Celebrating International Women’s Day (IWD) proffered HewardMills with a renewed focus to disintegrate bias and amplify the voices of women. In a time where Artificial Intelligence (AI) is rapidly advancing, there are growing concerns to address this powerful tool. Although AI is incredibly effective, it may potentially cause intrinsic harm to an individuals’ privacy if not properly controlled. To curb bias discrimination, policy development is integral to regulate use, increase benefits and mitigate risks.
As a global B Corp Data Protection Officer (DPO), we support clients to demonstrate compliance and continuously maintain safeguarding systems to shield personal data. We assist organisations to implement security safeguards such as ISO Standards, Privacy by Design and Cyber Security measures. In line with our aim to disintegrate bias and amplify the voices of women, now more than ever, it is critical to address data discrimination in the use of AI technologies and most importantly, balancing the benefits and risks of AI intelligence.
In 1955, John McCarthy, famously coined the term Artificial Intelligence and defined it as “…the science of and engineering of making intelligent machines.”
Since discovery in 1955, AI has demonstrably evolved to be a prominent branch of computer science involving algorithms and machine learning systems with capability to automate personal data processing and resulting decision-making. The Council of Europe’s Convention 108 1981, the first legally binding international data protection legislation, required organisations using personal data in computerised form to have a social responsibility to secure the data of individuals affected by decisions made about them based on their data.
The main premise of the UK Information Commissioners Office (ICOs) 3-year ICO25 strategic plan is to empower individuals of their rights in sharing their data with organisations for products and services and empower organisations to responsibly use data to innovate and navigate requirements to demonstrate fairness, transparency and accountability. As a result, this move by the ICO is fostering important discussions relating to addressing AI related discrimination and biased negative outcomes.
In her presentation, ‘Responsible AI: Principles and Practical Applications’, Tsu-Jae Liu, Dean and Professor at UC Berkeley explained, “While AI can improve efficiency and productivity, it can also undermine efficiency, effectiveness, and equity.” The sharp rise of ChatGPT instigated a sense of urgency to analyse the ethics of AI and the need to swiftly mitigate the potential bias intrinsic in training data.
There are two types of AI: general2 and narrow. General AI has capability to manage generalised human-like tasks, e.g., ChatGPT, and AI chatbot. Narrow AI, the most used form of AI, is essentially a machine learning system limited to performing very simple tasks such as question and answer engagements and basic calculations, e.g., Apple’s Siri or Google’s voice assistant, Alexa.
Machine learning systems use algorithms and analyse patterns of data that reflect behavioural patterns of individuals to develop recommendations on decisions, e.g., Netflix, YouTube, and Instagram platforms offers personalised recommendations of content.
In balancing the benefits and risks of AI, it is critical for organisations to analyse potentially discriminatory information gathering patterns. Due to human error, data sets used in algorithms and machine learning systems may be inaccurate. Since machine learning systems are a result of using patterns of data sets, there is bound to be ingrained biases that will naturally produce biased results that further amplify discrimination toward a particular group of people. Where organisations rely on such inaccuracies for decision-making to offer services to consumers, there is bound to be discrimination and a sense of automated biases that must be avoided.
As a DPO, we assist organisations to maintain compliance to global data protection and privacy regulations. Most notably the advent of the General Data Protection Regulation (GDPR) 2018, provided stricter rules on accountability and transparency requirements in the use of personal data. Therefore, we promote businesses to ensure that their systems demonstrate fairness, transparency, and accountability in processing personal data.