To explore different perspectives on generative AI, Compare hosted a networking breakfast, where insights from research and concrete examples of how companies are successfully using generative AI were presented.
Around eighty participants gathered on 18 January at Innovation Park. For just over an hour, they gained knowledge about generative AI, based on the latest research at Karlstad University and concrete examples of how the companies Uddeholm and Forgebyte Studios use generative AI in their businesses.
Ingrid Ivars, Digital Transformation Manager at Compare.
Research and user benefits
Bestoun S. Ahmed Al-Beywanee, Professor of Computer Science at Karlstad University, guided the audience through the evolution of AI research, from basic data analysis to advanced machine learning models and formal processes. Today, the focus is on managing AI in a holistic way, which means taking into account the entire process or lifecycle that an AI system goes through, from start to finish.
Johan Netz, Senior Lecturer, Centre for Service Research at Karlstad University, stressed that it is important to put user benefits in focus when it comes to services and the use of AI. Johan believes that there are many good opportunities and that the costs of service development can be reduced by using AI, provided that we understand the problem well.
- “It is important to understand and identify problems and then develop solutions using AI,” says Johan. “Start small, test and scale up gradually,” he recommends.
Johan Netz, Senior Lecturer, Centre for Services Research at Karlstad University.
Improving efficiency with Data Analytics and AI
Uddeholm is a leading tool steel manufacturer and uses data analytics and AI to generate added value in the industry. Erik Hallin, Head of Data Analytics & AI at the company, exemplified this by describing how Uddeholm uses data analytics and AI to optimise the start-up time process of an electric arc furnace where scrap is melted for steelmaking, resulting in a reduction of energy costs.
- “We always start the process by identifying what the problem is, then we take in data, store data, transform, model and visualise,” says Erik. “The competences we need to do that include domain management, data warehousing, simulation and machine learning.
To create an AI model, an algorithm is trained by feeding it with relevant data and informing it of the correct answer. The algorithm learns to identify patterns and make predictions, resulting in a model that can be used to make similar predictions on new examples.
- Through data analysis and AI, we go from idea to real savings,” says Erik. A key aspect of AI is good control and management of the data, which is not as fun to talk about as ChatGPT, but we work a lot with it at Uddeholm because it is necessary to actually use AI.
Image generation with AI
Staffan Olsson, CEO and founder of Forgebyte Studios, an indie game development studio, gave a practical talk on how to work with image generation through the use of AI, showing how he created a visualisation of a watch brand that did not previously exist by training his own model. Staffan does not believe that AI will lead to the replacement of staff in companies such as advertising agencies; instead, he believes that by integrating AI into their work, they will be able to reach a wider customer base and maintain a higher quality of service.
- “As I see it today, no one needs to be replaced, but instead the quality will increase,” says Staffan. “But we all need to adapt, and if you work with the visual medium, you should adopt AI tools today.
Compare’s next networking event
On 7 March, we welcome those interested in the Internet of Things (IoT) to our next networking breakfast. Compare’s networking breakfasts are open to all and free of charge. Keep an eye on compare.se for more information about the next event.
The networking breakfast on 18 January was arranged in collaboration with our project partner Region Värmland.
