“Cars are not good for jokes”: Carlos Fiolais in an interview

In the last week, how many artificial intelligence (AI) systems have become part of your daily life?
When I want to use my mobile phone, it recognizes me thanks to artificial intelligence technology. When I search on Google, my previous activity is taken into account. Anti-spam system prevents unwanted mail. When I look at Facebook, the AI ​​selects posts for me. It’s the same when I check Amazon: they anticipate what I want to buy. E-shopping fraud is prevented using AI. I get better at getting from one place to another using AI. Artificial intelligence systems make it possible to translate faster and also correct spelling. And that’s not to mention ChatGPT, which I rarely use, but can be useful for a first search on any topic. Therefore, AI has been in our lives for a long time.

Want to explain what an algorithm is and how it affects AI?
An algorithm is a sequence of instructions to obtain a specific result. It doesn’t have to be a computer: for example, the Gomes de Sá cod recipe. Computer programs are examples of algorithms. “Machine learning” AI uses algorithms. The difference between AI programs and traditional ones is that the former learn; that is, they change with use. Computers imitate humans, although to a limited extent. In terms of input data, a given algorithm does not always produce the same result. One of the issues discussed today is the transparency of AI algorithms. Some of them are “black boxes”, which makes many people uncomfortable. They work, but no one really knows what’s inside. Similar to what happens in the human brain, we witness automatic decisions without being able to indicate how new they are.

What do you think about investors in artificial intelligence companies?
The largest artificial intelligence companies are the five that dominate the global economy: Microsoft, Apple, Amazon, Alphabet (from Google), Meta Platforms (from Facebook) are all in the top 10 largest companies. This top also includes hardware company NVIDIA, which started out making gaming cards and has now ventured into artificial intelligence for huge profits. OpenAI is a new platform, it is developing quickly, but it is largely owned by Microsoft. There are few European companies that are competitive in the field of AI, and some of them work with American capital (for example, the French company Mistral). China is a different story because there are companies there that are controlled or controlled by the government. In AI, as in other areas, there is a confrontation between the United States and China. These Western companies, based on the West Coast of America, are more powerful than most countries. For example, only six countries have a GDP greater than Microsoft’s. In this era of global capitalism, obvious questions of sovereignty arise: power no longer rests solely with the States. They may be democratic, but big companies are not. Their goal is the greatest profit, not the common good. States, which are supposed to protect the common good, respond late to technological innovation…

Does it make sense to create rules to monitor AI systems?
Regulation not only makes sense, it is necessary. The companies themselves are asking for this in order to play by the rules. In 2023, President Biden signed an executive order on AI safety that protects privacy and citizens’ rights. The European Union has created regulation based on the type of risk. UNESCO has also provided ethical guidelines regarding AI. The Pope called on AI to serve the cause of fraternity and peace.

What are the main ethical and legal issues raised?
Technology is developing much faster than ethics and law. They are forced to chase innovation, sometimes “chasing losses.” Europe has made regulation a priority, which is absolutely necessary. We must pay more attention to moral and legal issues and reconcile economic interests with human values. What can be done is one thing, and what we should do is another, completely different thing. We must prevent the misuse of AI, even banning some of it. Among the dangers is violation of confidentiality. Second, we are complicit because we are giving away our data in exchange for something that should be free. But for me, the biggest danger is the massive poisoning of fake news on the Internet, which not only undermines democracy, but can also provoke conflicts. The problem of regulation is not new: when cars were invented, we had to create codes to avoid chaos on the roads. We should also be concerned about democratizing access so that the already glaring inequalities between rich and poor do not grow.

AI is also an opportunity for development. How should Portugal position itself?
The field of artificial intelligence is growing rapidly with applications like ChatGPT. The line between the real and the virtual is blurring. There are huge opportunities and caution must also be exercised. In this area, we could mobilize researchers to create a national interdisciplinary center. There will be room for people from mathematics, physics and engineering, as well as from the social sciences, humanities and the arts. Many scientists support each other, and there is an opportunity to work together.

“Artificial intelligence depends on millions of poor microworkers,” sociologist Antonio Casilli told Expresso. As a scientist, how do you interpret these effects?
It is difficult to make predictions about the impact of AI on employment. But great innovations have always brought changes in the profession: some ended, others appeared. Overall it wasn’t bad. Thanks to steam engines, we no longer needed animal power.
or person. And professions began to require greater qualifications. Now, thanks to AI, repetitive tasks are completed automatically, even if they are office tasks. Professions that are more concerned with human affairs, such as food or health, will remain. We may have less time to work, and that’s not necessarily a bad thing. Casilli raises an important point: There are many low-paid human operators who help train artificial intelligence systems: for example, they confirm whether the machine scanned an image well. Economics is a political field and therefore lies outside the boundaries of science and technology.

What are the areas where AI is expected to have the greatest impact in the next decade?
The impact spans multiple areas. There is great potential in healthcare, such as automated diagnosis, drug discovery and psychological support. Justice services can be faster and cheaper. This will impact education and teachers, who are always needed, will have to adapt. There are and will be more self-driving cars. Services such as banking, insurance, energy and other government services can be more automated. In contrast, those areas that involve more human creativity, such as art and entertainment, will be least affected. Machines can’t joke…

If you had to give advice to a teenager about their future career, what would you say?
Follow your dreams.