By Rodrigo C. Barros
What do AI and chloroquine have in common?
*
The reader has already understood the astronomical impact artificial intelligence (AI) has on businesses and governments, forcing large economies to make strategic plans for the technology. What not everyone understands yet are the real risks posed by the technology.
A historical overview of artificial intelligence takes us on a roller coaster ride of exaggerated promises and gigantic disappointments. One of their milestones is the emergence of artificial neural networks (ANN) in 1958, when Frank Rosenblat invented the “Perceptron”. However, such networks only became a driving force in the region in the 2010s. Thanks to a favorable combination of catalytic factors, such as the explosive availability of data and the ability to use specialized hardware for matrix multiplication, KNNs have sparked an amazing revolution and surprised the world with their ability to handle complex tasks. The area has been renamed “Deep Learning”, an allusion to the increasing number of layers of neurons in network architectures that are now deeper.
When “deep learning” penetrated our everyday lives, quite a few futurologists came up with the old-fashioned prophecies: the uniqueness and the revolt of the machines, rightly Schwarzenegger in his Terminator costumes. But don’t make a mistake. The likelihood of a current RNA becoming conscious is as small as the size of a biological neuron.
Amazingly, the great threat to AI is reproducing human behavior too well. Incidentally, reproduce what we have the worst: prejudices. It must be clear that ANNs are correlation machines, not cause and effect. Furthermore, in a country where the President of the Republic does not understand that “correlation does not necessarily imply a question”, we need to be didactic and warn the public that there may be multiple correlations in the data, but that it is good science one that views categorical statements about causality with suspicion. Otherwise, we would have to admit that US government spending on science is responsible for the number of strangulation and hanging suicides in the US.
The biggest example of how much of the population doesn’t understand the difference between correlation and causality is the pseudoscientific crushes in Covids CPI in defense of chloroquine’s use to fight the virus. It is true that those responsible for the sanitary tragedy we are witnessing acted out of ignorance: they do not know the difference between correlation and cause, and do not understand the specifics and nuances of the scientific method.
We are at the same risk if we blindly trust ANNs. When we train such methods to discover patterns over different data, the generated models reproduce the disparities. A classic injustice case of AI is that of the COMPAS tool (Correctional Offender Management Profiling for Alternative Sanctions), which helped US courts to assess the likelihood of criminal recidivism. Would anyone be surprised if the algorithm found black people to be more likely to relapse?
The area of “fairness in machine learning” has found its way into the academy and warns everyone who benefits from AI: It is not enough for models to learn the existing patterns in the data well – it must be prevented from spreading prejudices. The AI justice drive is just beginning, with many opportunities to combat harmful prejudice. Models can be developed that consciously combat previously identified disruptive factors. Work can be done to develop synthetic databases adapted to take into account such factors. What you can’t do is pretend there is no prejudice. Or that it is not a problem for any of us if machines reproduce them.
In times of far-right governments that exude and promote prejudice, it is clear that the main battle within the AI is the same that we wage on a daily basis: the fight against injustice and prejudice.
*
Rodrigo C. Barros is a computer scientist and has a PhD in artificial intelligence from the USP. He is an AI researcher at PUCRS and research director at Teia Labs.
Subscribe to Serrapilheira’s newsletter for more news from the Institute and the Ciência Fundamental blog.