The White House this morning revealed what it popularly refers to as an “AI Bill of Rights,” which aims to establish principles around the ways in which AI algorithms should be deployed, as well as guardrails for their applications. In five bullet points prepared with feedback from the public, companies such as Microsoft and Palantir and human rights and AI ethics groups, the document outlines security, transparency and privacy principles that the Office of Science & Technology Policy (OSTP) – which has the AI drafted Bill of Rights – states that this will lead to better results while reducing harmful effects in real life.
The AI Bill of Rights mandates that AI systems must be proven safe and effective through testing and stakeholder consultation, in addition to continuous monitoring of the systems in production. It explicitly calls for algorithmic discrimination and says AI systems should be designed to protect both communities and individuals from biased decision-making. And it strongly suggests that users should be able to opt out of interacting with an AI system if they wish, for example in the event of a system failure.
In addition, the proposed White House blueprint states that users should have control over how their data is used — be it in decision-making or development of an AI system — and be informed in plain language about when an automated system is in plain sight. language is used.
According to the OSTP, recent history is filled with examples of algorithms that have gone haywire. models used in hospitals to inform patients, treatments were later found to be discriminatory, while hiring aids designed to exclude candidates for jobs has been found to reject mostly female applicants in favor of men – because of the data the systems are trained on. However, like axios and wired note in their coverage of today’s press that the White House is late to the party; a growing number of bodies have already weighed in on AI regulation, including the EU and even the Vatican.
It is also completely voluntary. While the White House tries to lead by example and bring federal agencies into line with their own actions and derived policies, private companies are not bound by the AI Bill of Rights.
In addition to publishing the AI Bill of Rights, the White House has announced that certain agencies, including the Department of Health and Human Services and the Department of Education, will publish guidelines in the coming months to curtail the use of harmful or dangerous algorithmic technologies. . in specific settings. But these steps do not comply with, for example, EU regulations under development, which ban and curtail certain categories of AI that are believed to be harmful.
Still, experts like Oren Etzioni, co-founder of the Allen Institute for AI, believe the White House guidelines will have some impact. “If done right, [a] bill could reduce AI misuse while still supporting beneficial use of AI in medicine, driving, business productivity and more,” he said told The Wall Street Journal.