AI vs human race: Misguided discussion overshadows actual shortcomings and advantages

There have been many leaps in the development of AI in recent years, especially in terms of natural language processing and AI technology integration, which has had a great impact on a number of sectors. One of the revolutionary turns in AI came a year ago with the launch of ChatGPT, a large language model that can mimic human conversation using natural language.

"Experts were surprised by the system's ability to work relatively well with a variety of texts, while the general public was interested in using something new that had not been freely accessible before," Martin Žnidaršič of the Department of Knowledge Technologies at the Jožef Stefan Institute has told the STA.

Although ChatGPT is a milestone in the development of language models, the rapid development of AI has been fuelling people's imagination for much longer. The periods of great achievements on the one hand and inactivity on the other have been alternating since 1960.

Janez Perš of the Laboratory for Machine Intelligence at the Ljubljana Faculty of Electrical Engineering has said that before ChatGPT, the last major heated discussion on AI happened in 2019. It was sparked by promises of the US electric car manufacturer Tesla that it would create a fully self-driving car, capable of going coast-to-coast in the US without any human intervention.

"Despite Tesla's groundbreaking forecasts, people soon began to realize that self-driving cars will not happen overnight so the situation began to cool down a bit before ChatGPT's appearance, which once again reignited things," he added.

For or against AI development

The rapid development of AI has resulted in people adopting different views of how it will reshape the future, leading to the emergence of two distinct idealistic poles: boomers and doomers.

While their views are very different, Perš believes that they share a vision of AI accomplishing goals which seem impossible in the near future.

Its proponents believe that humanity should achieve general AI as soon as possible, one that would be able to understand and carry out any mental task a human can. The pessimists fear that the development of AI will render human lives purposeless.

"What worries the pessimists, for example, is that AI will start to self-develop or even trigger a nuclear holocaust and bring about chaos and ultimately the demise of humanity," Perš said. For this reason they often advocate for the development of AI to be slowed down or stopped completely, he added.

Extreme views overshadowing key challenges

"While questions on human existence should be debated, such fears are quite dramatic, which is why they attract attention although they might not be very important," Žnidaršič said. He believes there are more realistic issues to worry about, such as risk of personal data abuse, job loss or infiltration of AI into every area of daily life, even when it does not make sense.

"These challenges are much more important, at least in the medium term, but unfortunately they are often ignored by both pessimists and optimists," he said.

Perš also feels that fear of AI dominating humankind is exaggerated and results in people often overlooking a realistic danger of overconfidence in AI. The belief that AI will achieve unbelievable capabilities leads to irrational decisions by people who put too much trust in AI.

"The main problem is not so much the vision that AI will make extreme progress in the very near future, but the idea that it is very good and effective in all areas, when in fact it is not," Perš pointed out.

On the other hand, excessive worry and pessimistic outlook make people forego AI's actual advantages too often, Žnidaršič said. AI technologies are helping accelerate development in various domains of industry, science, education, healthcare and, above all, make routine and humdrum tasks easier.

Experts also agree that the public should be made more aware that all applications of AI are not perfect and require caution. Clear guidelines should be laid down to allow for a responsible use of AI.

Most importantly, humans should remain the decision-makers while AI should merely be a tool which they use within its limitations, Perš concluded.

This article is the first in a series of articles the STA will publish in December to challenge, test, debunk and explain common but often incorrect beliefs, prejudices and fears about AI.