The Ethical Concerns of AI

 Artificial Intelligence, more commonly written and pronounced as AI is defined as any piece of code or software which can simulate or emulate human intelligence. Nearly everybody is familiar with the concept of AI. After all, when you play Tic-Tac-Toe against your phone, your phone is running a dedicated AI module to make the seemingly intelligent moves. However, when discussing AI, it is usually considering “neural networks” - a series of virtual nodes which emulate the neurons in an animal’s brain, the interconnection of which allows intelligence.

 Recently, Artificial Intelligence has had an immense surge, with new software techniques and more powerful hardware allowing much more advanced AI to be developed. With this surge, and skyrocketing adoption in a variety of domains, application and use of AI has turned into a serious ethical problem, principally due to the impacts it may have on the population in coming years. As such, it may be time to properly evaluate the benefits and drawbacks of AI when such a powerful technology is deployed haphazardly in society.

Artificial Intelligence is generally associated with revolutionary, cutting edge discoveries. In fact, AI has pushed the limits of science and technology beyond what we previously imagined possible, ushering a new era based on the work of machines. It is extremely fitting that us humans, laden by our own limits, have sought to extend our gift of intelligence into a machine - to fuel electricity with logic and reasoning. And this is why one of the most dominant fields employing Artificial Intelligence on a regular basis is the scientific research domain.

  Rather recently, news came out that Google’s DeepMind deep learning AI had completed the sequencing of all possible protein combinations. All living things are made of proteins, it is one of the fundamental building blocks of animal life. Understanding all the different proteins which may exist means understanding the chemistry of life itself, and offers us a more profound outlook into the natural world. Knowledge on proteins can be applied in drug creation, in order to better simulate and understand all the potential side effects.

Along the same strain, Artificial Intelligence has been a blessing to the medical field, with highly specialised neural networks being in the works for a long period of time. Trained on huge amounts of medical knowledge and experience with real world patients, numerous specialised AI have been born to automate the medical field.

 Imagine, in the near future, there is no need for scheduling an appointment with a doctor. Instead, all you need to do is call an AI medical assistant over the internet and describe your condition. The AI will then be able to provide a thorough diagnosis. Or even when it comes to surgical operations, AI may very well overtake humans through their heightened precision and reasoning. Invasive operations may become safer as there is less reliance on humans - humans who can make mistakes. Furthermore, the logic of an AI is always straightforward - analytical and logical, instead of presumptuous and based on intuition.

Further uses of AI in the healthcare sector include genetic analysis in order to determine certain genetic traits which may make a person more susceptible to developing certain illnesses. Complications such as cancer, angina and diabetes may be influenced by the person’s DNA and through genetic sequencing, AI may determine exactly what they are at risk of. This opens up the path to preventive treatment or, in the future, genetic therapy to alter the nefarious genes.

 The AI also translates to the microscopic world, in particular bacteria and viruses. Analysis of the genetic code of pathogens is a crucial step in developing a cure for them. For example, the Covid-19 mRNA vaccines were developed using AI which was able to leverage computational methods and elucidate numerous biological pathways of the virus. Whatever the Earth can throw at us in terms of diseases, the AI is able to not only comprehend, but also provide a cure.

Artificial Intelligence is also generally associated with a higher standard of living, mainly due to the AI based facilities which we now possess. In fact, AI is everywhere around us, from being embedded in our phones to the algorithms which selects which cat pictures to show you on your Facebook timeline. Personal assistants, such as Siri and Alexa, are now more common than ever and synchronise seamlessly with smart home appliances to create a “smart home”. This has offered us heightened quality of life, since essentially anything can be done automatically through voice recognition and Artificial Intelligence. Furthermore, similar AI is used in applications such as Google Maps, in order to help people go from one place to another in the least time possible. It has even been integrated into social media to suggest new content based on posts and comments you have previously interacted with.

 AI is also able to effectively bridge the gap between practicality and public safety, by automating numerous aspects of our lives which are dangerous to us. The most prominent example of this is self-driving cars, which use machine vision and AI judgement to navigate on roads. Such technology, adopted on a large scale, promises to reduce vehicle-related accidents, thanks to the AI’s superior judgement and reflexes. Currently, Tesla’s and Volvo’s self driving vehicles have performed admirably on the road, with superhuman reaction times and pinpoint accurate judgement. Volvo’s trucks come with automatic breaking capabilities, which can apply the brakes before the human can even realise that they need to be pressed.

However, AI may be extremely dangerous if misused. For example, there is a very large push towards automation and robotics on the battlefield, with this technology being accelerated by the Ukro-Russian war. This includes kamikaze drones and unmanned aerial vehicles (UAVs) and it is not the first time that AI is being given autonomy on a battlefield. During the prolonged war in Afghanistan after the 9/11 attack, the US deployed UAVs to pinpoint and eliminate terrorist hideouts in the Afghan mountains. Recently, they installed AI pilots in fighter aircraft which were then able to out-maneuver the best of the human air force.

 The push for AI in the war zone is driven by how “perfect” machine can be. Unaffected by the flaws of the human body and mind, AI have the potential to revolutionise the battlefield. They are not affected by stress or fatigue and high processing speeds allow for AI to always take the optimal decision, be it at sea or miles in the sky. AI is simply faster than us, better than us and stronger than us. Furthermore, AI are able to be copied, meaning that there will never be a shortage of combatants, regardless of how many instances may be destroyed in battle.

