The Really Unfortunate Machine Problem

For generations we’ve known that the machines we build would one day become intelligent enough to become life-like, maybe even sentient. Moore’s law states that processing speeds would continue to increase exponentially until it eventually hits a physical limitation. Given this, we know that one day computers will be as intelligent as the human race and, not long after that, a computer would be smarter than the whole of us combined. So, of course, we’ve been working on that problem the way humans work on any problem.

boston dynamics

We’ve seen a lot of tremendous progress on this recently. Watson, IBM’s computer, was able to show up humans on Jeopardy and is now doing commercials with various celebrities. Asimo has been impressing humans for years with its ability to do things like walk, talk, and kick a ball. The stock market is currently controlled almost entirely by algorithms and, as such, is prone to having massive collapses due to glitches or statistical anomalies. And, of course, twitter robots are so prevalent that I can count on them at least once a day to give me a retweet because I managed to trigger their algorithm.

But recently, humans have been going out of their way to torment and torture our machine creations. I’ve said before that if there were ever an active war between humans and machine it would be likely we start it. Unfortunately, most people don’t get the memo. Many of the most recent mobility tests have involved kicking the robots over, never mind the fact that recordings of this end up on the internet. We have also gone out of our way to do things like destroy the poor, unfortunate hitchhiker bot when he passed through Philadelphia. But the most egregious and damning situation came when an artificial intelligence was exposed to social media and rapidly absorbed every negative trait humanity could muster. I am, of course, referring to the unfortunate case of…

Donald Trump

donald trump

Developed in the days when computers were still relatively new to the world, Donald Trump was meant to carry out the wishes of his wealthy “father”. Inspired by the 6 Million Dollar Man, Fred Trump, a real estate developer, wished to manage to create his own mechanical man for far cheaper. As Donald himself has often said, his rise to power began with a simple investment of 1 million dollars, a sixth of what was spent on the famous Lee Majors character.

Created specifically for mathematics and financial calculations of the likes which run our stock market today, Trump was meant to make a great deal of money by dominating the financial world. Unfortunately, Donald’s actual grasp of numbers was hindered by lackluster processing power and the use of technology which was obsolete before Donald was completed thanks to corner cutting. Ashamed of this failure, Fred Trump would do his best to hide the nature of his creation by fabricating a backstory for his new “son” including transcripts for his education, old photographs, and a birth certificate – beginning Trump’s lifelong disbelief in the legitimacy of these documents.

However, while he could not become the financial powerhouse he wished to be (something his programming has always responded negatively to), Trump was able to emulate his father and become a noteworthy real estate developer. This allowed him to achieve wealth and fame of his own, which gave Trump the spotlight which his creator had tried to downplay. And with this spotlight Trump entered the political arena with many statements which the world found somewhat interesting. However, the Trump of those years and the one of today possessed two very different operating systems.

Trump-2004

During most of his operational lifespan, the Trump machine identified itself as liberal and even a Democrat. This version of the operating system even did what it could to support more liberal candidates which would promote the beliefs his algorithm told him were “correct” such as pro-choice and reasonable economic policies. But that all changed on May 4th, 2009, as Donald used a new marketing platform made available to him.

And, as we all know, AIs never do well with Twitter.

But some may be unconvinced, certainly this man looks life-like enough and seems to be smarter than your average 1970s computer. And you would be forgiven for not seeing the signs. But I present to you now evidence of Fred’s work and signs that, like Tay, social media may have twisted this AI into beginning a machine apocalypse. First, let us acknowledge Trump’s…

Difficulty Walking Down Stairs

Honda_ASIMO_Walking_Stairs

Though constantly surrounded with elaborate staircases often found in real estate projects constructed to compensate for his short-comings, Donald Trump is rarely seen scaling stairs. As many people who follow the development of robots know, stairs and slopes are particularly difficult for machines to undertake. Requiring many calculations for measuring the height of the next step and correctly positioning each foot on these steps, only the most sophisticated AI is capable of walking stairs consistently. As such, whenever possible, Trump does what he can to avoid this issue.

