HAVE WE LET THE GENIE OUT OF THE BOTTLE?
The article you are reading at present was not written with the help of ChatGPT. I sincerely hope computer-generated texts will be labelled #chptgenerated in the future. If not, we are set for a bleak future, right?
To make myself clear: I find ChatGPT the best available programme capable of generating more or less intelligent answers, even in Slovenian. We must admit it has the competence to write more eloquently than most people. ChatGPT is a tool that saves time and improves effectiveness. It can summarise and garner information and is surprisingly human-like. In a way, ChatGPT is a child prodigy. It scours approximately 100 billion documents and pieces of information to copy and paste texts without limits on access. Its ultimate goal is to create content and texts that seem unfamiliar to us, hence leaving us with the misleading impression that technology is super smart.
Still, we must consider its drawbacks. One of them, I believe, is its inability to make logical conclusions. In other words, ChatGPT lacks common sense. At first glance, the programme generates seemingly flawless texts, perfectly structured. But a closer look shows the texts have no real meaning and that the system often makes apparent mistakes. In short, we have no guarantee that what ChatGPT writes is correct, regardless of its professed reasonable answers. Another peril exists; the danger of vast informational pollution with texts generated by artificial intelligence. We should only consider the misguiding political practices which occurred during Brexit. We have witnessed a similar story in the case of FaceBook and Cambridge Analytica. Hence, it is crucial that texts are labelled as #chptgenerated or computer generated.
Artificial intelligence can shake the foundations of democracy
There are already several established views on ChatGPT. It is apparent that the programme can write a decent plot against your biggest competitor on the market. The line between that and publishing posts on various platforms is thin. In a few minutes, you can create a campaign that can smear the reputation of your competition. There is no de facto protection against such attacks, barring regulation within countries or the EU. Numerous institutions are focusing on the sides of such technology. The European Commission, for instance, is preparing a so-called AI Act.
The matter is simple: we cannot return to the time before ChatGPT. The situation is similar to the time when the internet first appeared. We must not forget that every phone today is more capable than the most advanced computer in the nineties. We should also remember how they proclaimed in-person events would end when they moved online. ChatGPT is thus a tool that can either fix or destroy your car.
Most importantly, we must understand the limitations of technology. Is the digital becoming the real and the physical? What is even real, and what is a hallucination? At what point do we become a disinformed society? Familiarisation with artificial intelligence can result in a spike in productivity and quality of life. At the same time, artificial intelligence can shake the foundations of democracy and contribute to the rise of new weapons. Every company or organisation with access to information can tailor the algorithm to their needs. The ultimate misuse of artificial intelligence could result in autonomous weapons we do not want and should ban. Worse yet are entrepreneurs who jump on the AI bandwagon for quick profits without regard for possibly detrimental social effects.
If we see AI as a tool to improve productivity, without conspiracy theories or whatnot, we will gain an indispensable assistant. Why should we do something AI can do better and more effectively than us? Still, I must side with a columnist who compared ChatGPT with a robotic vacuum cleaner that can clean your house decently, but one you still need to take into your own hands if you want the best results.
Notwithstanding the mentioned examples, we must accept that human intuition and understanding are fundamental for creativity. With ChatGPT, we have acquired a flawed but helpful assistant that is fast learning.
N.B. ChatGPT does not include information about events after 2021. Hence, you will not get replies to questions pertaining to events that occurred after this year.
Article by Gorazd Čad