I am often asked why the Government of the Generalitat of Catalonia needs to invest public budget in what we call “Catalonia’s Artificial Intelligence strategy” (CATALONIA.AI). Where is the need to do this, why does Artificial Intelligence deserve this attention? The answer is both fascinating and worrying: AI has long since ceased to be a simple technology and has become a political issue, a strategic issue, a country issue.
We must be aware (and often we are not) that each of us is constantly interacting with artificial intelligence systems. They are invisible systems that are here even though we are not really aware that they are here. There are AI systems that decide who our friends can be or what news you will read on social media. AI systems that decide what the next product you need most is or which movie you’ll be most interested in (or the one you’ll never even know exists…). We talk on our phones and chat with our voice assistants, and obviously we can do this thanks to AI. There are also AI systems that decide how to get from point A to point B based on traffic conditions or weather conditions or, perhaps, based on commercial interests. We find AI systems in financial markets (deciding when to buy or sell stocks), in industrial processes and in the “autonomous” cars that are starting to circulate on our roads. And Artificial Intelligence is increasingly present in our healthcare system and even in our legal system. So, systems that can decide what kind of treatment you will get when you are sick and visit the doctor or systems that decide whether yours is a suitable profile to get a certain job or a certain insurance or even if you are guilty or innocent of a crime!
This whole panorama could be magnificent if it weren’t for one small detail: Artificial Intelligence is not (far from) perfect. AI presents us with immense challenges that we must keep in mind and face not only from a technological perspective but also from a more social and ethical perspective. Challenges like that of privacy (several studies demonstrate how AI can obtain personal data, such as a person’s sexual orientation, from non-personal data exposed on the Internet… and much more accurately than a human!), the discrimination and bias (reproducing and amplifying the stereotypes and patterns we have as a society, such as discrimination based on gender or race), the asymmetry (a few technology companies have the data of billions of people, opening the door to the manipulation of the population, with examples as clear as Facebook and Cambridge Analytica or as subtle as the autocomplete of Internet searches), the veracity (Let’s not believe everything we read or see… AI can create completely synthetic text, images and videos but we will hardly be able to distinguish them from real text, images and videos), the lack of diversity (both from the point of view of algorithms, which end up giving you only what you like and hiding what you don’t like, creating a reality around you that doesn’t have to be reality, and from the point of view of the teams that develop AI, basically made up of middle-aged white men), or the challenge of security (where the “adversarial AI” uses an AI to “hack” another AI by adding imperceptible marks to the images that completely confuse the AI system that identifies, for example, the image of a “stop” sign for a “give way” sign!).
That’s why AI is no longer just technology, but a political issue. And that is why the most countries in the world (and not just Catalonia) already have strategic plans, roadmaps, action plans, and best practice guides to address the deployment of AI in their respective territories. Let us have no doubt that this strategic vision (social and ethical) beyond the technological vision is essential if we want (as the scientist Stephen Hawking said) that artificial intelligence is the best thing that has ever happened to us (and not the worst thing that has ever happened to us) as a human species.