Our supervisory activities have shown that the use of artificial intelligence (AI) is increasing.

The Norwegian government's national strategy for AI was published in 2020. It emphasises the need for sectoral supervisory authorities to ensure that the application of AI in all industries is in compliance with the principles of responsible use.

Definitions

There are many definitions of AI, which tend to vary in step with technological development.

At Havtil we have not developed our own definition, but have chosen to adopt those set out in the aforementioned national strategy, where AI is described as systems that perform physical or digital actions based on the interpretation and processing of structured or unstructured data with the aim of achieving a given goal.

AI encompasses a number of different approaches and techniques, such as machine learning.

Risk-based

The EU has developed new rules for the application of AI (the AI directive). Political consensus has been achieved regarding the directive, and the plan is to incorporate it into the EU system. It is founded on a risk-based approach to regulation and deals in terms of risk categories.

This means that the requirements for a given system will depend on the risk context in which it is applied. Most of the requirements set out in the directive focus on high-risk systems.

Whether or not an AI system is defined as high-risk will depend on factors such as the degree to which it represents a risk to critical infrastructure, the health and safety of its users, or their fundamental rights.

Milestone

The directive represents a key milestone in the regulation of AI. The risk-based approach demands that sectoral supervisory authorities monitoring the major accident industries have to make the necessary preparations and appraisals.

At the same time, there have also been developments in the fields of accessibility research, standardisation and best practice in this field.

AI models currently used in operations, planning and decision support will be able, directly and indirectly, to impact on safety, the working environment, emergency preparedness and security.

Distinct

In the ocean industries, we are anticipating that AI systems that can influence risks linked to major accidents, security or safety systems, barriers or critical infrastructure will be classified as high risk. AI-related risks are distinct from traditional industrial risk factors and from those linked to standard IT and software systems.

New systems such as AI can also help to reduce risk, but their effectiveness will depend on the industry becoming aware of inherent and contextual risk factors, as well as the need to apply a structured approach to their identification and management.

In this context, and based on experience from previous studies and our supervisory work, Havtil has launched activities focusing on establishing an effective regulatory framework for the proper use of AI in the industry.

Activities in 2023

In 2023 we have held meetings to address the companies’ verification and risk management activities linked to the development and use of the technologies currently being integrated into a digital system controlling the use of safety-related equipment and systems.

We invited selected operator and supplier companies to share their experiences of risk monitoring linked to digital technologies. We have also been analysing fundamental legal challenges linked to the use of AI in the petroleum sector.

This will help us to clarify future regulatory needs in anticipation of developing a sound regulatory framework and preparing the industry.

Awareness

We are currently in dialogue with Norwegian technology centres with the dual aim of identifying collaborative opportunities and establishing an AI network. Our intention here is to assist in focusing awareness in both academia and the industry on safety-related risk factors linked to developing trends and the use of AI in the petroleum sector.

The International Regulators’ Forum (IRF) is working with digital technology and automated systems as one of its three highest priorities.

As digital technology continues to advance and is being applied by the companies, collaboration between the authorities, the International Association of Oil and Gas Producers (IGOP) and the International Association of Drilling Contractors (IADC) will aim to help the industry to put safety and the working environment high on the agenda.

Impact

These activities have enabled Havtil to focus awareness on AI and safety-related risk in the ocean industries. We have consolidated our dialogue with the industry and research communities on issues related to the risk management of AI systems and have encouraged the companies to express any specific challenges they may be facing.

This has contributed to fruitful exchanges on our collective approaches to AI-related risk management. The companies have responded positively to the way in which Havtil has been putting the spotlight on AI.

This article has been taken from the 2023 Havtil Annual Report, submitted to the Norwegian Ministry of Energy. Read the Annual Report in full here: Annual report 2023: The Norwegian Ocean Industry Authority