As technology is becoming more widespread and people are using it at increasing speed to power all types of processes, lawmakers have increasingly observed the need for legislation to ensure that the safety of everyone using the systems is well-guarded. In 2020, the United States, documents were released pertaining to the guidance related to regulating AI. While that approach had many positive aspects, including the reasonable scope and the understanding that regulations must exist independently from the development of new systems, some aspects were also lacking.

The first and most important was the omission of the potential harms of AI, and an overfocus on the positive aspects. The documents stipulated that the United States was determined to invest in the further development of artificial intelligence technology. Many analysts, however, have felt that the paper was overly focused on the economic aspect, as the guidance was mainly focused on the fact deployment shouldn't be hampered due to regulations.

However, in 2022 the situation has changed. The Bill proposed during the last few months of 2022 has remedied the shortcoming of the previous plan. But what exactly is the AI Bill of Rights, and what can you expect from it? According to ExpressVPN, it aims to protect the public from the harm that automated systems can produce through their various algorithms.

What is the AI Bill of Rights? 

The Bill is, first and foremost, a description of AI's possible risks and harms. While technology has been steadily developing over the past few years and will undoubtedly continue to do so, it's essential to determine, assess and seek to remedy its possible risks. Many concerns directly relate to expanding artificial intelligence solutions in education, financial services, healthcare, surveillance, hiring, etc. However, it isn't meant to offer universal truths but guidelines that can be used when discussing AI infrastructure and how it relates to those interacting.

While AI has developed significantly over the past few years, it is still far from achieving its full potential, which means there's plenty of room for improvement in the future. As technology develops, there are bound to be many modifications that consider these changes and find ways to mediate their possible weak points. The Bill is similar to the Artificial Intelligence Act proposed in April 2021 by the European Commission, aiming to introduce a regulatory framework for the development of AI.

AI bias 

One of the main reasons why regulations must be implemented in AI is due to the presence of Ai bias. It refers to human prejudices that have found their way into the programmed system. While some can be relatively harmless, others are downright problematic and can signal genuine problems that can impact the well-functioning businesses and organizations that use them. In some cases, AI bias can even be discriminatory in nature. For instance:

• An AI system used to hire prospective employees can have a misogynistic or racist bias that makes it overlook the resumes of women or people of colour and choose those of men or white people, despite equal or even better qualifications.

• Despite being high earners, people can get turned down repeatedly based on their ethnicity regarding mortgage approval.

• People of color can get labelled as high-risk offenders despite committing a relatively minor offense such as shoplifting, while a white offender might get a more lenient pass for a similar or more serious crime simply due to their ethnicity.

There are many reasons why AI bias occurs. The creator bias is arguably the most common. Software and algorithms are designed by humans, who sometimes subconsciously insert their preconceptions into the system, which the AI perpetuates. Data-driven biases appear due to the AI observing patterns in large amounts of data. Therefore, artificial intelligence will adopt prejudice as its own if prejudice exists. Latent bias can occur due to the system correlating different terms with stereotypes.

What does the Bill hope to achieve? 

The AI Bill of Rights blueprint contains five basic principles that come together with a set of technical companions that can aid in the responsible implementation of the system. The principles include:

• Safety and efficiency: Safe and effective systems are the ones that are protected against inadequate practices.

• Data Privacy: Over the past few years, the issue of data privacy has become increasingly important. Users must be protected via built-in tools and have complete agency over how their private information is used or by whom.

• Notice and explanation: This refers to the fact that users must be fully informed that an automated system is used to perform a specific process. They should also be aware of how this can affect the final results.

• Alternatives: You should have the option to opt out of using an AI-powered system and request the help of a person that can remedy any issues you're encountering.

The primary intent of the blueprint is to help develop and use the systems from now on in a way that won't cause any infringement on the rights, security and well-being of the general public.

Moving forward 

AI technology is here to stay, which means that the ways it operates into the future must keep in line with all the new developments. It's very likely that the regulations won't come without any pushback from the private sector, particularly tech companies who will push for self-regulation to not lose their investors.

As such, everyday users will only benefit from the Bill if companies and enterprises are willing to do their part and put ethical considerations above profit and higher revenue. Without this commitment from institutions, it's unlikely that the general public will be protected. Regulatory agencies will have an additional task in enforcing measures against contraventions. They will also have to adopt complementary legislation that helps increase responsibility and accountability in the business sector.