Our Global Agreement on AI Can Reduce Bias and Oversight

A CCTV camera

NYstudio/Getty Images

Artificial intelligence is more present than ever in our lives: it predicts what we want to say in emails, helps us navigate from A to B and improves our weather reports. The unprecedented speed at which vaccines for Covid-19 have been developed can also be attributed in part to the use of AI algorithms that rapidly processed data from numerous clinical trials, allowing researchers around the world to compare notes in real time.

But the technology is not always beneficial. The datasets used to build AI are often not representative of population diversity, so it can create discriminatory practices or biases. An example is facial recognition technology. This is used to access our cell phones, bank accounts and apartment buildings, and is increasingly being used by police forces. But it can have problems accurately identifying women and black people. For three such programs released by major tech companies, the error rate was just 1 percent for fair-skinned men, 19 percent for dark-skinned men, and up to a whopping 35 percent for dark-skinned women. Prejudice in facial recognition technologies has led to wrongful arrests.

That’s no surprise when you look at how AI is being developed. Only 1 in 10 software developers worldwide is female and only 3 percent of employees at the 75 largest technology companies in the US identify as black. But now there is hope that the world is about to turn to a much better approach.

Yesterday, 193 countries at UNESCO reached a groundbreaking agreement on how AI should be designed and used by governments and tech companies. UNESCO’s Recommendation on the Ethics of Artificial Intelligence took two years to complete and involved thousands of online consultations with people from a wide range of social groups. It aims to fundamentally shift the balance of power between people and the companies and governments that develop AI.

Countries that are members of UNESCO – and this is almost every country in the world – have agreed to implement this recommendation by enacting legislation to regulate the design and deployment of AI.

This means using positive action to ensure that women and minority groups are fairly represented on AI design teams. Such action can take the form of quota systems that ensure that these teams are diverse.

Another key principle that countries have just agreed to is to ban mass surveillance and other invasive technologies that violate fundamental freedoms. Of course, we don’t expect CCTV to be completely abolished everywhere, but we do expect such mass surveillance to be limited to uses that are compliant with human rights. UNESCO will use peer pressure and other multilateral tools that UN agencies use to enforce global standards.

Over the coming months, UNESCO experts will work to create a set of monitoring tools to ensure AI development and deployment complies with human rights, but does not stifle innovation. This will be a difficult balance to achieve and will require the full commitment of the scientific community.

The new agreement is broad and ambitious. It tackles online bullying and hate speech and obliges countries to reduce their technology carbon footprint – the amount of energy used to store our data has increased significantly since AI innovation began to expand.

All players in the AI ​​world are aware that they cannot continue working without a rulebook.

UNESCO now expects two things to happen. First, governments and companies will voluntarily begin to make their AI systems conform to the principles set out in the recommendation — similar steps took place after the UNESCO Declaration on the Human Genome set standards for genetic research. Second, governments will legislate using the recommendation as a guide. UNESCO will monitor the progress of the legislation and countries will be required to report on their progress.

With this agreement, we believe we can deploy AI where it can have the most impact on the world’s biggest challenges: hunger, environmental crises, inequalities and pandemics. We are optimistic that we have built the momentum for real change.

Gabriela Ramos is UNESCO’s Assistant Director General of Social and Human Sciences

More on these topics:


Leave a Comment