting an interview, you have to document it in a report. This implies returning to the police station, listen to the audio, and manually transcribing the interview. This is a time-consuming process, and this is a process we hope to make more efficient. During a period of one year, the pilot will test the use of AI transcription from audio recording to text. Let me walk you through the workflow. First, the officer records the audio of the interview using an app called Capture. This app is available for all police officers in Norway, and it’s used to record audio and capturing photos. However, only the dedicated pilot users have access to the transcription ability and permission to use it. The audio file is sent to our AI service, Scriber, and Scriber transcribes the audio record and produces a text file. This is automatically made available to the officer in a personal folder. Using AI, we must always have a human in the loop. The officer must read through, verify, and ensure that the text reflects the content of the interview. The police officer is always responsible for the content of the report. And finally, the interview report is created and submitted into the criminal case system. Today, the pilot currently involves 20 patrol officers. They’ve all received appropriate training and received support from pilot support team whenever needed. What we learned so far is that the transcription tool is simple to use. It is not obligated to use the transcription text and even informal conversations, like in public disturbance cases or verbal insults, can be recorded and transcribed. That said, we also face some challenges, like background noises, dialects, intoxication, and stress can affect the quality of the transcription. But we’re learning and the pilot is helping us identify what works and what needs improvement. Looking ahead, we hope to integrate AI-generated summaries into the process that could streamline our work even more. The overall goal is not to replace the officer, but to support the work we do. Faster and better tools means that reports can be written even closer in time to the event, which is a win for investigations, victims, and the public.

Ragnar Thorsen: A selection of crimes in investigators in Oslo and West police district including myself are these days using a headset like the one I’m wearing now as a part of the project. The headset is IP66 certified, has Wi-Fi, Bluetooth, and GPS, a 48 megapixel camera, and a screen with a 7p resolution, and a microphone with noise cancellation. The headset gives us the ability to record videos at crime scenes, and hopefully soon we’ll have a video conferencing with forensic experts. Best of all, it’s hands-free, it’s voice activated, and this is essential for us crime scene investigators as we need our hands clean and free. And as I mentioned, the headset has the possibility for video conferencing. We have had several international exercises calling for instance Denmark, Australia, USA, and England. In addition, we can see the possibilities of a conference with the investigative command center and the operations center. The high-resolution recordings of the crime scene, in addition to our forensic work on-site, is a benefit to the rule of law. In addition to the crime scene itself, we can record or examination of the evidence at laboratory laboratories. And during our work at the crime scene, we verbally describe what we’re doing and what we’re seeing. This gets recorded on the video. We then use the same speech to text engine that Kina mentioned, Scriber, to produce texts out of the recordings. We are also testing out an AI-based app that transforms unstructured texts into different reports based on a template. This shows potential for saving up to 80% of the time spent on paperwork. And finally, in the process, quality assurance must be done by humans before submitting the final reports to the criminal case system. And the speech to text and recordings of crime scenes are used these days in a pilot on real criminal cases. The AI app is a prototype that’s being developed. We are currently developing a proof of concept for it. For this, we are only using test data and not using real criminal cases. If anybody would like to try the headset and get some more information about it or what Kina talked about, feel free to visit us on our booth straight behind the stage here, booth 9. And thank you for your attention.

Moderator: Please welcome to the stage from Huawei Technologies, Monica Cheng.

