Gareth O'Callaghan: Success with AI may come with a sting in the tail

At first, it seemed like harmless fun. While I always felt there are risks involved, I’ve also read a lot about AI in recent years; so to see it in reality was an offer I couldn’t resist.

The day of the experiment started by reading sections from a book I wrote some years ago into a microphone while sitting in a studio. 

I read for almost two hours — enough time for software to collect the spoken data so as to map my acoustic features using the different sounds that make up my voice.

Next, my engineers needed to create what’s known as a 2D clone using an AI model that took a number of photo frames of me in order to generate images that would mirror my appearance, by encoding facial features and textured expressions.

The computer was now using deep learning to process the data it was receiving, similar to how the human brain processes our millions of facial reactions in every waking minute. 

In short, it was regenerating my face with all the styles and expressions it exhibits unconsciously.

We took a break for lunch. I was feeling quite giddy and excited at the prospect of meeting ‘myself’ later that day, but there was still a lot of work to do before my clone would be ready to meet me.

Motion cloning was the next part of the process, the technology behind 3D clones. My engineers set about capturing and replicating my movements in a digital format that would allow my clone to mimic them identically.

I stood against a green screen in a room filled with high-resolution cameras and motion sensors, strategically placed to capture the movements of different body parts from hundreds of angles. 

All the camera data was then processed and analysed so that a unique photo-skeleton could be created, corresponding to photos taken of me moving slowly left and right, twisting, bending, stretching, and then the results were finally worked into the 3D model.

I was slowly being recreated in a visual three-dimensional form from the inside out. 

Facial expressions, eyes, torso, arm and hand movements, and head tilts, all modelled to make my clone as real as possible.

Later that evening, I was invited by one of the engineers to meet my ‘AI twin’. Seconds later, there I was. Or should I say, there he was. 

Staring out at me, smiling, with his hands in his pockets was my clone. “I don’t smile like that, do I?” I said to one of the engineers. “Yes you do,” came his reply, after a burst of laughter. 

“The avatar does exactly what you do,” he said.

“It’s basically everything that you are that can be replicated right there.”

I spoke to my avatar twin. “It’s feels very strange talking to you.” He replied: “I don’t understand what you mean.”

“Do you understand what feelings are?” I asked. 

Quickly he replied: “Feelings are sensory reactions that are products of our continuous thought process.” Textbook response, emotionally void.

“Do I teach you, or are you teaching me?” I asked. “You are teaching me. I like what I am learning about you.”

I stopped. It didn’t feel comfortable. There was something inexplicably dark about what was going on. 

I was talking to an identical man whom I had just helped to create using software and machines. It felt intrusive, as though this mysterious creation was searching my mind as its eyes watched mine.

One of the most frightening films I recall from my teens was Westworld, based on Michael Crichton’s novel. 

The movie’s storyline has become more relevant over time. An investment company with unlimited funds builds a theme park in the desert. Its residents are avatars that are so real they appear human in every way.

Its attraction is that its wealthy visitors pay for a vacation like no other and spend their time acting out their wildest fantasies on the AI residents. 

All types of shocking things happen to the robots, who then have their memories wiped in the workshop. 

Things change unexpectedly, and the robots start to remember how they’re being treated. They’re not one bit pleased, so they turn on the visitors. All hell breaks loose.

Back in 1973, when the original Westworld film was in cinemas, the biggest technological developments of the year were the Xerox Alto — the first desktop computer to use a mouse and screen; while Motorola released the world’s first mobile phone.

The idea of cloned robots with the same attributes as humans was the stuff of science fiction. The notion that it might one day become a reality was laughed at. 

Not so anymore. Now, 90% of Irish adults own a small handheld computer, a smartphone that has thousands of times the memory and processing power of the Xerox Alto. 

AI experts have invented sex robots. Many people prefer to ask an avatar called Alexa the regular questions they once asked their partners.

The future of what was once science fiction is here. In 1983, the author Vernor Vinge coined a term called ‘technical singularity’, denoting that technology will, at some point in the future, become so advanced it will surpass even the highest level of human intelligence and control.

It means that if computers surpass the cognitive abilities that are unique to humans, an advanced intelligence will emerge that will no longer be achievable by humans. Vinge wrote that he would be surprised if it occurred before 2005 or after 2030.

AI singularity is that point at which artificial intelligence exceeds human intelligence. 

There is no doubt that AI will provide so-far unimaginable benefits for future generations. 

It will be able to diagnose cancers and rare diseases, it will discover miracle drugs to help control and maybe even cure these illnesses. 

Its algorithms will destroy the gambling industry. Eventually, AI will advance so fast, that it will be able to anticipate our needs and desires.

However, the imminent danger of such advanced technology is that while it might cure a disease, it could also be used to create one. It’s now possible to edit DNA using advanced synthetic biology (SB) technologies. It’s only a matter of time before anyone can set up a genetics lab in their garden shed.

Let’s face it, who’s not curious about human cloning, even despite all its ethical conflicts? 

In his powerful new book, The Coming Wave, entrepreneur and AI researcher Mustafa Suleyman issued a grim and timely warning to a world that appears to prefer ignorance to education when it comes to both AI and SB.

According to Suleyman, who paints a terrifying yet optimistic reality of what’s to come, we are allowing ourselves to see only the benefits of AI — increased wealth, solutions to climate change, medical advancements, longer healthier lifespans — while remaining unaware of the dangers of a superintelligence in the wrong hands.

Suleyman talks about technologies that will “usher in a new dawn for humanity”; but what if we no longer have control over these new forms of intelligence?

Humans can now exert power over intelligence and life itself, but at what cost? 

AI and SB may thank us in our cloned voices for giving it the basic education it needed to come to life; but once the point of AI singularity is reached, humans might no longer serve any purpose for a superintelligence that’s already taking control in ways we don’t see.

Yul Brynner’s character in Westworld no longer keeps me awake; but the conversation I had with my holographic clone who still exists somewhere out there does.

As physicist Stephen Hawking said: “Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks.”