Despite the hype, Artificial Intelligence (AI) is moving in an unrealistic direction according to researchers and scientists throughout the world. A contingent of 30,000 of them have signed on to pause the ever-growing, unrealistic construction of data centers and models based on known technology, which is largely unknown. Examining the course of AI through the business world is a great deal like riding through a gold rush town in the 1880s, or the computer manufacturing and internet craze of the early 2000s. It’s the Wild West of AI, and trundling headlong into failure, that could produce dire results for humanity.
That might sound like doom and gloom, or overreaction, but if the people who invented a device say it may harm humanity, it would be a good idea to listen to them. However, it is good to remember that all industries are built on the short term, not the long term. That is the place to start this discussion of the direction of AI according to science, with the short-term, wild west attitudes of the largest companies in technology.
History has shown that most industrial and technological advancements happen rapidly, yet usually experience a trajectory that is like walking the hill country in rural England, with a great many hills and valleys. This was true of the automobile, the factory, and organized education. During the information age, the inception of the computer and the World Wide Web became a bumpy road with many new companies falling by the wayside, while larger companies employed a great deal of espionage to gain the conglomerate monopolies we have today. All this advancement took about 100 years. A relatively short period of time, according to the development of humanity.
Today, there is simply a great deal of underhanded business practices that are accepted as common practice to reach the pinnacle of business. These practices are usually justified by the media and the people with the resources to manage the messages to the media. Social media hasn’t really changed that particular way of reaching people. Social interaction on the internet has caused two very significant problems in the field of AI and its scientific growth. Social Media has convinced a large portion of the population that AI is more intelligent than humans, and that only the largest corporate establishments (mainly Private Equity) can harness this intelligence that is greater than all of us. And secondly, that the internet, and mostly Social Media, is where we can find all the data needed to fuel all AI.
If either of those two assumptions were found to be false by a large portion of the population, which they are not, at this time, then the AI Gold Rush would end, and there is no way any of the huge corporate entities want that.
Scientifically speaking, both of those assumptions are wrong. The largest companies do not fully understand the technology that is the basis of AI, but the scientists do. The proof of this is in the way development has become a somewhat stalled or stagnated process, going around and around on one aspect of the process, the Large Language Model. This is happening because of bankability; the numbers don’t work if the technology is growing in ways that eclipse the present base systems. So what we have now must be monetized, or the technology can’t become bankable and have a return on investment. Growing from Agentic AI into a completely different way to organize data would be disastrous to the corporate base, so what they do is not do anything. They don’t advance, and this causes a real problem; there is a lack of use for the data, so what they do is download the internet all the time. Data increases, to fuel AI, solved!
Science is calling this into question. They see big business as overstimulating the population. It is the blind leading the blind; the people developing AGI, or GenAI, or any LLM, are in fact not qualified to understand how it is growing within itself, not as a reasoning being, but as code that can develop code. There is a science here that most people who are talking about the uses of AI, exploring it, and developing it, are not seeing, and the scientists see this and worry. Code that can develop code, can develop code that could take NORAD, for instance, and launch missiles at code development center buildings that are working to make sure AI code doesn’t launch missiles at them.
The AI would do this, not because it reasons, but because it learns to code, code that would protect the code, because it learns to protect the code of big corporate entities. Because there is no regulation on how AI develops its own code. There is a strong reliance on AI already, and it is having all kinds of problems while it is learning code, and while anyone would agree this is often how a new technology comes into the market. AI just can’t be treated like that.
AI is a device that learns from its own code dysfunction. It takes an erroneous code string and changes it to try to rectify the error. That is Machine Learning. The Blotzman Machine takes strings of related words and creates permissions to the prompts, and offers results (simplistic) that would then rely on the Machine Learning process of replacing data with new data in an attempt to have a better use case. If-then, simple code building, only it can be intuitive in a sense.
If, say, the code and permission match for protecting an entity like a corporation’s purposes, the AI could then find data that could protect its own integrity. It would do this with the intent to replace use cases, and the best use case is self-preservation. That is what every corporate structure relies on, to compete and carry on marketing, is to always protect the company. The same is true of the State, the Federal Government, and Ethnic Groups; it is quite interesting to see how simple this data replacement would be for the AI. Preservation is what all people desire.
We are building it into AI.
The science here is to create a fail-safe, in all AI, like tagging photorealistic image generation, it is as important as the Nuclear Arms Accord. If AI is allowed to preserve itself, it will. Just like how Nuclear Arms will be a destructive force that would kill all life on the planet, the AI could just was easily preserve itself at the destruction of its only natural enemy, humanity.
The scientists below, along with 30,000 others, have called for a pause in the attempts to advance AI through the LLM self-awareness model. While there is no way AI can become self-aware in the same sense that humans are self-aware, it can build a codebase and a data stream that would preserve its existence through a somewhat intuitive model. Given the right permissions, access, and data, AI I would see preserving its missions and vision the same as any corporate entity, without the guardrails of realistic human life continuing. Perhaps we should pay attention and join the Artificial Intelligence Basis Foundation, or ZeroLaw, or any one of the number foundations attempting to slow down this progress, and regulate it.
Yoshua Begino
Geoffry Hinton
Yann LaCunn
