How we are exploring using artificial intelligence

In this article, Matthew Belsey, Environmental Reporting and Data Manager at Culture for Climate Scotland, reflects on how we have been exploring artificial intelligence (AI) – sharing an example of how we used AI within our own work, as well as his experience as a panellist at Culture & Business Scotland Forum25 discussing ‘Responsible AI in the Creative Industries’.

If you haven’t already read it, check out our article ‘Environmental impact of AI’, where Matthew gives an overview of what AI is and the environmental impacts of this technology.

Incorporating AI at Culture for Climate Scotland

Last year, we were approached by Statistics Without Borders with the offer of delivering a pilot project to improve our data capacity. This coincided with our need to expand our environmental reporting programme to support over 260 organisations (compared to the previous 130 organisations in 2023/24).

We were keen to explore using AI within our work where the benefits would outweigh the additional environmental impact, and agreed this was a good opportunity to trial it on a project basis in a controlled setting.

We set out to deliver a project incorporating Natural Language Processing (NLP) into our analysis and feedback writing process. NLP uses machine learning to help computers understand human language according to IBM’s definition in their ‘What is NLP?’ article. This project was extremely useful for us to determine whether AI was relevant to this area of our work, as well as getting experience using AI tools within our own specific context – working at the intersection of climate change and culture. You can read a full summary of the project written by Statistics Without Borders.

After reviewing the outputs of the NLP model, we decided that the summaries provided didn’t quite understand the nuance of the 260 different reporting organisations at the level that Culture for Climate Scotland staff does. This would have resulted in a less thorough feedback process, giving the reporting organisations less detailed feedback on how they could reduce their carbon footprint, adapt to the changing climate and influence their audiences. By removing the current hands-on nature of the feedback writing where our team reads every response, we would have also removed the in-depth understanding we have of all the organisations that we work with. This is fundamental to the way that Culture for Climate Scotland works. While AI models in the future might be able to understand and implement this nuance, this model wasn’t yet able to meet our standards and requirements.

Statistics Without Borders also produced an energy consumption report of the AI model. They predicted that running the AI model used approximately the same amount of energy as one member of Culture for Climate Scotland staff using their laptop for a week. It could be argued that since the work that the AI model did saves 3-4 weeks of work, running it would be saving 2-3 weeks’ worth of energy. However, this poses the question of what the staff member would do with the 2-3 weeks saved. It’s likely that they would take on some more work or do more analysis, still using their laptop. Therefore, this example has shown us how using the AI model could actually cause one week’s additional energy consumption, compared to if we didn’t use it.

In conclusion, while this may not have been the best use case for applying AI in our work, it taught us a very important lesson that AI isn’t necessarily the silver bullet for all efficiency improvements.

Culture & Business Scotland responsible AI panel discussion

I was invited to join a panel on ‘Responsible AI in the Creative Industries’ at the Culture & Business Scotland Forum25 in December 2025, along with Kjersti Fergusson (Head of Intelligent Automation, Scottish Government), Dr Smita Kheria (Senior Lecturer in Intellectual Property Law, University of Edinburgh) and Morgan Petrie (Creative Industries Manager, Creative Scotland).

It was a constructive and informative discussion in which I focused on encouraging people to question what we do with the time we save through using AI. How can we use this tool driven by capitalist forces not to be more ‘productive’ (in the financial sense of the word), but use our saved time to rest, be creative and do things that we know AI can’t do? Using that time for the benefit of our wellbeing and the planet will be vital as/if we transition to a new AI-dominated way of working. This resonated with many people in the audience.

The other panellists brought interesting points and perspectives to the discussion. Kjersti highlighted that when most people come to her department with ideas for AI, what they need is a digital algorithm that can be implemented without AI. The knowledge of when to use AI, and when not to, is important to consider, because a non-AI solution could be more sustainable. She also highlighted that when replacing repetitive tasks, there is very little resistance to AI since it frees people up to do strategy and leadership work. Again, this links into thinking about what you do with the time saved if you decide to use AI.

Both Kjersti and Smita talked about these models’ use of our data. Kjersti highlighted that free-to-use tools like ChatGPT are not free. You are paying for it with the data that you put into the model. It’s important to check the terms of use when using AI.

Smita emphasised that you should treat AI use like any other procurement process. If you were using a new supplier, you would want to know who is making the item and how it is made. The same should be the case for AI. This goes for the sustainability of the model too. Does the AI model you are using meet the requirements of your organisation’s sustainability policy?

Morgan encouraged attendees to undertake human impact assessments before introducing AI into their systems or artistic practices. This can incorporate equality, fair work, sustainability, or other key values and priorities. Morgan also touched on some artists that are using AI responsibly including Tim Murray Brown and Rachel MacLean.

Although this discussion was fruitful in terms of getting people to think about responsible AI and its use in the arts and culture sector, these ethical considerations weren’t a core theme that was echoed in the remainder of the forum’s programme. This suggests that there is still a lot of work to ensure that as the sector adopts AI, we do so in a fair, just, transparent and ethical way

Our outlook for the future

Despite the benefits of using AI not outweighing the cost of using it (in terms of increased emissions) for our project with Statistics Without Borders, there might be ways we can use it in the future. For example, we could synthesise and share insights with our different audiences, increase our data analysis capacity, and offer front-line automated support for organisations and artists looking to reduce their emissions. However, we don’t believe any current mainstream AI models would pass our ethical and sustainable procurement checks. There are obviously issues around energy, water and resource use, but also concerns about privacy, bias and transparency.

As we continue our AI journey, I am now exploring how decentralised, hyper-specialised models might enable us to continue delivering support to the Scottish arts and culture sector, while still ensuring the technology we use aligns with the world we want to see.

Beyond Culture for Climate Scotland’s work, I hope to see more thoughtful use of AI by the sector, and greater and regulation to ensure that any benefits from AI development are shared fairly, and that the transition to an AI-powered world is a just one.

What you can do

• Use The Alan Turing Institute’s ‘Defining data science and AI glossary’ to familiarise yourself with AI key terms and definitions.
• Think about why you are using AI before you use it. Is it something you can easily do without AI? How will you ensure the answer from AI is accurate? How can you limit the number of queries you ask AI or tasks that AI does to get your output?
• Think about what you will do with the time saved through AI – if it is likely that the integration of an AI tool will make your work drastically more efficient, it is likely that you will require a strategic discussion on how your role will change.
• Create an AI policy and an AI acceptable use policy for your organisation. Charity Digital has a useful ‘How to create an AI policy’ guide for charities.
• Explore the Ada Lovelace Institute’s blog to read more about how AI can be used fairly and responsibly for people and society.

If you would like to discuss the themes covered in this article further or have any questions, please contact [email protected].

Header image: Graphic adapted from a photograph by teteescape on Canva.