Oct 05, 2022 12:30:00

White House announces `` AI Bill of Rights '' that protects citizens from AI harm and discrimination and allows opt-out as necessary

In recent years, various changes have occurred due to the advancement of AI technology, but it has been pointed out that the

accuracy of the facial recognition technology used by government agencies is low , and the illustration generation AI has been criticized for infringing on the rights of artists. There is also friction between humans and AI. Under such circumstances, the White House Office of Science and Technology Policy (OSTP) announced on October 4, 2022 local time, a draft of the `` AI Bill of Rights '' to protect the American people from the harm and discrimination caused by AI.

Blueprint for an AI Bill of Rights - The White House
https://www.whitehouse.gov/ostp/ai-bill-of-rights/

Blueprint for an AI Bill of Rights: A Vision for Protecting Our Civil Rights in the Algorithmic Age - The White House
https://www.whitehouse.gov/ostp/news-updates/2022/10/04/blueprint-for-an-ai-bill-of-rightsa-vision-for-protecting-our-civil-rights-in- the-algorithmic-age/

The White House moves to hold artificial intelligence accountable with AI Bill of Rights | VentureBeat
https://venturebeat.com/ai/the-white-house-moves-to-hold-artificial-intelligence-accountable-with-ai-bill-of-rights/

White House guidelines for AI aim to mitigate harm | Reuters
https://www.reuters.com/article/tech-ai-biden/white-house-guidelines-for-ai-aim-to-mitigate-harm-idUSL1N31426R

White House unveils artificial intelligence 'Bill of Rights' | AP News
https://apnews.com/article/technology-business-artificial-intelligence-7a39848340d210592aeea2478225f489

In its draft AI Bill of Rights, the OSTP acknowledges that AI-powered automated systems have the potential to bring enormous benefits to agriculture, weather forecasting, medical diagnostics, and more, and to improve the lives of all. I'm here.

On the other hand, AI causes harm to citizens, such as undesirable pre-existing discrimination in AI used for employee recruitment and personnel evaluation, and racist prejudice remaining in systems used for medical diagnosis. OSTP points out that there is also. 'Too often these tools limit people's opportunities and hinder access to critical resources. These issues are well documented.' It threatens the rights of millions of people and harms people in historically marginalized communities.'

Difficult to remove racial bias in algorithms built into medical systems - GIGAZINE

Therefore, the OSTP said, ``Important advances made by AI must not come at the expense of civil rights, democratic values, or the fundamental American principles that President Joe Biden affirmed as the foundation of his administration.'' It sets out 'Five Principles' to guide the design, use and deployment of automated systems to protect the public. The five principles, which the OSTP established in a year-long consultation with more than 10 departments, combined with feedback from civil society groups, engineers, industry researchers, and high-tech companies such as Palantir and Microsoft, are:

・Safe and effective system
People need to be protected from insecure or ineffective systems, and automated systems must be developed in collaboration with various stakeholders and experts to identify risks and potential impacts. it won't work. Automated systems used by companies and organizations are required to undergo independent evaluation and reporting, with the results made public whenever possible.

・Protection from algorithmic discrimination
Citizens should be protected from discrimination and mistreatment based on race, color, ethnicity, gender, religion, age, national origin, background, or genetic information by algorithms, and automated systems should be designed and implemented in an equitable manner. must be used. Designers, developers, and users of automated systems need to take proactive and continuous steps to protect people from algorithmic discrimination.

・Data privacy
Users should be protected from data breaches by safeguards built into automated systems, and data should only be collected when absolutely necessary. In addition, users of automated systems must clearly communicate information about data collection to users.

・Notice and explanation
Users have the right to know that automated systems are being used and to understand the impact on them. Users of automated systems are required to inform users of the use of AI, its impact, and explain why they are using AI.

・Human alternatives, consideration, withdrawal
Users have the right to opt out of automated systems and choose human alternatives if necessary. If an automated system fails, or if a user disputes the consequences, there should be prompt access to someone who can review and resolve the issue.

Dr. Alondra Nelson, Assistant to the President and Deputy Director of Science and Society for OSTP, told The Associated Press: 'The Biden-Harris administration is committed to working not only intergovernmentally but across all sectors to promote fairness. 'We need to create, use and control technology,' he said. 'We should expect and demand better from technology.'

The draft AI bill of rights announced this time does not mention specific AI that is actually controversial, such as AI that uses biometric data and facial recognition, and autonomous weapons and smart cities using AI. It has also been pointed out that it does not address the issue of In addition, the AI Bill of Rights is not binding on high-tech companies that develop automated systems and does not include new legislative proposals, so law enforcement agencies will continue to deal with automated systems according to existing rules.

The OSTP claims that the draft of the AI Bill of Rights is a guideline for the effective functioning of automated systems in society. 'We look forward to blazing this trail with the American people,' he said.