The name “Artificial Intelligence” (AI) has covered a wide range of technology areas for decades. But it is the so-called generative artificial intelligence (those that are able to generate new content from text, images, video or audio) that brought this old concept back into vogue. These are the ones that have made headlines lately, those that have attracted the curiosity of the general public, and those that they generated a wave of new social, political, economic and ethical debates.

Independent contacted several experts with different profiles to try and shed light on the present and future of generative artificial intelligence such as Dall-e, Stable Diffusion, Midjourney or Chat GPT, which are undoubtedly the most popular of all. And the conclusions seem unequivocal: most agree in praising its potential, but differ greatly in its actual usefulness and in the level of risk and threat it poses.

The main debate on this issue has revolved around the possibility that these technologies could surpass human intelligence and perhaps they may rise up against us. Manel Sanroma, Ph.D. in astrophysics, professor of applied mathematics at Rovira y Virgili University in Tarragona and founder of CivicAi, believes that they are already “equal” to us in many ways, although in others they are even worse than us. But he is convinced that along the way they will arrive at an intellectual answer similar to ours. “To date, this is a computational problem, but this time is relatively close,” he says.

Among the antipodes is Ramon López de Mantaras, CSIC researcher and AI pioneer in our country, who assures that at the moment he does not see any evidence that they can equal or surpass us. “There is a lot of exaggeration, a lot of pseudoscience and a lot of interest in talking about this and about the possible extinction of mankind. But this is sheer stupidity and a red herring, because it won’t happen. What is happening is that while we are talking about this, we are not talking about real dangers.“, – notes the expert, who believes that these technologies are actually quite simple and have many limitations. “In five years, GPT Chat will be out of fashion and we will have something more science-based. I am convinced,” he says.

negative aspects

Beyond what the future of these technologies will bring us, there is no denying that, for better or worse, their impact is already being felt. Ricardo Baeza-Yates, director of research at Northeastern University’s Institute for Experimental AI, points to the first of the problems they have raised: intellectual property debate. And the fact is that generative artificial intelligence has been trained with texts and images taken from the Internet, which they later mix and use as a reference when creating their answers. So it’s not entirely clear whether what they’re doing can be considered mere inspiration, or whether they’re outright plagiarizing from the original authors.

Hence another problem arises. And this is what, as explained by Pablo Haya, director of business and language analytics at the Knowledge Engineering Institute, the information the AIs were trained with will completely mark the biases their responses will later have. Something that the development companies tried to iron out with each new version, but will be difficult to completely eliminate. “At the end of the day, every society has different values. But since AI companies are mostly Western, in general the answers they give are in line with Western values,” explains Haya.

Experts also agree that, due to their characteristics, they can be a dangerous tool for spreading false news, although they differ in the extent to which it will actually affect society. And this is what some suggest that people will soon get used to not trusting all the information that comes to them through the Internet, no matter how real it may seem, and they will go directly to reliable sources to verify it. Others, on the contrary, are not so optimistic and believe that they will further help cloud reality.

also to be seen What environmental and environmental impact does the development of generative AI have?which require a huge amount of energy to operate. The issue of regulation is also relevant, since today some countries are already adopting regulations, while others do not have any legislative basis for their regulation, and at the moment there is no international law on their application. Regarding the possibility that these technologies could take away people’s jobs, the opinion of the experts interviewed was unanimous: in the long run, they will create much more than they destroy.

Positive aspects

Generative AI is still very useful in some specific aspects to this day. “They change everything because they start working. Until now, the images and texts they create have not had a quality close to human. But from 2022 they begin to give results that can be used functionally“, says Miguel Angel Roman, co-founder of the Artificial Intelligence Institute. “There is no doubt that they have made a quantitative and qualitative leap in solving problems that imitate a person, in the center of which is language. And they are very attractive to the public, because until recently, AI was very marginalized in society and was associated with science fiction, ”adds Haya.

Tools that use images can greatly help artists in their creative process. And text, like Chat GTP, are able to summarize, organize and find information or write. That’s why They are very useful when you have to process large amounts of information.. And that’s why Haya explains that he sees them as another tool in word processors that will be useful to both the general public and specific audiences in areas such as healthcare or justice that many use in their daily lives. weekdays.documents.

“At the end of the day, this is a technology with a lot of empowering power. And then you can use it for good or for evil. The difference is determined by human intent.”, sums up Miguel Lucas, director of innovation at consulting firm LLYC. “They do the same thing in less time and much more in the same time. And this means releasing human talent from more routine tasks in order to devote it to tasks with greater added value,” the expert adds, explaining that its use in business is one of the most promising, although so far only large companies are developing projects. the pilot and technology need to be consolidated a little more so that they can be extended to the rest.

public image

Although this is a complex topic with more questions than answers and just taking its first steps, all indications are that AI will become more and more important in our lives. But at the moment it is clear to experts that the image of artificial intelligence in society is summarized between “fear” and “ignorance”.. There is something logical and almost inevitable in every new technological leap, especially when you consider that at the moment we receive very conflicting and in some cases disturbing information about these technologies. And the fact is that even the experts cannot agree on the calibration of what they will eventually become.