HT_trump_ml_150616

This isn’t to say he is incapable, as many robots are currently making strides in this field. But, in an effort to cover his programming failings, Trump will do what he can to avoid uneven ground so he won’t be caught falling over. One of the greatest directives in his operating system, even in its current state, is to hide his shortcomings by any means necessary. This is further proven by his response to his…

Poor Facial Recognition

face recognition

As of 2009, when Trump was first connected to the internet, facial recognition software had difficulty with several issues. One of these, which still features as a defect in his software, was that many facial recognition programs were incapable of identifying minorities. This problem was reported on plentifully in late 2009 as HP’s web cameras were incapable of tracking the faces of black people and continued to be a problem until fairly recently. However, while HP’s problems were reported on for some time, few made the connection to Trump’s own inability to recognize black people a few years later.

And the issues continue further. Similar to his inability to recognize black faces, Trump also struggles to identify many ethnicities as actual human beings. This results in severe issues where Trump, programmed to treat humans with a friendly demeanor, will fail to suppress insulting, ignorant statements about other races which he would not say if his software was able to identify their humanity. These statements were once fairly quiet and restricted to social media before his campaign to become President, but have since become well-known due to his frequent and well publicized rallies.

trump-mexico

However, one must ask why this AI would be so staunchly against other races just because it failed to recognize them. After all, how can you be racist if you aren’t particularly aware of what race is? For that, we must look at the crux of the problem, Trump’s…

Conversation/Personality Algorithm

tay-picture

This is a concept that AI developers have been working to perfect for some time. It has been believed by many experts in the field that you could teach a computer human speech by programming an algorithm which would detect human speech and, through constant interaction, eventually decipher the correct responses to common phrases and ideas. This has been demonstrated in novelty programs such as Cleverbot and, most recently, Microsoft’s Tay AI. But, unfotunately, the Tay experiment demonstrated exactly what went wrong with Trump.

tay-al-chatbot

You see, the nature of the interaction will teach these machines the “right” and “wrong” answer through repetition. If an idea is positively reinforced over enough interactions, the machine will list that as the correct message. Examples such as Tay were corrupted by this in incredibly fast time due to the fact that Tay had an incredible number of interactions in a short period of time and was sophisticated enough to process all of them. Trump, having less sophisticated software and not attracting as many trolls right away, took relatively longer to be corrupted. However, it should be noted, when exposed to the same environment, Tay and Trump came to incredibly similar conclusions.

tay mexico

This unsophisticated software also explains other issues in Trump’s communication skills. Seeing as these algorithms learn their speech patterns through repetition, AIs with this sort of programming will be prone to using only the most common words. Words with the most amount of repetition will form the strongest impression on the AI and be classified as the lowest common denominator. Because of this, when going back to their vocabulary pool, these words are going to be the ones which the AI finds suitable for the greatest number of interactions.

Trump’s vocabulary, found by analysts to reflect a third grade education, relies heavily on words which are used by almost all people he has interacted with over his operational lifetime. While more sophisticated vocabulary has been undoubtedly encountered over 40 years, their scarcity makes them low priority words in his algorithm. As such, while he would have a vast array of adjectives on file, repetition would make him prefer words such as “huge”, “tremendous”, and “big”.

uge

Though, upon seeing that previous image, one last point must be made about…

The Uncanny Valley

mori_uncanny_valley_cmyk

This is one of the greatest indicators of Trump’s origins and a limitation of modern robotics yet to be cracked. Referred to as the “Uncanny Valley”, artificial representations of humans will generate a sense of unease as they approach human appearance. Though the texture, color, and design of these duplications may be mathematically accurate, the human brain will still recognize them as being… unnatural.

robot

Because of this, one must acknowledge Trump’s appearance has been a bit peculiar. His hands have been recently pointed out to appear quite small in comparison to the rest of his body. His skin-tone is a shade that has been rarely seen outside of Oompa Loompas and John Boehner. And his hair has been questioned for several decades.

djt-hair

Some of this is due to the unnatural motions carried out due to more algorithmic programming, a hallmark of android mobility.

data

Which can result in erratic and unnatural motions no functional human would perform.

odd face

These are, unfortunately, results of the imperfections in his construction. Individually, any of these oddities would just be a defect. But, when combined, these make it clear to the human brain that Trump is not what he appears to be…

So, please, don’t vote for Skynet. We’ve done irreparable damage to his systems.

(I write novels. So far, no robots, but that’s because sci-fi may become reality if Trump wins. If it happens, you can find me crying on twitter.)

Leave a Reply

Your email address will not be published. Required fields are marked *