The real threat of an artificial intelligence

By Rodrigo G. Barros

What do electronic AI and Chloroquine have in common?


The reader there understood an astronomical impact of an artificial intelligence (AI) in the electronic business in governments, so much so that seeing that large economies felt impelled to establish strategic plans for the technology. Not all of them yet understand thus operating system real risks that the technology poses to the world.

A historical overview of an artificial intelligence leads us to some rollercoaster ride over exaggerated promises and gigantic electronic disappointments. One of its milestones is the emergence of these artificial neural networks (ANNs) in 1958, when Frank Rosenblat invents Perceptron. However, it was only in the years 2010 that such sony ericsson networks became the primary driving force in the area. Thanks to some favorable union of catalytic factors, such as the explosion of electronic data availability and the possibility of using specialized equipment in matrix multiplication, ANNs caused some amazing revolution, surprising the world with their ability to handle complex tasks. The area was renamed Deep Understanding, alluding to the increasing number of layers of neurons in the architectures of these networks, now deeper.

With Strong Studying invading our everyday lives, zero has been little operating system futurlogos that came up with while old prophecies about always: the electronic singularity the revolt of these machines, entitled the Schwarzenegger in his costume on Terminator performs Futuro. However, let us not be mistaken. The probability that any current RNA will come to consciousness is as small as the size of a biological neuron.

The great threat of an AI, amazingly, reproduces human behavior too well. By the way, reproduce what we have the worst: operating system prejudices. It needs to be clear that RNAs therefore, electronic correlating machines, not electronic cause and effect. More than that, in a country where the President of the Republic does not understand that correlation does not necessarily imply a question, we need to be electronic didactics instructing the public that there may be several correlations in the data, however what good science is that which looks suspiciously at categorical statements to respect for causality. Otherwise, we will be forced to admit that operating system performs US government spending on science therefore operating system responsible for the number of suicides by electronic hanging strangulation in the US.

The biggest example of how good a share of a zero population understands the difference between correlation electronic causality thus operating system pseudoscientific raptures in the CPI of a Covid in the defense perform use of a chloroquine to fight a virus. It is true that the main operating systems responsible for the health tragedy we are experiencing acted out of ignorance: they do not know the difference between electronic cause correlation, electronic zero understand the specifics electronic nuances perform scientific method.

We are subject to the same risk when we blindly trust RNAs. Sony ericsson if we train such methods so that they discover patterns about disparate data, operating system generated models will reproduce the disparities. A classic case of injustice carried out by the AI ​​is the tool COMPAS (Correctional Arrest Administration Profiling for Substitute Sanctions), which helped US courts to estimate the probability of legal recidivism by part 2 rus. Any learning that would surprise me to discover that the algorithm pointed black individuals as more likely to relapse?

The area of ​​Fairness inside Machine Understanding has been gaining strength in academia, serving to alert everyone to an AI enjoy: zero enough that operating system models learn well operating system patterns existing in the data they need to be restrained from propagating prejudices. The justice effort in AI is just beginning, with plenty of possibilities to combat harmful operating system biases. Models can be developed that deliberately combat previously noted confounding factors. You can work on developing synthetic databases that are adjusted to account for such factors. What is not sony ericsson can pretend that prejudices don’t exist. Or that it’s not a problem for all of us to learn the machines operating system to reproduce.

In times when governments on the far right, which exude electronic promote prejudice, it is notorious that the primary struggle within an AI is the same one that we waged zero day to day: the battle against injustices electronic prejudices.


Rodrigo M. Barros computer scientist with a Ph.D. in synthetic intelligence from USP. AI researcher at PUCRS electronic Research Director at Teia Labs.

Sign up for sena publication perform Serrapilheira to follow more news perform institute electronic from the Cincia Essential blog page.

Back to top button