menu Home chevron_right
Ground ZeroRecent Shows


Ron Patton | June 8, 2018
Sponsored By:



The other day, someone asked me about why I have been focusing a lot on how technology has taken over our lives and how in a lot of ways the new Artificial Intelligence has backfired in ways that the creators have not anticipated.

I said that the reason is simple – I predicted that this year would be the year in which we would start to see the roll out of new technologies that would give us an indication of what we should be expecting as the technocrats start ironing out the technological wrinkles that exist with Artificial Intelligence.

Six months ago, as the new year began I decided that it would be important to talk about the role of technology in improving human existence and how unequal access to such technology poses a major threat to the overall welfare of the human race.

My show was entitled 2018: Living From Data Day and I focused on what I could see as a malicious A.I. and how it would be implemented and allowed to run our lives from this year forward.


Well, six months into the year and it is apparent that while many people will claim they are not affected by Artificial Intelligence or the internet of things – the majority know that it is becoming inescapable and that everyone will have to admit that we are allowing for malicious A.I. to rule our lives and that the prospects of peaceful technological Pinocchio’s wanting to be human are right now out of the question.

For example, this week, researchers at MIT unveiled their latest creation: Norman, a disturbed and psychotic AI.

It was named after the killer in Alfred Hitchcock’s film, Psycho.

The statement about Norman from MIT read as follows:

“Norman is an AI that is trained to perform image captioning, a popular deep learning method of generating a textual description of an image. We trained Norman on image captions from an infamous subreddit that is dedicated to document and observe the disturbing reality of death. Then, we compared Norman’s responses with a standard image captioning neural network on Rorschach inkblots; a test that is used to detect underlying thought disorders.”


While there’s some debate about whether the Rorschach test is a valid way to measure a person’s psychological state, there’s no denying that Norman’s answers were creepy as hell.

The ink blots according to Norman showed anything from a man being pulled into a dough machine to a man being murdered in broad daylight in front of his screaming wife.

Every interpretation in the test became even more graphic in detail.

The point of the experiment was to show how easy it is to bias any Artificial Intelligence if you train it on biased data. The team wisely didn’t speculate about whether exposure to graphic content changes the way a human thinks. They’ve done other experiments in the same vein, too, using AI to write horror stories, create terrifying images, judge moral decisions, and even induce empathy. This kind of research is important. We should be asking the same questions of artificial intelligence as we do of any other technology because it is far too easy for unintended consequences to hurt the people the system wasn’t designed to see. Naturally, this is the basis of sci-fi: imagining possible futures and showing what could lead us there. Issac Asimov gave wrote the “Three Laws of Robotics” because he wanted to imagine what might happen if they were contravened.

It appears that testing A.I. for confirmation bias and cognitive capabilities of all kinds is part of the creation process of having Artificial Intelligence learn how to deal with all situations and personalities, including those personalities that can be sociopathic or even psychopathic.

This is all part of the training to make A.I. more human, making the simulation seamless.

The new wave of A.I. enhanced assistants — Apple’s Siri, Facebook’s M, Amazon’s Echo, are all creatures of machine learning; however, they still remain sterile in a sense and when they try to sound human they fail in sounding like they have no real ticks that make them sound like genuine human beings.

I have always wondered as to how people feel about that.

How do most people feel about the idea that technology companies are at the moment trying to simulate personality and putting it into enhanced Artificial Intelligence?

Alan Turing proposed a test for an artificial general intelligence: a computer that could, over the course of five minutes of text exchange, successfully deceive a real human in the course of dialogue.

Once a machine can translate fluently between two natural languages, the foundation has been laid for a machine that might one day “understand” human language well enough to engage in plausible conversation.

Obviously, Artificial Intelligence creators are now working on having a machine sound natural and mimic that human cadence so that humans feel comfortable talking to it. In fact they want it to be so advanced that humans would not be uncomfortable arguing with it or asking it for advice.

