AI Is Going to Kill Us All

Just not today or anytime in the foreseeable future.

Alan Turing, who likely contributed more than any single person to the Allied victory in World War II, is also well-known for his test to determine if a machine is artificially intelligent. The Turing Test, which he called the “imitation game,” proposed that a computer would need to convince a human interlocutor - who would not be told whether they are speaking with a human or a machine - that the machine was in fact human, based on text only conversations. The test is well depicted in Alex Garland’s “Ex Machina,” a brilliant movie which takes the Turing test to the next level by challenging a young technologist to determine if an advanced android has achieved Aritificial Intelligence. I highly recommend this movie as it asks many of the right questions about technology traps (like Frankenstein) and what it means to be artificially intelligent.

Compare that to today’s AI frenzy and ChatGPT. While there have been many headlines that claim ChatGPT has passed the Turing test, ChatGPT is best described as a weak AI that can mimic human dialogue through text but lacks reason. My view of ChatGPT is that it is a chatbot that reads the internet and makes probabilistic language decisions that it does not actually understand. Here is Ian Bogost’s view on the matter in the Atlantic in a piece called “ChatGPT is Dumber Than You Think”:

First and foremost, ChatGPT lacks the ability to truly understand the complexity of human language and conversation. It is simply trained to generate words based on a given input, but it does not have the ability to truly comprehend the meaning behind those words. This means that any responses it generates are likely to be shallow and lacking in depth and insight.

This is why people find such a high error rate in ChatGPT (mistakenly calling it a liar, rather than an error-prone toy). It has absolutely no idea what you or it are talking about, it just is really good at pretending. A facsimile. A highly technical and advanced parlor trick.

But to listen to Big Tech and corporate leaders, AI is taking over the world. AI will bring on the singularity soon, a doomsday scenario not unlike the Terminator movies. Popular Mechanics reports AI may take over by the end of the decade, and even Elon Musk has said “AI is more dangerous than nuclear weapons.” I think he may be right, but at least 100 years in the future.

Similarly, when you watch Musk’s peers, CEOs and other business leaders selling their stock on cable business news stations, they are always telling stories about AI. Truly, “Generative AI” is the new catchphrase, replacing “digital transformation,” and “cloud computing,” as the secret elixir that will empower growth. Generative AI is the magic bean that will grow into the beanstalk. AI is Excalibur, leading us to a promised poet-king. Or something like that. As I write this, the ServiceNow CEO Bill McDermott is pitching his version of Large Language Models and Generative AI to sell his company’s stock on CNBC. And just look at these quotes at the Aiifi website, as they sell you their AI story:

"AI does not keep me up at night. What does? The idea that we might not use AI to its fullest, to help us collaborate and understand each other better." - Astro Teller, Entrepreneur, Scientist, and Head of Google X

"In the age of AI, human creativity and innovation will become even more valuable in the workplace, as machines take over routine tasks and allow people to focus on generating new ideas and solutions." - Sundar Pichai, CEO of Alphabet

Automation is good, so long as you know exactly where to put the machine." - Eliyahu Goldratt, Physicist & Management Consultant

That last quote indicates what I believe wholeheartedly - that “AI” is a buzzword which is actually being used by business leaders to denote “automation.” Automation has been around since humans first invented machines like the hand axe and the inclined plane. Certainly the Industrial Revolution began a process whereby humans sped up the ability to automate repetitive tasks to increase efficiency. ChatGPT and other large language model chatbots seem to be to me more of the same - a new tool being employed to increase human and worker efficiency. But ChatGPT is incapable of understanding what SkyNet is, let alone being able to develop it to murder the human race.

Indeed, the only model we know of that created consciousness is our own. Consciousness is difficult to define and a matter of open debate (ever since there was debate), but at its simplest the definition seems to me to be an awareness of internal and external existence. I know myself as a living being, and I know the world and others as living beings as well. Self-awareness, autonomy and sentience seem to be necessary ingredients to consciousness - which in my mind is interchangeable with Artificial Intelligence and is the only proper definition of AI.

Humans were not created and did not develop using algorithms. We developed from very small, single-celled organisms that through the pressures of survival grew over billions of years to self-aware primates. The issue with AI is that it attempts to build something using tools (algorithms) that are unlikely to mimic life, let alone the prerequisites for consciousness. In my opinion, not enough attention is paid to the biological aspects of intelligence, which again is the only example of true intelligence as far as the evidence is concerned. In truth, robotics engineers are now researching the embodiment of AI and are discovering that in order to develop AIs to better function as self-aware individuals, they likely need to be embodied.

Automation is not Intelligence, just ask a printing press or a manufacturing robot about Springsteen. We’re flabbergasted by ChatGPT solely because the machine is now automating language and communication rather than physical tasks. It’s novel, but it’s a trap. The current, bastardized definition of AI - “automation of language” - is the latest “Current Thing,” convincing humans through the bias of “availability cascade” that it will either destroy the world or lead us to Nirvana. Like a Nevada voter, I vote “None of the Above.” Don’t believe the hype.