We are already in the age of algorithms. Some of the major decisions in our day to day lives – from getting a personal loan or not, how much we pay for health insurance – are being decided by mathematical models rather than humans. Sounds perfect! The system is fair, denies privilege to a certain set of people, applies the same rules for everyone and thus is unbiased.
Welcome to the other side of the coin. Cathy O’Neil, the American mathematician, data scientist and author, reveals in her book titled ‘Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy’ that the models being used today are opaque, unregulated and uncontestable, even when they’re wrong. Models are propping up the lucky and punishing the downtrodden, creating a “toxic cocktail for democracy.”
Take, for instance; the model used to define the risks associated with a person applying for a loan – what if a poor student applying for the same is marked as “too risky,” it’s nothing less than reinforcing discrimination. This is where the concept of Human-Centred AI (HCAI) is gaining prominence. In this article, we will delve into what exactly is HCAI? Why do we need such AI systems?
Need for human-centric systems
HCAI is an emerging field dedicated to developing AI systems that complement rather than replace human abilities. HCAI aims to maintain human control in a way that guarantees artificial intelligence fits our requirements while also being transparent, equitable, and respectful of privacy.
Despite the increasing levels of automation offered by AI — whether it’s AI driving our vehicles, deciding the kind of content and information we see, finding new drug molecules, or even deciding the best stocks for investment – the human aspect remains a constant and critical thread throughout these systems. The long-term success of AI is predicated on our acceptance that people play a crucial role in its design, operation, and application.
Machine learning systems have failed in unanticipated ways, making them unreliable in important applications. A bad product or content recommendation, for example, could have a minimal influence. Consider the recent case where words like “beach” become “bitch”, “buster” turned into “bastard”, or “combo” morphed into “condom” when the automatic speech recognition system displayed such inappropriate transcriptions in a video meant for children on YouTube. But a rejected loan or job application, on the other hand, can be far more detrimental, and an inappropriate diagnosis of a medical condition can be fatal.
Benefits the AI ecosystem as a whole
Enhances reliability and scalability: The human-centric approach paves the way for the technology to handle heavy computational needs while relying on cognitive and emotional input from humans. This enables the scaling up of processes and information without jeopardising data integrity and giving due respect to privacy.
Better decision making: The goal of human-centered AI is to augment human skills rather than replace humans. Human-centered AI enables businesses, to be precise, humans in the business to make more informed decisions and build trustworthy models and solutions by combining the precision of machine learning with valuable input and human values.
Towards Artificial General Intelligence (AGI): Human-centered AI extends the currently limited boundaries of artificial intelligence solutions to bridge the gap between machines and human beings. Developing machine intelligence with the purpose of understanding human language, emotion, and behaviour can lead us toward AGI.
This broader perspective to incorporate HCAI can help shape technology’s future to better meet human needs. Educators, AI researchers, designers, coders, software engineers, product managers, and even government agency employees can all contribute to the development of breakthrough technologies that improve people’s lives.
Source: indiaai.gov.in
