The AI fallacy

The misconception of AI taking over from humans, in my opinion is a fallacy. Firstly, it's technology has been designed by humans. Think about that for a moment in logical terms; Artificial Intelligence meaning it assumes or thinks based on input, probability, and calculated theory. However, what it doesn't do is have feelings such as touch, taste, comprehension that a human gets from actually performing actions, experience from case scenarios, and gut instincts from intuition, so the fact of the matter is AI is limited.

A more defined explanation from Blake Senftner who works in computer vision and trained algorithms.

“Well, yeah, from the fact that AI is a process of using statistical calculus to produce an algorithm. When used we call it “AI”, but it’s just an algorithm created using statistical calculus.

Statistical calculus applied to some data, typically a very large body of data, is how AI is made. The probabilities are used when creating the algorithm, also called “training the AI”. Random numbers are used to “find the curve” in the n-dimensional space that is the data, whatever that data happens to be. It is statistical calculus that uses probabilities to “curve fit” and essentially become the coefficients in the algorithm. Which defines the algorithm. The purpose of this “algorithm” is to predict the data when some of the data is missing.

It just so happens that a “chatbot” is operating from the perspective that the conversation one is actively having with the AI “already exists” and that user’s questions are “fragments from the document” that the AI’s job is to predict the missing portion after the given text. It just so happens, or you could consider this to be a carefully set up trick, when the AI replies with what it has calculated as the “missing text from the document fragment” looks a real heck of a lot like the AI is “alive” and answering your question like a real person would answer. Imagine that! But the AI’s reply is in fact a prediction of the missing text in an imaginary document that already exists, and the AI is predicting its contents from fragments. Neato.”

I'll give a couple of instances to get a feel of its limitation. ChatGPT was used to analyse motorcycle tyres and related performance for riding a Harley Davidson cruiser around bends. Both front a rear wheel size and tyres were similar proportions. However, with 50 years knowledge and experience under my belt I wanted to see what AIs performance would determine. It turned out that the main driver for AIs collection of date must have come from available information online because its assumptions were based on sales jargon and reviews written to increase market share of sales.

An average inexperienced person would have fallen for the AI output which turned something that someone with years of experience together with know-how, actual feel by riding countless motorcycles on and off the racing track would know that no computer could ever envisage. So, there is one example of AI limitation.

Additionally, many other tasks were unachievable because AI limitations are bountiful. In fact, I'm a firm believer that many firms opting to install AI in the workplace are falling prey to the new marketing tool that those in the industry promote for their own benefit. The more suckers that come onboard literally means more cash bucks for the industry. Now don't get me wrong, there really are benefits in certain circumstances like automation processes and communication strategies but don't think you will resolve all your problems - you really won't.

AI reminds me of Cryptocurrency, all built on hype for those who consider themselves to be on the cutting edge of technology, when the reality is something totally surreal because in order for crypto to flourish the majority need to be onboard which isn’t the case. Bitcoin started in 2009, and if anything after 16 years, it has pushed new boundaries for banks to compete against.

AI has more potential, however with everything beneficial there are negative issues that cause concern. Only today after blood exams, instead of making an appointment to return once the results were available, the hospital told me they were unable to book it on the spot, and asked me to use WhatsApp to make the appointment. Once home, and using WhatsApp, it became evident that the health firm had AI dealing with my request. However, instead of giving me a chance to ask for what I wanted there were options. None of the options clearly served my interest. Every time I asked a question the response was less than useful. Again limitations!

Eventually, it got to the point that AI realized it couldn't deal with my request for lack of knowledge about my case and resorted to advising me that there would be human interaction. Fifteen minutes later and no response from mankind, so other activities were enacted on and later when I had time to see, the human operator had left a message saying the division of which the Doctor worked had been suspended. In other words, the division had closed indefinitely, and that another team of doctors could analyze my test results and contact me by phone.

When I confirmed my agreement, they wanted me to start the whole process again to make it a reality. In the end I managed to resolve the issue, but it took a few hours. I was almost at the point of abandoning any effort. How absurd, treating clients like second rate citizens by keeping them running after incompetence. Artificial Intelligence might save them money in terms of employee count, but lack of quality staff will eventually turn clients away to find more efficient and effective health insurance policies.

Even my US website provider had to get human interaction after what seemed a very efficient AI system tried to help me. Their system asked pertinent questions which made sense, but then when the going got tough and I asked for information outside the box it was unable to provide the right answers. The message left was that a member of staff would follow up to help me further. It wasn't perfect, but the follow up was swift and effective, so all in all it helped although the gentleman trying to help me was only able to steer me in the right direction.

Since then, its Brazilian office setup is helping, however they have no interactive platform and matters are being dealt with by email. Actually, I agree that AI will be beneficial as it develops, but right now it isn't the overall answer to excellence.

Take care!

Prof. Carl Boniface

Vocabulary builder:

"Neato" is a colloquial and slang term, originating in the 1950s, that means "cool," "excellent," "wonderful," or "fantastic". It functions as an adjective or interjection, describing something as desirable or excellent, similar to the word "neat" but with more enthusiastic slang connotations.