Monica Cheng: Good morning, everyone. Can you hear me? My name is Monica Cheng. I’m working as the head of public affairs of Huawei Norway. So today my speech topic is AI for good, tech for all. And I’m going to use a specific project to show you how we are using AI technology to protect the environment challenge currently going on in Norway. Because today, indeed, it’s very difficult to talk about environmental protection without talking about modern technologies. Because these technologies like AI, 5G, cloud services is indeed providing a powerful tool that we have never seen before and helping us to make the quick actions and more targeted actions. So before I go into the details, I would like to show you a small video. We have a problem. If we can’t stop it, it will take over Norway. Our island is gone. And we have been in chaos again. This is a responsibility that the authorities have. Of course, they have to do something about it. We are doing something about it here, together with Huawei. And we are happy about that. We are the ones who initiated this project. We are the ones who are in charge and have been driving this and brought in these cooperation partners, Troll Systems and Simula. And apart from that, we are the ones who contribute with the servers and applications around this with AI. The whole point of this project is to get the Bokelaxen out. And by doing that, we will have much more control over our rivers. In addition, this system will provide a very detailed information about all the fishing in this river. And that is something that these hunters and fishermen associations tell us that it will almost be just as important for them as to get the Bokelaxen out. The case itself or the principle of the case is relatively simple. The fish simply goes into a cage as a sign and stands there until someone manually sorts it out. It is the Bokelaxen we are talking about in the main case that comes every year. In addition, there are other unwanted species such as sea bream that will also be sorted out. It is a very fun cooperation. We are in a split between one of the world’s largest technology companies and a team from Berlevåg. Fantastic people in both camps. I am very happy in this part of Norway, up in the north, up in Finnmark. It is a fantastic part of our country. You have everything from midnight sun in the summer, like I am now. You have our winter, covered in snow. It is dark, and it is magical with northern lights. And I am also very concerned that we work in the care of nature and our rivers. The Bokelaxen project is supported by the Norwegian Government. Thank you. I would like to use some more numbers to show you how quickly this problem has become in Norway. Like all the other invasive species, the first time people have ever recognised or seen a humpback salmon in the Norwegian rivers was between the 1950s to the 1960s. And then, it remains a very low number for past years or past decades, until it is not. In 2017, people have seen more and more of these invasive species of humpback salmon, of more than 6,000 in total, in all Norwegian rivers. And this is a two-year life cycle species, which means they only come to Norway in odd number year. In 2019, all of a sudden the number raised up to 20,000. Can anyone guess how big is number in 2021? It’s 200,000. So that’s actually to everyone’s surprise and that’s when people realize if they’re not doing anything more, it’s very likely the whole ecosystem in the greater rivers will no longer exist. And in 2023, the number raised up to almost 600,000. But that’s actually not the real number because in many rivers, because of this invasive species becoming too big a problem, they stop counting. The fish experts believe, they have the reason to believe, the actual number is between 1 to 1.5 million. And we don’t know what’s going on in 2025. And as we speak, the fish is entering the fish, entering the Norwegian rivers from beginning of July or between middle of July. So that’s when they enter in the river. And right now, the rest of my team is in the northern part of Norway installing our system with the partners. Why we are trying to find a new way? This is how we used to do things before. Using diving method, people are diving into the rivers and see if there’s any invasive species. And once they see it, they capture it and they count it. But even though the water is really crystal clear, but it’s not spa temperature, it’s between 4 to 15 degrees. And it’s extremely cold for people. And you can imagine, when the number is raising up to such amount, it’s impossible for people just to use the traditional way to stop the invasive species. That’s why we have been approached by the local river association, trying to ask us if there’s anything that we can do with technologies. And we know it’s not easy, but after conversations, this is what we got. We first have a sketch on a napkin. And this is one year later. We have built the first system in Badevog in the north part of Norway. And then last year, we have another one. And this is what we have in a wider scenario. And we’re using also solar panels providing the non-stop electricity to the river. And this is the result we got. This is humpback salmon that we captured with the help of the system. 60,000. And this is how it works. Firstly, we block the whole river. So we create a channel that fish would like to swim in. And there, we put a camera at the beginning of the tunnel. We capture the fish. And the machine will be able to realize, identify the fish species within 0.01 second. And then, send a command to the ultimate door to decide if it’s going to close or open. If it’s a local species, we’ll open the door and allow the local species to swim further in the upstream of the river so that they can spawn. And if they are unwanted species, either it’s humpback salmon or maybe in the future, farmed fish, escaped farmed fish, we’ll guide them to the fish holding tank. So there, they’ll be kept alive. And because even though it’s an invasive species, it still has its value. We’ll give it to make dog food. We’ll give it to sell to the local market. So the local fisher association, they can have some income to give back to the community and continue the tremendous work that they’re doing. Furthermore, we’re providing a monitoring system. The front end will show 24-7 statistics in the river. So people will know what’s going on in their river so that they can know, make the right decisions for the river. If it’s going to be open or if it’s going to be closed. Because it’s important to keep the river open. I asked my local partner a very interesting question. If you think your local species stock is really decreasing and almost extinct, why don’t you just close down the river and don’t allow people to fish out? And their answer is also very interesting. They say, if they stop fishing, nobody will care the problem anymore. So they have to keep the river open and have the people still come into the river to create value for their local community and people will care. And this is how the fish is entering the fish tunnel box. So this is for the good fish, which is local Atlantic salmon. They’re going in the tunnel. Sorry, this is humpback salmon. They’re going to tunnel. We recognize it. The door remains closed and they will swim into the fish holding tank. And this is when a local salmon is identified. You can see the door opens before they’re approaching and they’ll close after they finish. So basically this way we allow the fish to go into the river and spawn without being touched. So they’re being separated without being touched. This is extremely important. Why? Because I think the fish species is quite sensitive and quite smart. It’s like people. When our economy is not good, when the world is not really as good as before, some families will decide not to have babies. So is the salmon. When they feel they will be handled and touched by human beings or other species, they will feel the river is no longer safe enough. So even though they’re in the river, they will not spawn. So that’s actually also a main reason why the local species is decreasing on its stock. So my point is, even though we have managed to find a way to build automated system for the river, but the AI part here is actually not the most fancy part. We’re using CNN. So it’s actually the basic AI model. But what is tricky here is how we communicate with local partners, with fish experts, with engineers to build up system that fish would like to go in because that is actually difficult. When you are doing the projects in the wild nature, it’s many, many unexpected risks. You will never see, you will never know until you face it. So you have to build up a system that is, you know, maximize the efficiency of solving the problem, but also minimize the disturbance on the nature and disturbance on the wildlife. So the fish will go in through the river untouched and unstressed. So that is actually the point and that requires effort. That requires back and forth dialogues with all different parts because we are speaking different languages. And to communicate and to understand each other requires enthusiasm. Only AI is not going to replace the enthusiasm of the local fishermen, of local community. So that is still the gold of how we can have the system that works best for the nature. And now we are also aiming for even more layers for the projects. We’re aiming for identify the farmed fish. So an ultimate target is to build up a total river management system. So here we are and we’re still working on it and we’ll try our best. My name is Monica, a proud project leader. Thank you for listening.

