April 28, 2024

The word "intelligence" should not be associated with machines or silicon. There are motives in defying that fact; but it also takes ignorance to assume anyone would miss the obvious.

Methodology shows the truth. Algorithms are not a methodology that replaces truth and awareness. Using algorithms means a procedure follows a pattern defined by the limitations of a programmer who learned how to connect processes.

Adding the word "generative" is supposed to mean the computer adds something to human intelligence. It's total fraud. All the junk does is create complex averaging of scraped garbage directed by algorithms.

There is a high degree of misrepresentation in promoting the results. A data set needs to be available to be scraped. The only significantly available data set is literature-type material stored on computer servers. Other types of material are not suitably available to be scraped. No concept of scraping scientific knowledge exists.

Without the scraping and averaging, computer processes add nothing but old fashioned automation. Automation does a lot of things humans cannot due; but it is not addition of awareness.

Yet the "generative" AI is supposed to be similar to, if not superior to, human intelligence. Showpiece gimmickry is used to impress persons who have nothing better to do than entertain themselves with the worthless absurdities. Impressing with machines is nothing resembling intelligence. A steam engine does impressive things that humans cannot do.

Where the folly gets more relevant is in claiming computer processes are superior to humans in driving. One of the most basic frauds is the pretense that measurable events can be superior to human awareness. The ability to measure events is disgustingly inadequate for replacing humans in driving.

The first problem with "self-driving" is the grossly inadequate detection devices. To match existing technology would require covering a vehicle with devices, while the purpose is to impress the public with something that looks like a normal automobile.

Consider optical cameras. To get suitable resolution on objects with a camera would require "close up" lensing. But close up lensing has a narrow field of view. That means a lot of cameras would be needed to detect what is happening outside the field of view of each one. How many? To encompass what humans see would require dozens or hundreds of cameras.

Then to have a computer evaluate the images while in motion would require about twenty to fifty frames per second, because a lot can change in a few milliseconds with a moving auto. That means, each frame needs to be evaluated in relationship to everything concerned with driving in milliseconds while comparing algorithms which determine what factors are relevant in the changing images.

Besides the unimaginable amount of data storage and speed required, there are impossibilities to the process. One is the inability to evaluate changing images to determine the meaning of differences in images. When a spot such as a light is in a different location, as motion occurs, is the different location in the image due to a different spot or the same one moving? Humans can determine what the same images are while in motion, but computers cannot, because motion is not continuous with computers, as it is for humans. Motion is broken up into discrete frames with computers.

You might think a good guess is close enough for a computer. But what about flashing lights? When a light was off and comes back on while camera images change drastically in that amount of time, the relationships are lost between images which are twenty to fifty frames apart with different backgrounds.

Humans do continuous evaluation of gradients. Computers do not. Gradient braking is the simplest example. Humans feel the effect of braking with continuous adjustments. No machines can do anything resembling that.

To program braking, a determination must be made for the amount of force required. But braking does not have a definable force. The force is different every time a brake is touched. That's because of variations in velocity and weight, which change the amount of inertia that determines the force. Then there are variations in opposing forces due to the nature of the brake linings and tire friction which are constantly changing due to wear, grime and moisture. With all of those factors constantly changing, a programmer cannot come close to determining in advance what gradient braking should be.

Then there is the problem of evaluating situations. Ask a programmer what driving situations consist of. What sort of god is he supposed to be? No two driving situations are exactly the same. There is no such thing as a definable driving situation. That's why self-driving is little more than following a white line, while any aberrations in the line can be deadly, as has occurred.

And the public is supposed to fantasize that result as the wave of the future? Only to keep incompetent power mongers in control of our lives, while they impose futurism onto society in place of identifiable realities that determine real solutions to existing problems.

The recent version of so-called artificial intelligence is a shift from relevance to nonsensical word salad, which is the trend in bureaucratic and administrative processes, as incompetent power mongers take over the social structures and convert them into power structures for their personal gain at everyone else's expense.

That's why power mongers glamorized artificial intelligence and poured endless resources into it. But the pretense of a result fell flat in every area. So the rationalizing required pretending that the age-old automation technology was being improved. But that pretense also fell flat, as automation technology is vastly different and was already pushed to its limits.

Automation technology differs from so-called artificial intelligence in that the first requires exact precision evaluated by human input, while the second rounds off everything to a point of irrelevance by scraping and quasi averaging descriptive literature which is designed to be generalized. Non-generalized material, such as scientific publications, cannot be scraped and processed by the useless methods of artificial intelligence.