ethics in ai

As computing technology continues to evolve, so too does its use in our everyday lives. For years, computer scientists have dreamed of greeting an Artificial General Intelligence (AGI), a programme that is all-knowing and capable of answering any question. In the 1950s, it was exceedingly expensive to operate computers; however, recent breakthroughs in technology have led to the creation of hardware with high computational power that is cheap and accessible to the average home consumer. While AGI is yet to become a reality, recent developments such as OpenAI’s Large Language Model (LLM) Generative Pre-trained Transformer (GPT) show that what was previously thought to be constrained to the realms of science fiction is now an increasingly unavoidable part of our modern lives.

Questions must be raised about ethics in the field of Artificial Intelligence (AI). LLMs are a type of neural network inspired by the inner workings of the human brain. They learn and evolve by being given new information. One might think that with around 75 years of research, scientists would understand how their systems work; however, LLMs are considered to be a black box in that, though we know what goes into the programme, we do not understand how the system processes it to create an output. This results in the system needing to be “trained” by humans, a process that is tiresome and is often outsourced to underprivileged countries where labour is considerably cheaper.

One thing that AI excels at is Natural Language Processing (NLP). With the introduction of the transformer in 2017, AI has been able to learn entire languages. The Universities of Glasgow and Edinburgh have recently been awarded grants from the Scottish Government to fund their ongoing Gaelic language conservation efforts using AI. As AI requires a great pool of data to learn from, researchers have been given the task of collating a large body of Gaelic language text to save one of Scotland’s indigenous languages from digital extinction. The goal of the universities is to create a Gaelic subtitling system for the BBC, which would be difficult to implement manually due to a lack of resources and manpower. The researchers are also working on a speech recognition tool for Ojibwe, an indigenous language of Canada.

While it is clear that there are positive use cases for AI, a particular use case has found itself the topic of frequent discussion, media attention, and derision in the creative sector: generative image models. Generative image models take libraries of images and use them to create new images, similar to how LLMs do this with text. This has been the subject of much discourse both online and in the professional world as already creatives have felt the real-world effects: companies will go above and beyond to avoid paying artists for their hard work. In the past year, the Writer’s Guild of America (WGA) and the Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) famously went on strike to protest against, amongst other things, Hollywood studios’ plans to replace writers with LLMs.

It is clear that AI has revolutionised the technology and has evolved past a point where it can be ignored. OpenAI’s ChatGPT has already demonstrated it is highly competent at writing code in various languages, making it an ideal assistant and learning tool for traditional developers. It is the writer’s view that as AI improves at writing code, developers will be able to automate more and more of their processes with minimal oversight, allowing them to spend their time on more important things, such as implementing new features and improving cybersecurity. In a world dominated by late-stage capitalism, AI is the new fad, just like cryptocurrency and even the dotcom bubble before it. As companies report record profits, they fire their workers in mass layoffs while claiming to be too poor to afford to pay them fairly. It is plain to see why AI, which does not demand a wage, is so popular with the corporate crowd. In an ideal world, AI would be used to automate busywork in order to promote a healthier work-life balance, however in practice it has been shown time and time again that companies would far rather remove the worker from the work with no regard for the harm it can cause to the individual and to the wider economy.

Despite this idealist corporate philosophy of automating tedious behaviour and removing the worker from the work, it has already been established that AI requires a shadow workforce of low-paid data input workers. Again the ethics of this should be considered: could corporations be willfully destroying local and global economies in order to line the pockets of their shareholders?

Though AI has clear shortcomings when placed in the hands of for-profit organisations, its use in medicine may just be some of the most important technological advances in the history of mankind. AI has been used to quickly and accurately predict the sequencing of genomes, a task that may take a dedicated group of PhDs as long as three months. Fascinatingly, advancements in the field of stem cell research have allowed scientists to combine the power of AI with synthetic organoids- lab grown human brain tissue- with the goal of creating more human-like AI. Many have voiced concerns about the ethics of creating new lifeforms with the sole purpose of being used as learning machines, though reports claim that while they are aware, they are not sentient.

While the moral use of synthetic brain organoids in general and in relation to AI has been debated, it is hard not to see the potential benefits of brain-computer-interfaces (BCI). A woman diagnosed with severe amyotrophic lateral sclerosis was given a brain implant that used AI technology to detect her brain signals, allowing her to speak after almost complete paralysis. The future of this technology, while it deserves close regulation to ensure it is doesit does not become dominated by a monopoly, looks promising.

With all the severe ethical dilemmas involving AI, it is understandable that many view it with a pessimistic lens: “AI is going to take over the world!” However, AI is a market trend that comes and goes, a cycle so consistent that there’s a name for periods where the market loses interest: the “AI winter”. With time, the public will accept the current innovations in AI as just another part of modern life, and AI tech startups will quickly go out of business. Businesses such as Google, OpenAI and others all benefit from sowing seeds of fear and paranoia in the general public as it keeps them relevant and profitable.

In what several people have dubbed the “AI-pocalypse”, it is this writer’s opinion that it has been proven that people are far more terrifying than Artificial Intelligence could ever hope to be- specifically corporations with more money than sense. Perhaps with strict regulation in regard to ethics and antitrust law, AI can be used for the betterment of society as a whole. As it stands, all the world can do is watch its development and have hope that corporate corruption will not tarnish what may be humanity’s greatest new invention.