Towards actionable governance on trustworthy AI

Today, artificial intelligence (AI) technologies are present in most spheres of our life, from driverless cars and finance to healthcare solutions. In 2022, IBM found 35% of companies were using AI in their operations, while 42% of businesses indicated that they are exploring the use of AI.

Moreover, Gartner predicts that by 2025 more than 30% of new medications and materials will be discovered using generative AI techniques.

This rise in AI across industries is also reflected in its projected market growth. In 2022, the global AI market was estimated at $120 billion; forecasts expect its worth to reach almost $1.6 trillion by 2030.

Whereas AI technologies hold numerous benefits which contribute to advances of society, as a dual-use technology it also contributes to profound economic and societal disruptions that can impact our privacy, safety and wellbeing and, as such, cannot be ignored by governments, the private sector or by individuals.

To amplify the benefits and mitigate the risks associated with AI, in recent years public and private organizations have developed governance frameworks intended to guide the development and implementation of ethical, fair and trustworthy AI.

To illustrate, the OECD Council Recommendation on Artificial Intelligence, adopted in May 2019, identify the five key principles for responsible stewardship of trustworthy AI that should be implemented by governments, organizations and individuals. Today, these principles serve as a reference framework for international standards and national legislation on AI.

Other examples of international and regional efforts on AI governance include UNESCO's Recommendation on the Ethics of AI, the Council of Europe’s proposal for a legal framework on artificial intelligence based on the Council`s standards on human rights, democracy and the rule of law, as well as a series of documents that define the EU’s approach to AI. The EU’s “Artificial Intelligence Act”, for instance, proposes a legal framework for trustworthy AI.

At the national level, governments have developed AI strategies, which, among other things, outline approaches for trusted and safe AI.

From the private sector perspective, a number of businesses, including Google and Microsoft, have developed governance tools and principles for building AI systems responsibly that also outline practical approaches how to ensure that the unintended consequences of the technology are avoided.

However, despite these different public and private governance efforts, research shows that a majority of organizations are yet to take actionable measures to ensure the development and use of trustworthy and responsible AI. For instance, more than 70% of organizations have not adopted the necessary steps to eliminate bias in AI, while an estimated 52% are not capable of safeguarding data privacy through the entire AI lifecycle

To move from policy to practice on trustworthy AI, organizations should devise and implement clear and structured programmes that guide the implementation of AI governance frameworks. Among other things, these programmes should:

As a technology that transcends industries and geographies, the development and implementation of trustworthy AI is not a one-man job and as such requires collaboration amongst researchers, developers, businesses and policymakers. Society at large has the responsibility to ensure that AI systems align with social values and promotes the public good.