AI ethics 101: the good, the bad, and the biased

AI has been the center of the social and political debate in the last months, and has inevitably become an important part of our lives. Like all new technologies, from new softwares to self-driving cars, AI has the potential to bring a tremendous amount of benefits to society. In this article, I’d like to offer an analysis of the “good, the bad, and the biased” of AI, as well as offering an historical perspective on technological advancement and censorship.

Being suspicious about new technologies and inventions is nothing new, change is difficult to accept and there has always been a certain level of hostility regarding the new “trends”.

In the early days of the printing press, rulers and European clergymen were worried that the easy dissemination of information and “revolutionary ideas” would lead to the erosion of the traditional orthodox Christian values and spread dangerous ideas around the continent.

The newly invented print suffered a similar fate in the Ottoman Empire as it did in Europe, officials and religious leaders ( do you see a trend?) feared new ideas would disrupt social norms and push people to challenge the authorities. This fear led, in 1727, the Ottoman Sultan Ahmed III to issue a decree banning the use of the printing press in the whole empire.

How did this fear of technology hinder, for example, the aforementioned Ottoman Empire? The ban prevented books and printed materials from being circulated, making it difficult for scholars to keep up with the rest of the world. Disseminating information and communication became harder, even for the government itself, as information had to be transmitted orally or through handwritten documents. In short, a nightmare. Lastly, the printing press ban also hindered economic growth, as printing was a rapidly growing industry in Europe and elsewhere, and the Ottoman Empire was unable to participate in this new and lucrative field. One of the reasons for supporting the printing press ban was the argument that it would leave the scribes without a job, as they were the ones responsible for handwriting books and documents. (again, sounds familiar?)

The ban was then gradually relaxed, leading to a flowering of Ottoman literature. Ok, that was a lot right, why the historical essay? My goal was to highlight how history has been repeating inside the same cycles for hundreds of years. The printing press, the telephone, the fridge, the internet, planes…the list goes on. And time and time again, history keeps showing up how many of these concerts were often unfounded, and that new technologies, when used right, can bring significant changes and benefits to society.

The good

One of the many benefits of AI systems is the ability to automatically perform repetitive and mundane tasks. From industrial power plants to finance, to healthcare, AI-powered robots can be used to assemble and run almost every kind of machinery or computer. Giving humans a break, and allowing people to focus on more complex and creative tasks rather than wasting time pulling a lever for 8 hours a day. Using chatbots and text-generating AIs can be useful when dealing with customers and easy “machine to human” communication, saving time and money for shops and companies.

Machines can also be taught to make decisions, or at least used to improve and make decisions that could be considered more reliable. In finance, AI algorithms can be used to go through large amounts of data and to identify patterns that a person might miss.

The bad

One of the main concerns around AI is the potential for it to automate jobs and displace human workers, as we have seen in the Ottoman Empire – automation has been displacing jobs for centuries – yet AI has the potential to automate a much wider range of tasks than previous technologies. If only the scribes had to apply for unemployment, at the rate we’re going, AI’s effect on sectors like writing, graphic design, coding and copywriting might ultimately end up displacing thousands of skilled workers.

AI software has the potential of being used for malicious purposes too, such as cyberattacks, propaganda campaigns, and the spreading of misinformation and fake news. The AI-generated picture of the pope wearing a tight Balenciaga jacket fooled all of us last week, and if we can’t tell apart such “unproblematic” images, who knows what will fool us next time. Criminals sporting an AI-generated face and voice will soon be able to create fake videos, audio recordings and content that might become almost impossible to tell apart from reality.

The biased

Again, AI-powered software is not inherently good or bad, biased or unbiased. They are merely the reflection of the people who wrote their code, designed and trained them. A biased word brings to life a biased code, taking into consideration that who wrote the code and the companies profiting from it are probably Western European and North American companies, it is easy to predict some biases will inevitably escape scrutiny.

Reducing AI bias means filtering harmful content and ensuring that the data used to train it comes from diverse and representative sources. Developers need to be transparent to the public about how their machines work and learn, and should be held accountable for discriminatory and harmful practices.

Accountability, transparency, and fair practices should be at the center of the debate around AI. We are all worried about the future of our jobs under the new “AI era”, but maybe instead of pointing the finger at a machine, we should focus our attention on the system that seems to be unable to offer an alternative to heavy, repetitive, uncreative labor instead.