We interact with it, it shapes more and more aspects of our lives, and yet most people are unaware of how it works. Although its use has become widespread at a crazy pace, artificial intelligence (AI) remains a great unknown. Jordi Torresresearcher Barcelona Supercomputing Center (BSC) and professor at the Universitat Politècnica de Catalunya (UPC), wants to help change this.

After more than two decades of scientific research, he has now presented the book ‘Explaining artificial intelligence to humans’ (Plataforma Editorial, 2023), in which he calmly explores the social, economic, political and even cultural impact it could have. technology. He is also the consultant of the ‘Artificial Intelligence’ exhibition, which will be hosted by CCCB from next October 18th.

Artificial intelligence emerged in the middle of the last century, but its popularity has skyrocketed in the last year thanks to ChatGPT. Was this a surprise?

Yes, the result was amazing. The technique wasn’t a surprise because the language model it used (GPT3) was already known in academic circles, but it was surprising how people jumped to participate in this experiment.

As we see with ChatGPT, the term artificial intelligence leads people to imagine conscious robots. What would be the best way to explain this?

It is a very old term, created in 1956, and what it means has changed significantly since then. This abstract concept allows us to always reference the latest developments in this field. Artificial intelligence is nothing but the evolution of computing.

It is trained with large databases of language models (LLMs) such as ChatGPT and responds to user questions from them. Artificial intelligence experts compare these chatty assistants to “random parrots”…

Yes, artificial intelligence algorithms are like a parrot that speaks very well but does not understand what it is saying or what causality is. And this is very important.

“Artificial intelligence is like a parrot that talks very well but does not understand what it is saying or what causality is.”

They answer reasonably, but they also make things up. Could this contaminate information on the internet?

That’s it. We have already used almost all of the data on the internet to train language models. The generation of synthetic data (those generated by ChatGPT) pollutes the environment, as the algorithm can learn to accept misinformation as good information. If you use this misinformation to train new models, you distort the original sample and increase the likelihood of a wrong answer. So you can increase the amount of data you train an AI with, but no new knowledge is generated. However, some information found on the internet before ChatGPT is incorrect. We’ve seen this with the rise of human-created fake news.

Proponents of AI systems like Sam Altman repeatedly assure that their target is General Artificial Intelligence, like the Terminator or the HAL9000 robot in ‘2001: A Space Odyssey’. Is it safe?

No matter how many movies are made, machines have no conscience. There is always someone behind you telling you what to do. Stuart Russell says we’re still a few Einsteins away from getting there. This is something that is not possible in the short term with current technology. It’s a matter that doesn’t concern me. I prefer to focus on the good uses it can provide for science, as well as its negative effects, such as the fake nudity we saw in Badajoz.

You guarantee that artificial intelligence will be the greatest human revolution comparable to steam, electricity or steam. Internet.

The world will change a lot, but I don’t know if it will only be because of artificial intelligence. Life before and after the internet is not the same, but we survived. We must assume that we need to understand this in order to decide where we want this revolution to take us. More collective awareness is needed.

“Life before the internet is not the same as life after the internet, but we survived.”

Despite this, major technology companies reiterate that this also threatens humanity. Is this a smokescreen?

Names like Elon Musk, who signed a letter asking to stop the development of artificial intelligence and at the same time invest in starting his own company, do not inspire much confidence. Artificial intelligence is a very powerful tool and you need to be careful. Its current situation already worries me because it could do a lot of harm by facilitating misinformation or being implemented in the military sector, which should be banned.

You joined BSC 18 years ago in 2005 as part of the founding team. What does a supercomputer do?

These are the most powerful computers today. They allow large volumes of data to be processed very quickly with algorithms. They do something that would take a simple computer about 3,000 years to calculate. There are only three of these supercomputers in Europe, and one is in Barcelona. And here all kinds of advanced scientific research can be carried out, for example for the advancement of healthcare.

More complex systems equate to greater data needs, computational capacity, computational speed, and therefore energy expenditure. How to integrate the development of artificial intelligence that can serve the fight? climate change With the growing ecological bill?

We cannot hide that the climate prediction models intended to stop climate change are exhausted to the detriment of climate change. Some studies show that GPT3 training is equivalent to 500 tons of CO2 emissions. There is a server infrastructure behind your GPS that consumes the brutality. Everything is very consuming, but we cannot give up on artificial intelligence because it can help us improve the future.

What kind of evolution will we see in artificial intelligence in the next 10 years?

He will change everything. Like? I don’t know. Over the past year, all industries have put forward hypotheses about how AI will affect them, some of them quite exaggerated. We’ve brute-forced our way through generative AI and more data and power, but I don’t think we can continue this exponential growth. Chips and data for this are missing.