Facebook, AI, yi,yi

Facebook AI, yi, yi

By Wes Hessel

Facing IT

One of the largest trends in technology today is that of “Artificial Intelligence” or AI. The Merriam-Webster dictionary defines AI as “a branch of computer science dealing with the simulation of intelligent behavior in computers.”

Predictive Models

The idea of bringing technological “thought processes” closer to human thought pattern isn’t a surprising goal. In particular, to aid companies in  handling repetitive tasks. But, more deeply, companies want to use a predictive modeling approach, to statistically glean an anticipated decision, occurrence, or reaction. Imagine a virtual coin toss. Statistically, how many times out of 100, 1000, 10,000 or ten million will a flip result in a heads or tails outcome? Or, that you will order a Marguerita, not a Tequila Sunrise.

Fuzzy Logic

A “fuzzy logic” algorithm evaluates the truth values of variables. It is employed to handle the concept of partial truth, where the truth value is measured across a range. Standard logic indicates completely true to completely false. Such “fuzzy logic” and other similar algorithmic programming methods are being used in many aspects of our life.  It is used from traffic management (be it vehicle, data, or human) to security (physical and virtual) to that “smart speaker” sitting on your table.  One place most of us (at least it seems) encounter such “machine intelligence” is with the king of social media, Facebook.

Book ‘Em, Mano A Mano…

Parts of our interaction with others through Facebook are “one-on-one” connections – Messenger and so on; others are more like broadcasting, such as the classic “Post” and “Share”.  But behind those various communicative methods are multiple “back office”-type operations. Artificial intelligence systems and similar procedures are used more and more to do things as “easy” as suggesting your friends “Tag” up to these complex structures. For example, such structures are designed to help security people keep social medias safe, monitoring for prohibited language, identifying illicit bots, hacking, or illegal enterprises.

On Point, On Guard

In a cio.com article (https://www.cio.com/article/3280266/6-ways-facebook-uses-artificial-intelligence.html) from June of 2018, author Terena Bell cites six ways the online giant incorporates artificial intelligence into its work.  Beyond the (near?) future research toward mind-tech interfacing, Facebook seeks ways to literally connect to your brain waves. The purchase of the virtual reality company, OculusVR underlines the seriousness of this endeavor. Facebook Watch uses machine intelligence to predict what you want to watch, based on what you like, what your friends like, and even where you’ve been.

Policing

Policing Facebook’s Live streams for crimes or other proscribed acts is another AI application, as is stopping pornography. Artificial intelligence is even used internally to manage physical security for the social behemoth, as well as dealing with threats against the Facemeister himself, Mark Zuckerberg.  Facebook confirmed that in 2018 The Zuck received 14 million threats of one type or another (and you thought those few dozen flames you got today were ridiculous?).

Big (Oh) Brother Is Watching…

Unfortunately, the same systems that seek to stop illicit activity can also interfere annoyingly with day-to-day normal activities, or even put a kibosh on worthy cause work.  How many times have you posted something to have it then taken down for supposed foul language use, flagged keywords, or “violating community standards”?  Sometimes it will come back reversed after an apparent review (presumably a human, but the way things are going, it could have been kicked up to a higher power-ed process…). Other times the innocent thing you shared is gone for good and you don’t dare try it again for fear of potentially being locked out of posting or other basic tasks.

Blocking

Facebook users report being prevented from sending a friend request to someone the AI system deems there is no valid reason for having a connection with. Aggravating, infuriating, and nonsensical. Or, this has happened to me-I make a post; someone else comments on the post, I block the commenter then I am locked out of responding to anything in that thread’s comments, even from people who I didn’t block. How can this be? It was my post!

In The Market

When using the Facebook Marketplace section to post an item for sale, if it is bounced as ‘violating community standards,’ there is little recourse. A request for a review of the restriction typically receives no response.  People cannot post unused, sealed vitamins, which are apparently classified as a drug, yet used cosmetic items or open food packages are permitted (used food, really?).  Raising the price of an item is forbidden, even if it is discovered to have more value than was originally thought. However, people can list items as “free” which really aren’t. Discovering the real price doesn’t happen until the listing is opened. Facebook Marketplace offers no way to flag such listings, nor are these posts challenged, apparently.

