Facebook’s AI Head Discusses Limits of the Field

One of the most exciting parts of the tech world in this day and age has to do with artificial intelligence (AI). A lot of companies are dipping their fingers into this new industry, and it’s making people discuss the seemingly endless possibilities that this could provide us with. From personal assistants all the way to simple stickers that are meant to personalize your social media experience, the progress that has been made has been nothing short of fantastic, but in spite of that some influential people that are working in the field don’t really seem to think that the sky is the limit.

One such person is the Vice President of Artificial Intelligence at Facebook Jerome Pesenti. Facebook has been a trailblazer in terms of the uses of AI that would actually improve the lives and experiences of consumers. Some experiments that Facebook has tried out may not have gone so well such as that personal assistant that didn’t really give people the experience that they have been expecting, but uses of AI in the algorithm designed to personalize your Facebook experience have been quite successful and have made the social media platform borderline too addictive for anyone to actually end up using in a healthy manner all in all.

There have been some controversial uses of AI as well. One big example of this can be seen in deepfake videos which are basically computer generated videos that look and sound like real life people, something that could lead to a fair amount of chaos since no one would really know what videos that are being created are true and which ones are actual fakes that would make it difficult for you to trust anything that you see on any social media platform.

Some progress that has been made according to Pesenti has to do with preventing bullying and hate speech. AI is becoming more advanced and is able to understand the nuances of human speech, something that allows it to prevent serious occurrences of hateful or violent behavior on the platform. AI is also becoming better at detecting deepfake videos, to the point where users are being encouraged to test these detection techniques out on new deepfakes that the platform was making on its own.


With all of that being said, Presenti seems to think that it is unlikely that AI would match human intelligence anytime soon. For one thing, the very concept of human intelligence is something that is rather vague, and it differs from person to person. The term artificial general intelligence is used to describe AI that can think for itself, but there are many limitations that are being faced in this regard.

At the end of the day, deep learning and other such techniques are made by humans, and the main limitation here is that they end up making the computers learn human biases before they pick up on things like empathy and other aspects of emotional intelligence. While Presenti didn’t outright say that AI will hit a dead end, he did say that progress will probably end up slowing over the years until we find a new way to advance the field of artificial intelligence.


Photo: Twitter / Thierry Zomahoun.

Read next: Facebook Renews Efforts to Prevent Discriminatory Ads

No comments:

Post a Comment