Agreement points

AI and technology should augment human capabilities rather than replace human workers

Speakers

– Silje Sande
– Kine Smordal Olsen
– Ragnar Thorsen

Arguments

Task-based labor model can solve both healthcare labor shortage and social exclusion by dividing full-time positions into small tasks accessible to community resources

AI transcription from audio recordings can make police interview documentation more efficient by eliminating manual transcription time

AI-based apps can transform unstructured texts into reports, potentially saving 80% of time spent on paperwork

Summary

All speakers emphasize that their AI solutions are designed to support and enhance human work rather than replace workers. Sande’s model creates new opportunities for excluded populations, while police representatives stress maintaining human oversight and responsibility.

Topics

Future of work | Digital standards | Capacity development

Human oversight and quality control remain essential in AI-powered systems

Speakers

– Kine Smordal Olsen
– Ragnar Thorsen
– Monica Cheng

Arguments

AI transcription from audio recordings can make police interview documentation more efficient by eliminating manual transcription time

AI-based apps can transform unstructured texts into reports, potentially saving 80% of time spent on paperwork

System design must minimize disturbance to wildlife while maximizing efficiency, requiring collaboration between technology experts and local communities

Summary

All speakers acknowledge that while AI can automate processes, human verification, quality assurance, and decision-making remain crucial. Police officers maintain responsibility for report content, and environmental systems require human expertise for proper implementation.

Topics

Digital standards | Future of work | Sustainable development

Similar viewpoints

Technology can significantly improve efficiency and productivity in public sector work while maintaining human control and responsibility. All three speakers present solutions that automate routine tasks to free up human resources for more valuable work.