Not For Sale

Even if there was no intent to sell something, it could end up in Facebook Marketplace.  My wife shared a picture of a cute foot stool we had just purchased at a local thrift store to a group and on her Timeline to show it off.  An automated procedure created a potential sale listing for the stool without confirmation from her, then asked her if she still wanted to sell the item. Head banging.

I Believe I’ve Been Locked Out

There has certainly been posts suppressed unintentionally that were religious, political, or personal in nature. Comments as simple as “Praying for you” or “Our newest…” have sometimes triggered the Facebook screening measures. People on both sides of the political spectrum will tell you they have had shares that were trapped out. It may seem like it only happens to items in line with your beliefs, but likely it is across the board, since these automated processes are written by fallible human beings, ultimately.

Fur-babies

Things on Facebook can get annoying if you are supporting a cause, or trying to offer help. For instance, my wife, who works with multiple volunteer networks on Facebook has encountered problems.  She often posts information and shares pictures of abused animals, or animals at shelters in danger of euthanasia. It seems the Facebook algorithm lacks awareness that many animals need “fur-ever” homes. My wife and her cadre of volunteer friends have saved many “fur-babies” lives. But, often, she or one of her online comrades have been locked out of posting or sharing, typically for a few days, because of posting too fast, or the sharing of the same animal too many places.  Yet the latest hoax posts, like the sick little girl who isn’t really getting a dollar per share, the bloody banana scare, and the inevitable COVID-19 related scams, keep coming around…

Learn While You Burn

No good deed goes unpunished, or so it seems sometimes. Presumably these mistakes occur because Facebook’s automated procedures are trying to cull out spam, mailshots and other unwanted contacts. One of the aspects of artificial intelligence that is intended to help – self-improving systems that “learn” when they did “right” or “wrong” in the decision-making process, is having human review of the machine choices.  For example, in the Bell article it was noted that Facebook in their attempt to prevent smut has accidentally censored mothers breast-feeding. Despite such errors, the social media colossus said their AI intercepts about 98% of porn posts.

Alternative Messaging

There are other applications to use instead of the big FB for broadcast/sharing or one-to-one messaging. Instagram, WhatsApp (are both owned by the social media giant), Pinterest, Snapchat, Hangouts, and so on, may be viable alternatives. It is expected assumed that these other applications also use artificial intelligence for screening.  Is there such a thing as trust worthy artificial intelligence? I suspect not, but I do imagine some platforms are better than others. I guess it depends on the humans behind the programming.

HAL-bot

The possibility of a computer misunderstanding its mission is another potential danger. Arthur C. Clark’s imagined computer, HAL (9000), in his classic science fiction novel “2001: A Space Odyssey” did just that. In an early Star Trek episode an M-5 computer, which was intended to replace humanity on interstellar missions, was installed on the Enterprise. However, when the M-5 was given control of the Enterprise it ran amok, endangering life and limb.

Artificial Intelligence – An Oxymoron?

The macro-microcosm of Facebook demonstrates that perhaps this burgeoning technology is a bit big for its britches. The use of artificial intelligence continues to grow. As with any maturing modus, it will take time for the benefits to outpace the detriments.  As machine intelligence increases in areas like government or medicine or warfare we must be increasingly vigilant. The point at which moral decisions come into play is a threshold at which we should pause and catch a breath to insure we don’t code in institutional bias. There must be a trigger, and ‘fail safe’ mechanisms that provoke human review and override.

Surrendering Control

Up to this time Facebook has been negligent in many areas, particularly assuring customer privacy.  It is clear there must be an incorporation of principled rules to manage AI applications. “Do no harm”, or Asimov’s, Three Laws of Robotics, could be useful to help stave off disaster in the many forms it could take. Ultimately there are many decisions that should remain in the hands of responsible people. What we gain in helping to extend the human reach in accomplishing things that would otherwise be very labor-intensive pose an obvious threat.  Humanity must keep a weather eye out for the threats of surrendering control to machines.

https://www.calamitypolitics.com/2019/01/29/super-artificial-intelligence8205/