Carlos L. Ariztegui: “We still don’t know how to make AI work like the human brain”

Artificial intelligence (AI) will reach all businesses, even if they don’t realize it. This was stated by Carlos López Ariztegui, a communications engineer specializing in the use of artificial intelligence in companies at the MIT Sloan School of Management, and currently vice dean of the Esic Business School. Ariztegui, who developed his career at companies such as Apple or Nokia, points out that artificial intelligence will help companies improve productivity, as the spread of the Internet and office automation did a few decades ago, but he is still far from being able to emulate it. Human intelligence.

Question: This week, moves at the top from OpenAI, the creator of Chay GPT, made headlines on Friday Reuters She stated that the company is dealing with a new tool that “could end humanity.” Should we fear artificial intelligence?

Answer: In artificial intelligence, as in everything, we take it to the extreme, and we find disaster deniers. But we need to be clear about what AI is: it is a system that, no matter how advanced we think it may be, trains with data and makes approximations. They are still statistical estimates. ChatGPT is trained on an infinite number of texts, so the average person cannot do it, it would take 20 thousand years. Machines outperform us there, but when you ask them questions that require the application of logic, they make mistakes because they are not trained to think. We still don’t know how to make AI work like the human brain, but we are hearing many apocalyptic predictions that border on science fiction.

Q: After ten years we won’t know either?

a: When we talk about intelligence it is because we have the ability to use our senses to understand the world around us. The human nervous system is the result of millions of years of evolution that allowed consciousness to emerge. It is this process that leads us to ideas and generates the abstractions and the world of symbols that characterize intelligence, and it does not know how to do that like a machine. Artificial intelligence, today, never ceases to be with us Programming And the computer, no matter how advanced it is. In the human brain there are no parts, ones or zeros, but rather biochemistry, and although we have some understanding, we do not yet understand it in a concrete way. Replicating the human brain without knowing how it works in a way that has nothing to do with it is very complicated. They will help us with productivity, but they are not smart.

Q: Leaving out the end of humanity, what risks do AI systems pose?

a: There are risks with very stupid apps. There are algorithms that reflect biased perceptions and can deny someone credit for being black. However, the ultimate responsibility lies with companies that do not protect their customers’ information well, which can lead to many reputational crises.

“AI will help us increase productivity, but it is not that smart.”

Q: Are Spanish companies ready to integrate AI?

a: Large companies and small companies move at very different speeds. For example, large insurance companies have long used data analysis and the use of forecasting systems, so-called Machine learning. It is the artificial intelligence that is being used today and transparently. An algorithm looks at your risk profile, expenses, saving capacity, income, etc. And budget setting. In the case of SMEs, those that use and generate data use CRM-type systems and platforms that have integrated advanced analysis and forecasting functions. And this is not artificial intelligence, but data management is already penetrating this type of company.

Q: Will the use of artificial intelligence in companies increase?

a: It depends on your need to handle the data. If you’re not managing large amounts of data, you can use your wits and Excel. If you use more advanced processes, you’re likely to inadvertently access data tools, things we call AI, but they’re not that advanced. In ten years, it will be used more and more, integrated into the applications we use today, and we will benefit from it more and more.

Q: Are professionals and companies showing interest in training in AI?

a: There is great interest in companies in training in artificial intelligence, in advanced fields and, above all, in non-advanced fields. The first step is the call Machine learning There is a lot of demand. It’s like the Excel of the future: everyone has to figure it out manually or semi-automatically.

Question: Beyond corporate self-regulation, what should the public sector regulate?

a: While departments tend to be overorganized, absence is often bad. Although before you start organizing, you need to understand its impact. The role that the European Union plays in this aspect is positive, because it is trying to raise awareness among companies while curbing some aspects such as identifying people via video on the street. The regulation also provides recommendations for use that can be useful to companies and users, who have a clear idea of ​​the ethical dilemmas involved.