Speakers

– Silje Sande
– Kine Smordal Olsen
– Ragnar Thorsen

Arguments

Task-based labor model can solve both healthcare labor shortage and social exclusion by dividing full-time positions into small tasks accessible to community resources

AI transcription from audio recordings can make police interview documentation more efficient by eliminating manual transcription time

AI-based apps can transform unstructured texts into reports, potentially saving 80% of time spent on paperwork

Topics

Future of work | Digital standards | Capacity development

Successful technology implementation requires understanding and adapting to the specific needs and constraints of the target environment, whether human social systems or natural ecosystems. Both emphasize the importance of collaboration and communication across different expertise areas.

Speakers

– Silje Sande
– Monica Cheng

Arguments

Micro-learning modules can educate people for specific tasks without requiring complex education, creating easier career entry paths

System design must minimize disturbance to wildlife while maximizing efficiency, requiring collaboration between technology experts and local communities

Topics

Capacity development | Interdisciplinary approaches | Sustainable development

Unexpected consensus

Community engagement and local partnership are essential for successful technology implementation

Speakers

– Silje Sande
– Monica Cheng

Arguments

Task-based labor model can solve both healthcare labor shortage and social exclusion by dividing full-time positions into small tasks accessible to community resources

System design must minimize disturbance to wildlife while maximizing efficiency, requiring collaboration between technology experts and local communities

Explanation

Despite coming from very different sectors (municipal healthcare and environmental technology), both speakers emphasize that technology solutions must be deeply integrated with local communities and stakeholders. This consensus is unexpected because it shows that both social and environmental technology challenges require similar collaborative approaches.

Topics

Cultural diversity | Interdisciplinary approaches | Sustainable development

Simple AI models can be highly effective when properly implemented

Speakers

– Monica Cheng
– Kine Smordal Olsen
– Ragnar Thorsen

Arguments

Automated fish sorting system using AI can identify and separate invasive humpback salmon from native species within 0.01 seconds

AI transcription from audio recordings can make police interview documentation more efficient by eliminating manual transcription time

AI-based apps can transform unstructured texts into reports, potentially saving 80% of time spent on paperwork

Explanation

All speakers demonstrate that relatively basic AI technologies (CNN models, speech-to-text, text processing) can deliver significant practical benefits when applied thoughtfully. This consensus is unexpected because it challenges the notion that complex AI is necessary for meaningful impact.

Topics

Digital standards | Future of work | Sustainable development

Overall assessment

Summary

The speakers demonstrate strong consensus on human-centered AI implementation, the importance of community collaboration, and the effectiveness of practical AI applications in solving real-world problems across municipal, law enforcement, and environmental sectors.

Consensus level

High level of consensus with significant implications for technology policy and implementation. The agreement suggests that successful AI deployment requires human oversight, community engagement, and focus on augmenting rather than replacing human capabilities. This consensus provides a framework for responsible AI implementation across diverse public sector applications.

Different viewpoints

Unexpected differences

Human-AI interaction philosophy

Speakers

– Silje Sande
– Monica Cheng

Arguments

Micro-learning modules can educate people for specific tasks without requiring complex education, creating easier career entry paths

System design must minimize disturbance to wildlife while maximizing efficiency, requiring collaboration between technology experts and local communities

Explanation

While both speakers advocate for technology solutions, they reveal different philosophies about human-AI interaction. Sande promotes using AI to simplify and standardize human tasks through micro-learning, essentially reducing complexity for human users. Cheng emphasizes that AI is ‘not the most fancy part’ and stresses that technology must adapt to natural behaviors (both fish and human community needs), requiring complex collaboration. This represents an unexpected philosophical divide between AI-simplified human adaptation versus AI adaptation to natural/existing behaviors.

Topics

Sustainable development | Capacity development | Cultural diversity

Overall assessment

Summary

The speakers show remarkable alignment on using AI and technology to solve societal challenges, with minimal direct disagreements. The main area of difference lies in implementation philosophy rather than goals.

Disagreement level

Low level of disagreement with significant implications for technology implementation strategies. The speakers represent different sectors (municipal services, law enforcement, corporate technology) but share common ground on technology’s potential. The subtle philosophical differences about human-AI interaction could influence how similar technologies are designed and deployed across different domains, suggesting the need for sector-specific approaches to AI implementation.

Partial agreements

Similar viewpoints

Technology can significantly improve efficiency and productivity in public sector work while maintaining human control and responsibility. All three speakers present solutions that automate routine tasks to free up human resources for more valuable work.

