“Technology has already had a huge impact on the way we create and consume art. Technology can be both a threat and an opportunity for artists creating art,” Ai-Da, a robot artist powered by Artificial Intelligence (AI), told the House of Lords committee.

There are several jobs that are being done by AI now, whereas they used to be done by people. Writing essays, diagnosing diseases, and operating cars were largely manual tasks now being automated with AI. However, the crux of the problem is this: why is AI replicating human prejudices? Researchers are trying to answer this question. These AI systems have problems with their functions being biased, mirroring gender, race, and sexuality biases that exist in the world now. For example, there was an issue with Apple’s Face ID unlocking the phone for two different asian women. Additionally, black people are proportionally misidentified by facial recognition technology too. Because of this, as well as the fact that researchers frankly do not know how to track this problem, experts have warned developers to slow down and examine how AI produces certain outputs.

Most AI systems are black box models. A black box refers to a model whose behaviour can only be observes by examining inputs and outputs– what goes in and what comes out of the system are the only things that can be seen. A black box model’s objective is to iterate and evolve until it reaches a final product we are happy with. However, because of the black box model, we cannot keep track of how an AI has been influence and, more importantly, who is responsible for any harm caused.

AI is the capability of a computer or a robot to perform tasks that are normally performed by intelligent beings. Scientists have worked for decades to create the perfect AI. However, it raises the question: what is the ideal AI? Can a flawless AI be created; one whose intelligence humans are unable to comprehend?

The complex capabilities of AI do not end there. In the medical profession, scientists from MIT (Massachusetts Institute of Technology) have discovered that AI can accurately identify a patient’s self-reported race using only diagnostic pictures like CT scans and x-rays. This is something humans are unable to do, and scientists are baffled as to why AI can do it but doctors can’t. For example, looking at the bone density showed that different races have different bone thickness, while numerous earlier studies have demonstrated that racial differences are not strongly predicted by sickness or physical make-up.

All dystopian-like thoughts aside, it shouldn’t slip our minds that AI continues to be a key figure in the development of technology. An important function of AI is that it enhances security and surveillance. AI can detect potential threats and come to the rescue when danger strikes, and can be used in unison with voice and facial recognition for biometric authentication. Smart systems give operators fast information on visitor behaviour anomalies and suspicious activity, as well as the identification of undesirable visitors.

Scientists are advocating for AI accountability. Researchers are developing an emerging term called "explainable AI" to address AI's limitations. These models are currently less potent than black box ones, which may affect the true desirable output of AI.

An alternative to this is AI that considers human feedback and uses models that are particularly designed to be bias-aware. Without society noticing, AI is still being powered by the workers of the world, those who are experts in their field.

AI specialists and academics are cautioning developers to take a step back and concentrate more on how and why a system gets particular outcomes rather than the fact that the system can accurately and quickly produce them as AI systems become more complicated and humans are less able to understand them. It's unlikely that we'll have ideal systems anytime soon, but trying to find the answers to our creations while simultaneously flourishing in the development process is a crucial first step toward creating AI systems that are more transparent and properly reflect the ambitions of humankind.

Sources

The most popular AI currently is ChatGPT (Chat Generatively Pre-Trained Transformer). It is a chatbot that could have a realistic human conversation, but it can also do much more than that: it can write essays, answer math questions, and even code. However, in an academic setting, using ChatGPT has struck a chord of concern amongst educators. Teachers are concerned about students using this free platform and using it as a tool to complete homework and write assignments, which endangers students’ abilities to write and research. There is also concern that ChatGPT and similar kinds of AI may also displace workers in ways that are mostly hidden from view; these robots don't need to be paid, and they don't get tired, which is the complete opposite of what people need.

The Artificial Intelligence Growth Spiral

Sources

InSight Scoop