Back in the 1960’s, it was proposed that Artificial Intelligence, if perfected, should be able to not be noticed as artificial. Back then, scientists believed that if you could program a computer to mimic higher-order cognitive tasks like math or chess, you were on a path that would eventually lead to something akin to consciousness.

However, what scientists and techies are looking for is A.I. with a sort of ambiguity. After all a computer or robot can be confident and they shrug and be arrogant about how advanced they are but one thing a human does that an A.I. can’t do is be vague, or show doubt, or even get frustrated or be sarcastic—these are traits that make the simulation even more realistic.

The goal is to make A.I. conversation indistinguishable from humans.

There are now many companies who are using some of the prototypes of these A.I assistants to carry out what are called robo calls. There are also companies with a lot of money that are using this technology to scam people into thinking that they are speaking to a person, when the reality is that they are speaking to a robot that is using the call to phish them for information in order to commit fraud and identity theft.

If you’ve picked up the phone only to hear the start of an automatic recording, you’re not alone. Roughly 16.3 billion of these calls have been placed just in the first five months of 2018.

Now of course these calls that offer recordings are usually sent out by candidiates seeking your vote; however, now sophisticated systems of calling are now being used and many of these systems are using what appear to be advanced AI assistants that sound human, interact with the person on the phone and with an authoritative voice can have the callers give their address, phone numbers and bank account numbers without even questing whether or not they are being phished by a robot.

The Federal Trade Commission has strict rules that, in its own words, make “virtually all” robo calls illegal. But we’ve still seen the number of complaints about robo calls explode over the last few years.

The big question is how are they supposed to enforce the illegal practice of businesses using robots to do the talking, especially those calls that come from foreign countries.

Last Monday, thousands of people in El Paso, Texas received a mysterious phone call from several different Russian phone numbers.

The El Paso 411 Media Center received a tip at around 5 a.m. saying that a large number of people at a local Walmart all received a cell phone call from a Russian-based phone number at the same time.

Just a few hours later, employees from Podium Finish Sport Boutique Cafe Indoor training reached out to say that about 22 customers all received a phone call between 5 am and 6 am as well.

Many people had claimed that they all received a phone call from the number +7 954 400 00 33.

When answered, the phone call would be silent then it eventually drops.

When the number was called back, a robot would answer and say “Someone sent you a love song” and then proceeds to play a song.”

There are some people who have said that when they call the number back they feel sick when they hear the song and so there are people who are worried that the song is a “brainwashing trigger.”

Keep in mind that if you choose to call the number to try it, you are doing so at your own risk and of course it is an expensive international call to Russia.

It’s unknown why so many people in El Paso are receiving these calls from Russia and what their purpose is; however, it is setting an eerie tone throughout the city.

Another incident that has been reported in that a call with an unknown ID will show up on your phone. When you answer and an authoritative voice will tell you that you are being audited by the IRS and they will connect you with a representative. When the representative comes on the phone, you realize that he has a Middle Eastern accent.

When you ask if this is a real call the person on the other end will get agitates and in some cases becomes belligerent and even goes as far as saying that they are watching your family and will kill them if you don’t do as they say.

There have been many people that have complained that they have been getting robo calls from a woman who speaks Mandarin Chinese. This is also under investigation because the robo call messages are usually a variation on the same theme: “This is the Chinese consulate, we have an important document that needs to be picked up, it may affect your status in the U.S., press a button to speak with a specialist.”

Most non-Mandarin speakers just hang up. But Chinese immigrants sometimes engage with the robo call. One reason is the number that appears on the phones is sometimes the number for the Chinese consulate in New York, giving the calls a semblance of credibility. This is known as “spoofing” and is a common and illegal tactic.

According to the NYPD and a security expert in Hong Kong, this scam has existed for over three years — the phone call is connected to someone in mainland China.

Once the scammers get a Chinese speaker on the line, they often carry out what’s known as a “parcel scam.” That’s where the caller pretends to be from a courier company and claims that a parcel addressed to the victim is connected to a criminal case.