 Unfortunately, it is the lack of humanity which makes AI so ethically flawed. AI is unemotional, which can lead to bad judgement, especially if it comes to weighing human lives against each other. Furthermore, errors in the AI’s code may lead to mistakes. A civilian car being misidentified as a hostile vehicle by the AI will lead to it firing at the car, potentially causing mass civilian casualties. This was the case back during the Afghan war, where an AI controlled drone fired at a building it had misidentified, killing a dozen civilians, of which there were 7 children. When you are putting human lives at the hand of a machine, it is of utmost importance that the AI is nearly always correct. There is no room for error.

Yet another unethical application of Artificial Intelligence is how it is taking over jobs, being more accurate than their human counterparts. It is estimated that, in the near future, some 75% of present day jobs will be semi or fully automated, meaning that employment opportunities will plummet. And for good reason too. Machines and AI are stronger, more efficient and more accurate than us measly humans. Furthermore, they do not need sick leave, can work over weekends and public holidays, do not need to sleep and do not ask for a wage. This is a direct result of capitalism, whereby the global elite seek to deepen their wealth, disregarding entirely the working class. AI would provide the means for exponential wealth growth, coupled with limitless scalability of operations.

 Currently, capitalism is supposed to have a trickle down system, whereby wealth accumulated by the rich is redistributed among the working class through wages and taxes go towards infrastructure which benefits everyone. However, with the introduction of an AI driven workforce, there will be no more wages going to the “working” class. This means that the average person now has absolutely no means of income as AI will have an edge in every domain. All the wealth created by automation will go straight to the pockets of whoever owns the company, not to the working class. This is shown in Silicon Valley’s largest companies, such as Google, Apple and Facebook. Google’s revenue has jumped from 37.9 to 256.74 billion US dollars in the span of 10 years, with no sign of slowing down. And yet, the employee headcount only increased by 5x, due to how automation and AI is pushing for unlimited business expansion.

Furthermore, AI has been seen to demonstrate biases and prejudices identical to those in the real world. This includes, but is not limited to, racism, homophobia and sexism. This is because AI needs to be trained with real world data. It is like a child. If you expose a child to a host of racist content, the child will perceive racism as normal and will themselves be racist. The same concept applies to AI. A lot of the time, the people developing AI do not care much about carefully selecting data for the AI to learn from and instead give it unfiltered access to the internet. Biases in this information induce similar biases in the reasoning and logical thought of the AI.

 A striking example of this is Google’s own translator AI, known as Google Translate. It is trained on sentences written in two languages and from there, determining how words are translated, as well as the context. What happened is that gender neutral pronouns, such as “they” in English, were translated to sexist gender counterparts. Hence, “they are in the kitchen” would translate to the French equivalent of “she is in the kitchen”, a sexist stereotype. This mis-characterisation and bad judgement, born from incorrect teaching, leads to unequal treatment and classification, mirroring real world prejudices. Furthermore, these biases in treatment may arise when it comes to minorities, because there is insufficient data to properly train the AI. For example, AI may be more likely to classify black people as criminals, further exacerbating the already racist judiciary system in many countries.

Another serious concern brought about by the rise of Artificial Intelligence is the risk posed to privacy, in particular on the internet. Artificial Intelligence has brought with it immense processing power. AI can classify millions of data pieces in an hour and it is rather disappointing to see that this raw speed and power is being misused for profit. With the advent of the internet, people have become more “open” to the world. Present day, everyone on the web leaves a “digital footprint” - a trail of data showing their internet activity. This may be stored through internet trackers or “cookies”. Google itself runs some of the largest data collection programs on the internet, called Google Analytics. It tracks all the user’s activity within their proprietary browser, Google Chrome, and may also gather location data. Google then feeds this mound of personal data into their Artificial Intelligence algorithms which is then able to create a digital profile based on the user, detailing internet habits, hobbies, interests, age, profession and so much more.

It is obvious that such profiling is detrimental to privacy and security on the internet and yet unrestricted use of giant artificial intelligence networks allows such immoral and unethical practices to take place. However, it does not stop at the internet.

AI is frequently used in mass surveillance systems to construct a database of real-world activity. China is the first to have implemented such a system and uses it to monitor the population, to deter any protests against the CCP dictatorship. AI powered cameras were widespread in Hong Kong during the riots against Chinese occupation, and they helped arrest a large number of protestors. Sophisticated facial recognition systems, working in tandem with a network of high resolution security cameras, allows the government to track the whereabouts of distinct people. And while this has helped make China one of the most “crime-free” countries on the planet, the degree of morality of such a blanket use of artificial intelligence for surveillance comes into question.

Artificial Intelligence is here to stay, whether we like it or not and for better or worse. From the aforementioned points, it is obvious that AI is not inherently flawed and it is only the humans in charge who apply the technology with malicious intent. In the computer science field, AI is referred to as a tool. And as with any tool, it can be used to drive humanity forwards or it can continue to be used as yet another weapon. However, this should not detract from the fact that even nowadays, Artificial Intelligence is a stunning technology which has brought about immense improvements to our lives.