Absolutely everything in excess is bad, no matter whether we talk about consumable products already nocive for our health as cigarettes and alcohol, or activities that normally might have a great impact as sports, eating and sleeping. Indeed, this time Google just demonstrated it while, in the middle of the fight against bullying, classism, sexism and racism, its AI algorithms are claiming for a pinch of acceptation in what concerns to human race and gender in order to develop and evolve faster.
Background scenarios from the timeline of our history on earth demonstrate how hard it’s been even for us to accept and tolerate. Of course, nowadays it could seems just ridiculous to degrade someone because of their color skin, but a few hundred years ago there were even murdered people because of that reason, and the change of narrow-minding it’s still fresh if we dare to talk about sexual identity, for example — Even though not only over the tech environment, but almost no matter which field, we may find exceptional, respectable gay people who stand out brilliantly, as it’s the case of Tim Cook, to name but one of my favorites.
That being said, throughout our personal and groupal battle against those old demons, we have to accept we’ve probably exaggerated a bit and lost the balance between just denigrating and unaccepting our gifted race and genders, which is turning into an obstacle to AI. By any chance, computing intelligence understands about our deep complexes? I don’t think so.
So now that computer vision is increasingly good at recognizing different facial expressions, Google realized through its new AI-based smile detector, that it’s time to embrace and accept our races and genders more than ever before, just because it helps to reduce the technological bias.
Google’s AI experiment: not racism, but racialism
In a nutshell, Google is experimenting with the racialism through their AI algorithms. Dusting off a little bit the concept, it’s nothing more than a scientific racism, or to face that human species are naturally divided into ostensibly distinct biological categories.
In this way, all the evidence is published in a paper at arXiv where Google researchers found that codifying or training a racial classifier on four race subgroups (Asian, black, Hispanic, and white) and two for gender, did has a quite important impact into improving smile detection algorithms. To put it another way, smarter AI that can be 91 percent accuracy upon state-of-the-art smile detection.
So far, naturally, numerous researchers have been wondering and are hesitant to consider classifiers of this kind, and the reason is the assumption that it’s easier to be guilty of bias — Or at least be accused of it when something goes wrong and the system shows explicit both racial and gender categories, and therein lies the intricacy.
Why do we need AI to recognize faces?
But lastly, you might be wondering why is it useful to polish artificial intelligence into face recognition skills. Far beyond of belittling anyone, it has to actually do with a tech-inclusion of the racial minorities as women with androgynous features, where algorithms have been underperforming until now.
On the other hand, as part of the coolest experiments the big G have been testing recently, a bright light over the path of Google’s researchers of futurism is twinkling: facial recognition through AI could have million of applications. In fact, Google researchers Hee Jung Ryu and Florian Schroff are about to present some sort of “electronic screen protector,” whose AI-based eye detection in the front camera will detect when someone is peeping at your phone over your shoulder.
In any case, in addition to the promising results and care taken to be aware of bias in all its forms, experts at Google included a section in their doc called “Ethical Considerations,” explaining race and gender identification as actually an end-goal — Or in simple lines, stating this is a moment to let the racialism to computers and AI systems instead of promoting the nonsense phantom of racism that have been haunting societies since we all can remember. Smile, artificial intelligence needs us to be proud of us!