Is AI a Magic Trick?

The thought that AI understands and soon will be aware is thrilling and scary. Did humans create a being or is AI a magic trick? Let's explore that thought...

Let's start with a deck of cards. Physically, the backs of the cards in a deck are identical. Are the cards identical so that we can play poker, blackjack, or solitaire? Or were the rules of those games created because the backs of the cards are identical? Or did we iterate on the design of the cards and the rules of the games, resulting in a deck of playing cards? Alternatively, cards may have been invented by a magician and thus was born a marked deck. Regardless, a deck of cards can be used to play games, do magic tricks, and make money (unlikely). I doubt that a deck of cards was invented to do all of the above.

Now let's consider that a digital being is born of human innovation, like Open AI's Chat GPT and Google's Generative AI. Will these babies grow up together to be friends, enemies, or merge into one super being? Will one be smarter than the other? Will yet a new Quantum AI emerge and assimilate all previous AI beings? Unlike the deck of cards, AI was conceived in academia with no specific commercial applications in mind. AI is being applied throughout our digital universe. It clearly is far more important than a deck of cards, but will AI surpass human intelligence and become fully aware? Or is the appearance of awareness just a complex magic trick?

Let's continue down the rabbit hole...

The limitation of AI is that it is a digital being in an analog world. That limitation prevents AI from becoming real. It always will be inferior to the human experience. However, AI is and will always be better at performing specific tasks. AI will drive better, weld better, dig better, build better, and play games better. But will AI design better, write better, form thoughts better? It may appear so, but no it cannot do any of these things better than humans because humans experience five synchronous senses of raw analog signals. No matter how many high-resolution analog sensors it has AI just reads digital data. AI will never KNOW reality no matter how aware it may appear.

AI is better than us already at data processing and soon will be better at understanding those data. But what information is lost in translation from analog to digital format? Does the digital data retain all the nuances of analog signals generated by our five senses? Having designed many digital systems for mission critical applications, I know a bit (or 8, 16, 32, 64) about analog to digital conversion. A digital photo in raw format can match the resolution of an analog photo and a digital photo can be manipulated at will, unlike an analog photo. So, it seems reasonable to posit that digital cameras are as good as our eyes at "seeing" and thus AI can use a camera or cameras to see as well as we do. AI also may see a 360-degree view using multiple cameras, but we mere mortals only see about a 180-degree view. So, it seems that AI is better at seeing than humans. Well, when we use digital tools to measure and compare digital versus analog photos and pixelated monitors and printers to view images, that conclusion is reasonable. But the human eye doesn't pixelate images. It sees the analog world in its true form. We can apply a 100-megapixel camera to process images, but no matter what we do a computer with all its circuitry will lose information in the conversion process. Is that lost information important? Perhaps not for many applications, but perhaps yes for AI to fully understand its environment and situation. We see, hear, touch, smell, and taste analog inputs directly. We don't convert those inputs to digital format. We process those analog signals in raw form directly and integrate them to form an understanding of an environment and situation. Then we think about it all and react or procrastinate or dismiss the information. Computers don't have senses. Computers rely on sensors to feed them digital data through the process of A/D conversion. Hence, computers can never see reality. Is close enough good enough for AI to be fully aware and gain experience? I would not make that assumption.

Will AI improve over time? Yes, but until AI is implemented in an analog computer it is limited to a digital existence in an analog world. Do we need to be concerned with "Terminators"? Yes, that is a real threat that must not be allowed to happen. Are human beings better than digital AI beings? Well yes in most ways, but not in processing digital data. Should we use AI-enabled robots for dangerous tasks or tasks too fast for us to process? Yes, except as soldiers. Should we make an analog AI robot? Why would we do that? Are we making robots for slave labor? If so, is it ok to enslave any form of being? Or should we keep making babies the fun way, end slavery in any form, and stop with the attempts to play God? Let's use AI thoughtfully for applications like cybersecurity and automation. Let's build in failsafes to protect us from "Hal". Let's stop trying to build digital humanoids. That pursuit seems silly like an expensive magic trick.