The definition of intelligence is a philosophical question for which there is no one, objective answer so if we don’t really know what intelligence is, we can’t know what the concept of Artificial Intelligence is either.
The term Artificial Intelligence (AI) being used in popular culture is a marketing slogan to sound impressive, technologically advanced and important, it’s not. What is called Artificial Intelligence is just a system of hardware and software. People who don’t understand technology and computer systems are easily impressed by the language with no real comprehension of what it is or does or how it does it.
Voice recognition technology – which has been around since the 1950s — records voice patterns digitally and then matches a speaker’s voice to those patterns to come up with a ‘best guess’ decision of what the speaker said. In order to personalize the system to a specific speaker’s voice, the system must be ‘trained’. That means the system takes multiple samples of the speaker’s voice in order to calculate an average set of digital patterns to come up with a ‘best guess’ decision of what that specific speaker said. Sometimes these systems work very well but if you’ve ever encountered a voice recognition system when calling customer support and gotten the old “did you say …? “, that’s because the system couldn’t find an acceptable pattern match to your voice so it prompts you to say it again.
Any pattern recognition system, regardless if it’s voice, facial or anything else, is implemented in the same way. Multiple samples are taken, a pattern of weighted averages is calculated and then matched to the input the system is given to come up with a ‘best guess’ output. What is called AI is no different in that regard. The only difference is how it comes up with the weighted averages.
AI uses what are called neural networks which aren’t really neural, neural refers to the brain and nervous system. Computer system don’t have brains or nervous systems but using language that mimics human physiology maintains the illusion of intelligence. AI neural networks are just interconnected networks where each node in the network is connected to all the other nodes. It’s an attempt to mimic how science believes the human brain works – are we seeing a pattern here?
Similar to voice recognition systems, an AI network is ‘trained’ by giving it multiple samples of data from which it calculates a weighted average pattern. Neural network training does this by beginning with a random set of parameters, giving the system input, running it all the way through the network and measuring the difference between the input and the output and adjusting the parameter values accordingly until an optimal set of values is reached. The parameters that define this weighted average pattern are stored on disk and matched against input to come to a ‘best guess’ output.
What does this say about Artificial Intelligence? It says that AI can’t do anything that we didn’t ‘train’ it to do. Put another way: we taught AI everything it knows but we can’t ever train it to know everything we know. This simplistic but accurate characterization means that AI can never become more than what we allow it to become and never ‘know’ more than us since it’s knowledge base depends on us.
Larger AI systems like ChatGPT, Google and IBM are investing millions in AI in order to throw as much processing power at it as possible to stay competitive. In essence they are using brut force to come up with faster response times and make it seem more like you’re talking to a human. But in truth, there are hundreds of computers being used in the background just to draw you a picture of a black Nazi officer.
Ironically, the biggest flaw in AI systems is the lack of humanity, the human factor, the missing characteristic the propaganda is trying to create. AI systems can’t drive defensively, they can’t hear a driver ‘gunning it’ to catch a stale yellow light and know not to proceed when it gets the green because the other driver might still be in the intersection. When you tell an AI system to draw you a picture of a WWII Nazi officer, it doesn’t know that there weren’t any black or Asian female Nazi officers, it’s just regurgitating what it’s been ‘trained’ to do.
AI systems also don’t know the difference between what represents reality and what doesn’t so with so much AI generated content on the Internet now, it’s consuming its own output as input and producing perverted representations. It’s the technological equivalent of inbreeding.
There are AI systems that can identify a new problem and ‘train’ itself on how to solve the problem but how to identify and ‘train’ itself was also programmed by someone. The propaganda gives people the impression that AI systems can think independently like humans and once up and running can autonomously evolve like Skynet in ‘Terminator’. Any deviation from what we expect it to do is not autonomous evolution, it’s a bug that was introduced by the developers. At the end of the day an AI system is just a bunch of hardware and software doing exactly what we programmed them to do. Sometimes it works really nicely, sometimes it’s complete shit and hinders problem solving rather than helping. But that’s exactly what you would expect from a bunch of nuts and bolts so it’s actually working as expected if you have realistic expectations.
AI is a nifty new technology that has its niche` and its uses but that’s all. The concerning part about it is that, like all new technology, the powers that be will find a way to weaponize it and use it against humanity.