Speakers

– Silje Sande
– Kine Smordal Olsen
– Ragnar Thorsen

Arguments

Task-based labor model can solve both healthcare labor shortage and social exclusion by dividing full-time positions into small tasks accessible to community resources

AI transcription from audio recordings can make police interview documentation more efficient by eliminating manual transcription time

AI-based apps can transform unstructured texts into reports, potentially saving 80% of time spent on paperwork

Topics

Future of work | Digital standards | Capacity development

Successful technology implementation requires understanding and adapting to the specific needs and constraints of the target environment, whether human social systems or natural ecosystems. Both emphasize the importance of collaboration and communication across different expertise areas.

Speakers

– Silje Sande
– Monica Cheng

Arguments

Micro-learning modules can educate people for specific tasks without requiring complex education, creating easier career entry paths

System design must minimize disturbance to wildlife while maximizing efficiency, requiring collaboration between technology experts and local communities

Topics

Capacity development | Interdisciplinary approaches | Sustainable development

Key takeaways

Technology and AI can address critical societal challenges including healthcare labor shortages, social exclusion, and environmental protection

Task-based labor models combined with micro-learning can simultaneously solve healthcare staffing issues and social exclusion by making work accessible to untrained community members

AI transcription and documentation tools can significantly improve efficiency in police work, potentially saving up to 80% of paperwork time while maintaining human oversight

Automated AI systems can effectively protect ecosystems from invasive species while preserving native wildlife through precise identification and separation

Western societies face a dual crisis of aging populations creating healthcare demands while rising social exclusion reduces available workforce

Successful technology implementation requires collaboration between technical experts, local communities, and domain specialists to ensure practical effectiveness

Human oversight remains essential in all AI applications – technology augments rather than replaces human judgment and responsibility

Resolutions and action items

Continue pilot testing of AI transcription tools with 20 patrol officers in Norwegian police

Expand task-based labor model testing beyond initial nursing home and home service experiments

Install fish sorting systems in northern Norway rivers during the current salmon season

Develop proof of concept for AI-based report generation from unstructured text

Integrate AI-generated summaries into police interview documentation process

Scale up automated fish sorting system to create comprehensive river management solution

Unresolved issues

How to overcome technical challenges like background noise, dialects, and intoxication affecting AI transcription quality

Scaling the task-based labor model beyond pilot programs to address the broader 18% social exclusion rate

Long-term sustainability and funding mechanisms for innovative municipal solutions

Integration challenges between new AI systems and existing criminal case management systems

Addressing the exponential growth of invasive salmon populations (potentially 1-1.5 million in 2023) across all Norwegian rivers

Balancing river accessibility for community engagement while protecting native species

Developing standardized approaches for technology implementation across different municipalities and regions

Suggested compromises

Maintaining human-in-the-loop approach for all AI systems to ensure accuracy and accountability

Keeping rivers open for fishing to maintain community engagement while implementing protective technology

Using invasive species for economic benefit (dog food, local markets) rather than waste to support local communities

Combining traditional methods with new technology rather than complete replacement

Balancing maximum problem-solving efficiency with minimal disturbance to natural environments and wildlife

The irony is not lost on us that we lack people in healthcare at the same time as the exclusion rates grows. So we need to address both problems at the same time, the theory being that they can partly cancel each other out.

Speaker

Silje Sande

Reason

This comment is profoundly insightful because it reframes two seemingly separate societal problems as interconnected solutions. Rather than viewing healthcare worker shortages and social exclusion as distinct challenges, Sande identifies them as complementary problems that can address each other. This systems thinking approach demonstrates innovative problem-solving that looks beyond traditional silos.

Impact

This insight fundamentally shifted the presentation from discussing a single innovation to revealing a comprehensive social innovation model. It established the intellectual foundation for the task-based labor model and elevated the discussion from operational efficiency to addressing systemic societal challenges.

Our job market is not suited for those who fail to deliver on these elements, and it is becoming a very excluding factor that we cannot afford, neither for human nor economic reasons.

