But why does a waste management company need an AI strategy?
“AI is everywhere and everybody is now using tools like ChatGPT and Claude,” says Heiðbjört Vigfúsdóttir, Sorpa’s IT manager.
Read this article in Norwegian on Arbeidsliv i Norden
“We want to protect all our data, but we also want to train our staff to do this. AI offers many positive and useful opportunities, but we have to have some ground rules.
“Sorpa will probably have to comply with the NIS2 directive on cyber security as well as the EU AI Act. So everybody in IT must adapt to that with a strategy and staff training in order to use AI correctly and in the best way possible.”
Sorpa is just starting to expand its use of AI.
“We are currently developing plans for our future use of AI. One possibility we are exploring is using AI-powered image analysis to improve waste management and sorting at our centres.
“For example, the public could take pictures of the waste, and the tool tells them where it is supposed to go and even where it ends up.”
More information on where waste ends up
This will lead to easier delivery and disposal and also improve transparency.
“A lot of people want to know what happens to their waste but there have been mixed messages on that. Some even believe that all sorted waste ends up in the same bin. That is a misconception, and we want to show in a clear way where each type of waste goes.
“The main aim for our strategic planning is to better inform the public. We want to review our data and make sure it is accurate so we can use AI for business cases in the best possible way.
“Reliable information about what happens to the waste will be accessible to the municipalities, business partners and the public,” Vigfúsdóttir says.
Part of the image analysis process will be implemented soon and will help waste management centre staff.
“When staff receive hazardous waste, they can take a picture of it and the tool can instantly analyse which category it belongs to.
“We are currently testing this, and the first attempts are promising. It will be a big help for the staff because a recent change in hazardous waste management legislation has increased the number of waste categories from two to eight.
“This tool will make that categorisation process a lot faster. And all this information will then be part of our data.”
Improving second-hand products store
Sorpa’s IT department is also looking at using AI to help maintain and update their web pages.
“That goes both for Sorpa and the second hand store Góði hirðirinn (Icelandic for Good Shepherd). We’d like the marketing team that supervises the pages to be more independent and able to update the pages faster,” Vigfúsdóttir says.
She adds that this and other use of AI could also lead to a more efficient operation.
One of the challenges of using AI is the safe handling of personal data.
“Security is the most important issue,” Vigfúsdóttir says.
“If we develop AI tools and use personal data, we have to make sure the data is categorised correctly. It needs to be encrypted so that it can’t be read without the right key.
“We have a data protection officer who regularly reviews how data is stored and categorised. Of course, we assess what data we work with on each occasion but the less data we need to use, the better.”
Vigfúsdóttir says the company has not started to use so-called AI agents to do a whole project – that is something they want to be really careful with.
“The first thing we want to do is raise staff awareness of AI. That takes time. We also aim to have our own local AI system that operates in a secure environment where each employee has their own access and can work with our company data in a secure way.
“That way our staff can be less worried about security. And in this system our staff quality manual will be available so they can ‘talk’ to it, so to speak, instead of searching through a big pile of files. This will also help train staff in using AI.”
Learning to challenge AI outcome
Another challenge when using AI is to know when to trust its output.
“This needs to be shown, trained and discussed,” Vigfúsdóttir says.
“We intend to have a special staff meeting bringing up some examples of where AI can simplify life but also how it can make mistakes.
“It is a big challenge to make people realise that AI is not always right. Staff need to assess it, look at it critically and even challenge it. That can be tricky, but right now we simply have to train our staff.”
Vigfúsdóttir says that the next AI tool that will be used is the previously mentioned image analysis, which she believes will increase the percentage of waste that is sorted correctly.
“This could mean that customers who arrive with waste that has already been sorted can use self-service instead of going through a manned service gate.
“It could save a lot of work since our staff don’t have to check beforehand what is being disposed of, and it also makes it easier for the public. In general, we want to use this to improve our service.”
The use of sensors to help with maintenance is another potential AI benefit, explains Vigfúsdóttir.
“AI can use data from sensors to tell us if something is not right, which allows us to repair equipment before something seriously fails.”
Suggesting product price
There is also an experiment going on at Góði hirðirinn.
“We are now working on its online store, which reaches a new group of customers: people who like to browse but don’t necessarily visit the store regularly.
“Now a staff member can take a picture of the second-hand item, and the AI tool suggests a description and a price. This is then reviewed by online store staff before it is published. It makes the process a lot faster.”
And the aim is also to help office staff with work.
“We want them to use AI to work more efficiently and thoroughly, while helping their colleagues embrace new ways of working. This also applies to our service centres, helping them channel enquiries to the right place more quickly.
“We are just getting started on all this now, preparing the ground to be able to make more use of the technology.”
All this should lead to a better service.
“We want to work in a smart way. Our office team is relatively small compared to the scale of our operations, which gives us a unique opportunity to adopt AI quickly through teaching and training.”
