Things weigh. It may seem obvious but it is not. The weight of things explains our behavior, the economy, geopolitics and religions. All. We might think that information or intelligence is ethereal and not subject to the laws of physics. But not.
Let’s take intelligence. We can define it as the property of an agent to establish objectives and approach them with their actions taking into account their environment. It is a definition generic enough to cover everything from an E. coli bacteria to a sapiens sapiens, and inclusive enough not to exclude artificial intelligence.
Note that it follows that intelligence is not a binary property but a continuum. Whether an agent exhibits more or less intelligent behavior depends on the data that its senses provide it, the ability to store and process them in accordance with its biological algorithms. If it is also capable of reprogramming these algorithms, we say that it has learning capacity. In other terms: data, algorithms and computing. Thinking in zeros and ones makes evident the subordination of information to physics: a bit occupies a space on a chip and energy is needed to change its state. It is precisely this combination of data, algorithms and computing that explains the current AI revolution/pop phenomenon. The large amount of data we have generated and advances in chip design have revitalized algorithms, such as neural networks, that have been in the freezer since the 1980s.
And what role does Barcelona play in all this? Barcelona is relevant in two of these areas: high-performance computing and algorithms. As for the first, “tenim un nom el sap tothom”: MareNostrum. In its first version, in 2005, it was the fourth fastest in the world. Today, a cutting-edge chip of less than 10cm2 has twice the power of MareNostrum 1. MareNostrum 5, in its accelerated partition, has 4,480 of these chips!
Large amounts of data and processing power are necessary but not sufficient conditions for AI. The hardware would be little more than expensive junk if it weren’t for the humans who use it, for the sapiens sapiens who give it the goals it doesn’t have. Here, Barcelona also has specific weight with top-level researchers such as Mateo Valero in supercomputing, Alfonso Valencia in life sciences, Ramon López de Mántaras in machine learning (ML), Carme Torras in social robotics, Karina Gibert in ML, ethics and explainability and Ulises Cortés representing knowledge.
That things weigh, and which ones weigh more, should be explained more often. This month, two books that go in this direction have coincided: Artificial intelligence explained to humans, by UPC professor Jordi Torres, and 100 things you need to know about AI, by Ramon López de Mántaras. They both enter the exam.