Speaker

Silje Sande

Reason

This comment challenges the fundamental assumptions of modern labor markets by questioning whether our current employment structures are fit for purpose. It’s thought-provoking because it suggests that the problem isn’t with excluded individuals, but with exclusionary systems that waste human potential during a time of labor shortage.

Impact

This observation provided the critical justification for radical innovation in employment models. It moved the conversation beyond incremental improvements to questioning fundamental structures, setting up the rationale for the task-based labor model as a necessary disruption rather than just an interesting experiment.

Using AI, we must always have a human in the loop. The officer must read through, verify, and ensure that the text reflects the content of the interview. The police officer is always responsible for the content of the report.

Speaker

Kine Smordal Olsen

Reason

This comment is insightful because it addresses one of the most critical concerns about AI implementation in sensitive areas like law enforcement. It demonstrates mature thinking about AI adoption – embracing efficiency gains while maintaining human accountability and oversight, especially crucial in contexts where accuracy affects justice outcomes.

Impact

This comment established a responsible framework for AI implementation that likely influenced how the audience viewed the technology demonstrations. It shifted the discussion from pure technological capability to responsible deployment, adding credibility to the police innovation project.

Only AI is not going to replace the enthusiasm of the local fishermen, of local community. So that is still the gold of how we can have the system that works best for the nature.

Speaker

Monica Cheng

Reason

This comment is particularly thought-provoking because it acknowledges the limitations of technology while highlighting the irreplaceable value of human passion and local knowledge. Coming from a technology company representative, it shows remarkable humility and understanding that successful innovation requires human-technology collaboration, not replacement.

Impact

This comment provided a philosophical capstone to all three presentations by emphasizing that technology serves human purposes rather than replacing human values. It unified the theme across all presentations that innovation succeeds when it amplifies human capability and community engagement rather than substituting for it.

Overall assessment

These key comments shaped the discussion by establishing a sophisticated framework for understanding innovation as a tool for addressing complex societal challenges rather than just technological advancement. Sande’s insights about interconnected problems set an intellectual standard that elevated the entire session, while the police and Huawei presentations built upon this foundation by demonstrating responsible AI implementation and community-centered innovation. The comments collectively shifted the conversation from ‘what technology can do’ to ‘how technology should serve society,’ creating a cohesive narrative about human-centered innovation across three very different domains – social services, law enforcement, and environmental protection.

How can micro-learning modules be effectively designed and implemented to educate community resources for specific healthcare tasks?

Speaker

Silje Sande

Explanation

This is crucial for the success of the task-based labor model as it determines how well untrained people can be prepared for healthcare tasks while maintaining quality and safety standards.

What are the specific processes that need to be automated to support the task-based labor model effectively?

Speaker

Silje Sande

Explanation

The first experiment revealed the need for process automation, but the specific processes and automation requirements were not detailed, which is essential for scaling the model.

How can AI-generated summaries be integrated into police interview documentation to further streamline the process?

Speaker

Kine Smordal Olsen

Explanation

This represents the next evolution of their AI transcription system and could significantly reduce the time officers spend on documentation beyond just transcription.

What technical improvements are needed to address transcription quality issues caused by background noise, dialects, intoxication, and stress?

Speaker

Kine Smordal Olsen

Explanation

These challenges directly impact the reliability and usability of the AI transcription system in real-world police work scenarios.

How can the AI-based app that transforms unstructured text into reports be developed from prototype to full implementation?

Speaker

Ragnar Thorsen

Explanation

This technology shows potential for 80% time savings on paperwork, making its development critical for improving police efficiency in crime scene documentation.

What will be the actual number of humpback salmon in Norwegian rivers in 2025, and how can this be predicted or monitored?

Speaker

Monica Cheng

Explanation

Understanding future invasion patterns is crucial for environmental planning and resource allocation for containment efforts.

How can the AI system be expanded to identify and manage escaped farmed fish in addition to humpback salmon?

Speaker

Monica Cheng

Explanation

This expansion would create a comprehensive river management system addressing multiple invasive species threats to Norwegian ecosystems.

What are the long-term ecological impacts and effectiveness of the automated fish sorting system on river ecosystems?

Speaker

Monica Cheng

Explanation

Long-term monitoring is needed to ensure the system successfully protects native species while minimizing environmental disruption.