The phone call is then transferred to a supposed police officer who closes the con by telling the victim it’s a money laundering issue and if they transfer money to a Hong Kong bank account, police will investigate and resolve the case.

There is also the weird case of a robo call that monitors your political views on Facebook.

People have received calls from a robot that actually attacks people who post derogatory comments about Donald Trump on Facebook.

Robo calls are a thorny problem to solve. Calls can travel through various carriers and a maze of networks, making it hard to pinpoint their origins, enabling the callers to evade rules. Regulators are working with the telecommunications industry to find ways to authenticate calls, which would help unmask the callers.

In the meantime, the deceptive measures have become more sophisticated. In one tactic, known as “neighborhood spoofing,” robo callers use local numbers in the hope that recipients will be more likely to pick up.

In fact, I remember a robo call that I received that mimicked my number and when I picked up I heard some electronic beeps and static—it made me wonder if it was some kind of triggering call and so immediately hung up.

However, I was once told that the longer I stayed with the call, the easier it would be for the bot to record where I am and what my personal information is.

There was a notorious scam where a robo call would use the voice of Paul Marcarelli – the man who used to do the “can you hear me now” slogan for Verizon.

A person would pick up and the call would identify itself as a representative from Sprint explain that Paul is now working for them – the caller then hears “Can you hear me now?”

The caller then says “yes.”

Their “yes” reply is recorded and used to authorize fraudulent charges via telephone on the victim’s utility or credit card account – it basically uses a robo call to fool an automated system using your recorded voice.

Here is another thing you probably didn’t know – if you receive a robo call and the recording says – “If you want us to take you off our do not call list press the pound sign” do not press the pound sign. Robots often use that tactic to identify and target live respondents meaning that they are now aware that the response was not from another robot but from a human being that they can call again.

Here is another example, the phone rings. You pick it up and say “Hello. Hello. Helloooo.” But nobody answers.

It turns out there could be someone on the other end of the line: an automated computer system that’s calling your number and tens of thousands of others to build a list of humans to target for theft.

That initial call you get, with silence on the other end, is essentially the first of what are called the robo reconnaissance calls. They are recording everything you say when you answer in order to determine if there is a human on the other end.

It picks up on everything from background conversation to coughs.

Later in the week you will then get a call with a prerecorded voice that tells you, for example, “We are calling with an important message about your debit card. If you are the cardholder please stay on the line and press 1. Otherwise please have the cardholder call us at 1-877-etc.

If you’re thinking about ignoring it, the message tries to scare you into paying attention with a warning: “A temporary hold may have been placed on your account and will be removed upon verification of activity.”

That number leads to another automated system that prompts you to share personal details like your date of birth, your card number and secure PIN, the expiration date, your Social Security number.

Researchers estimate 1 in every 2,200 calls is a fraud attempt. And they’ve observed an interesting detail about the fraudulent 1-877 numbers. If you call back from your phone, which the criminals dialed, you get the prompt to enter personal data. If you call back from somewhere else, you get “this number has been deactivated.” So a regulator or police officer that’s trying to crack down will think, incorrectly, it’s out of commission.

The best thing to do is just hang up on a robo call. In my opinion, if a business wants my attention they will hire a real human to do the job. I believe that if they have to use Artificial Intelligence, than they are an artificial business.

Written by Ron Patton

Search Ground Zero


  • play_circle_filled

    Ground Zero Radio

  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled

    Episode 86 – How To Succeed In Faking An Alien Invasion Without Really Trying!

  • cover play_circle_filled

    Episode 85 – TIN FOIL HATE

  • cover play_circle_filled

    Episode 84 – BLOOD AND SOIL

  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled


  • cover play_circle_filled

    Episode 78 – METEOR RIGHT

play_arrow skip_previous skip_next volume_down

Ground zero


get all the ground zero news
directly to your inbox