Alex Moltzau

Have you considered an algorithm register for your organisation? There are several platforms to share your AI models, with perhaps the most common being Hugging Face. If you can create a place to access your models it may both make it simpler for your team to collaborate and others to navigate. AI is governed in organisations, and it could be helpful to describe models with model cards. A model card explains what a model does, its intended audience, and who maintains it. If you are selling to the public sector and transparent decision-making is a wish by the buyer, this could be one helpful approach. It could also help build a sharing culture. Governments should also carefully work to share algorithms when it impacts citizens lives.

Quality of AI models will be influenced by the organisational culture and statements that are constructed as truth. The 'move fast and break things' attitude was fine until it broke democratic discourse in some countries and partly contributed to a genocide (Myanmar, Meta). What are the qualities beyond the accuracy of your model? Governance of quality is about more than a risk assessment, it is how you govern an organisation and what responsibilities you can take. How you treat people as human beings and not machines in all areas of your AI value chain. That will show not only the quality of a single AI model, but the qualities of those who work in your organisation. Of course quality management is important, but how is your culture built?

Much like individuals may want to have an overview of their portfolio of algorithms, presentation of this may be important for a company in how they manage their algorithms. As such, as an added benefit, it could equip individuals with AI governance skills. Many companies do not have an overview of either their software, algorithms or models. In this way, having to communicate your algorithms may be a vital skill needed in companies or organisations with a range of stakeholders. How can an organisation govern well without this level of coordination? Of course principles are important and autonomy for developers, yet increased ability to coordinate may have added benefits in terms of organisational learning or avoid duplication of work.

In Norway basic training for police officers is a three-year university college education. In the US it could be 22 weeks. If we throw advanced AI technology at the police force without either the technical competencies or deep ethical training how can we expect AI to be used responsibly? We must consider what mandatory requirements we should implement for using AI systems that attempt to assist in police work. Also what drastic limitations that may follow. If certain software or AI models are too discriminatory to minority or vulnerable groups the use of these should be banned. However, to discuss this topic countries need advanced capabilities or collaboration on AI safety that includes a human rights approach.

Generative AI (genAI) in the public sector needs careful consideration depending on context, as there added legal responsibilities — for example in Norway — that both the buying entity and vendor needs to consider. When it comes to implementation domain knowledge is highly important. Consider running ChatGPT for all teachers and students in a given municipality. Furthermore, with young children content filtering may provide greater challenges and the potential fallout in the media should accidents or omissions occur. This does not mean generative AI should shy away from the public sector, instead including teachers and workers in different areas to adapt the genAI product. Nations that want responsible AI need coordination.

I do not think I need AI to be a part of the communication with my mentor as a tool, that would not be very helpful. Let us respect what it means to be human. However, when it comes to connecting with a mentor on the topic of AI it may help to be specific and ask who could provide help towards your progression. I asked my leader who could help me with science policy on a national level in the field of AI and he found a perfect candidate with a background as a scientific attaché to the US willing to help. He had shaped a bilateral agreement on AI, yet had deep knowledge in both the field I was interested in and the structures surrounding it. You do not need AI to make it clear to your leader that you wish for mentorship suggestions.