The race for AI is not without risks: what are the main dangers?

In recent months, artificial intelligence (AI) has made a sensational entry into our lives, through ChatGPT, but also many other tools. If this technology promises to make our lives easier and to provide us with many services, many specialists warn us about the risks to which it exposes us elsewhere. This is particularly the case of Sam Altman , who is neither more nor less than the creator of ChatGPT, and who believes that tools using AI can become “potentially frightening”. Geoffrey Hinton , former Google engineer and AI leader, goes even further! He regrets having created “terrifying” technology and warns the general public of the risks of artificial intelligence. Here is an overview of the different threats linked to their development.

Generating incorrect information

One of the main concerns of AI users is the reliability of the information created by this new technology. If we take the example of ChatGPT, the conversational tool created by OpenAI, the database it uses to respond to our requests does not support data generated after 2021. The tool therefore does not have the capabilities necessary to respond to a request relating to recent news, which will not prevent him from presenting assertions to us with a certain aplomb. Some users also noted that it sometimes provided imprecise or even inaccurate answers, particularly when it comes to technical subjects.

Whether we use ChatGPT, or another tool using artificial intelligence , to respond to our requests, it is therefore essential to verify the veracity of the answers it provides us. It would indeed be detrimental if you in turn distributed erroneous information to your customers, or any other interested person, or if you made strategic decisions on the basis of erroneous data.

A threat to many jobs

Historically, technological progress has made people fear that the automation of the processes they enable will lead to job losses. In recent decades, we have observed the closure of most industrial factories in developed countries. The recent explosion in AI capabilities reinforces these fears about the future prospects of many professions. On the other hand, and contrary to previous advances, it is not only unskilled jobs that are threatened by ChatGPT and others. Thus, professions such as those of translator, mathematician or financial analyst could experience a real upheaval in the coming years .

The American bank Goldman Sachs recently estimated that nearly a quarter of workers could be threatened by these new technologies in the United States and Europe. Worldwide, the number of jobs at risk is estimated at 300,000 . If AI has yet to prove itself to replace so many human workers, it is certain that the tools that use it will profoundly disrupt almost all professions, sooner or later. For humans, one of the major challenges of the coming years will be knowing how to adapt to artificial intelligence, to be able to use it effectively at work, while maintaining control over what it produces.

The fear that AI could overtake its creator

What if the main danger of artificial intelligence is that it surpasses the human who created it? In recent weeks, many tech entrepreneurs and academics have expressed fears that AI could spiral out of control. Some of them, including people actively working on its development, have even expressed concern that it could become a real threat to the future of humanity.

This fear lies in the fact that AI is learning, that is to say, it becomes richer as it processes our data. It could thus quickly become more intelligent than the human brain, which would then have difficulty muzzling it. The fact that humans rely too much on AI to respond to all the issues they face could also lead them to reduce their ability to think and reflect .

The widespread dissemination of fake news

In addition to ChatGPT, the image-generating AI Midjourney has been a hot topic in recent weeks. She is able to create images from scratch by answering all our requests, even the most absurd. We could see Pope Francis wearing a Balenciaga brand coat, or even Donald Trump in handcuffs. It thus becomes easy for anyone wishing to disseminate fake news to create images supporting their words. Text-generative AIs could also be used to quickly create a multitude of content that would reinforce the credibility of false information .

This possibility of generating fake news easily, and of being able to disseminate it on a large scale, raises fears of a weakening of our democracies. Indeed, a person with insurrectional objectives could create information from scratch, and disseminate it with the objective of arousing the rebellion of the people. In addition, it is important to be aware that AI could be used for actions aimed at discrediting a person or a company, by generating false opinions on the Internet, for example.

A new weapon for criminals

While the dangers of artificial intelligence are currently poorly regulated, criminals could take advantage of this legal vagueness for malicious purposes. For example, authorities are concerned about the creation of falsified video or audio content, also called “deepfakes”. These could in particular be used to impersonate someone else. AI would also pose a significant threat if terrorist organizations were able to subdue it, creating killer robots or autonomous vehicles programmed to dash into crowds . It sends shivers down your spine!

AI flaws could also lead to personal data leaks, making it easier to collect by more or less benevolent organizations. We can imagine that by collecting as much data about you as possible, scammers will be able to fool you more easily. You could also be the victim of blackmail, aiming to make you pay a sum of money, in exchange for the non-disclosure of compromising information concerning you. The possibilities of criminal uses of artificial intelligence therefore seem endless, which makes it a problem that our governments must quickly tackle head-on.

The dangers of artificial intelligence are therefore varied and should not be overlooked. The human will have to learn to master this tool, to make the most of the opportunities it provides, while limiting the risks to which it exposes it. States will be on the front line, with the objective of putting in place regulations to control AI and the threats it creates in criminal matters and in terms of employment .