Impact Assessments: Supporting AI Accountability

Impact Assessments: Supporting AI Accountability

Artificial intelligence’s applications span everything from healthcare and education to manufacturing and energy.

Automation is influencing strategies across industry, but the rapid pace of deployment raises as many issues as these technologies claim to solve.

Law enforcement, mortgage lending, and video hiring processes are just some of the areas where the limits of current systems have been exposed, with significant consequences. Tools like deepfake software threaten to accelerate the spread of disinformation even further, inflaming geopolitical tensions and jeopardising national security.

The world stands at a regulatory crossroads. For all that algorithms have promised to augment our decision-making, judgements on how to mitigate the negative impacts of AI while preserving its value rest in human hands alone. Where do we go from here?

Access Partnership’s latest report, ‘Impact Assessments: Supporting AI Accountability & Trust’, helps to guide this debate by providing an overview of existing processes, as well as recommendations on how to proceed. Developed in partnership with Workday, the report outlines three main accountability frameworks – algorithmic impact assessments (AIAs), third-party auditing, and conformity assessments – to explain why AIAs should become the global standard.

To date, a patchwork of regulatory responses has emerged. New York and California have proposed mandatory third-party auditing of AI systems. Companies in Virginia, Colorado, and Connecticut must conduct ‘data protection assessments’ for activities that demonstrate heightened consumer risk. The EU’s landmark Artificial Intelligence Act will require conformity assessments for ‘high-risk’ systems.

This report analyses existing and emerging frameworks in detail, drawing on our industry-leading expertise to identify the limits of each in terms of technical standards and adaptability.

Policymakers hold the key to harmonising AI’s future. The challenge of balancing transparency with the need to drive automated technologies’ advantages for economic development is essential to societal progress. To learn how to do both, download our report.

Download report 

Related Articles

Access Alert: Peru’s congress introduces bill to regulate AI

Access Alert: Peru’s congress introduces bill to regulate AI

On 12 February, lawmakers from the Podemos Peru party set forth BoL 07033/2023 to establish a legal framework for AI...

22 Feb 2024 Opinion
Connecting Fiji: The opportunity for growth and prosperity from subsea internet investment

Connecting Fiji: The opportunity for growth and prosperity from subsea internet investment

The widespread adoption of digital technologies, such as Artificial Intelligence, will require investments in digital infrastructure to ensure connectivity and...

7 Feb 2024 Reports
Access Alert: Yes, there is an AI Standard, and its number is ISO/IEC 42001

Access Alert: Yes, there is an AI Standard, and its number is ISO/IEC 42001

A standard for managing AI in organisations was released in December 2023. This standard, ISO/IEC 42001 Artificial Intelligence – Management...

7 Feb 2024 Latest AI Thought Leadership
Access Alert: UK’s “Pro-Innovation Approach” to AI – Outlier or Opportunity?

Access Alert: UK’s “Pro-Innovation Approach” to AI – Outlier or Opportunity?

The UK government has unveiled its response to the AI White Paper consultation, through which the Department for Science, Innovation...

6 Feb 2024 Latest AI Thought Leadership