ARTIFICIAL INTELLIGENCE – REALITY OR ILLUSION?

And what if, in reality, the Artificial Intelligence (AI) companies are just playing with words? And they use the word 'intelligence' to denote something we would call it otherwise?

Our thesis

Ordinary people and scientists do not attribute the same meaning to the word 'intelligence.' Furthermore, even among scientists, there is disagreement, as they understand intelligence differently depending on whether they focus on a 'human' or 'real' profile. However, companies producing sophisticated software do not concern themselves with such nuances. They boldly declare their programs as 'artificial intelligence' and sing its praises. But is this truly about 'intelligence'? And if so, why? We will delve into this further in the following discussion.

Our Categories of 'Intelligence'

Warning: We will not reproduce here either the general definition of the word 'intelligence' or its categories as described in dictionaries or scientific works. We fully acknowledge that we are not complying with established rules on this matter, just as we will do for everything that follows. The opinions expressed here are ours and not those of a scientific article. We recognize that we cannot compare ourselves to scientists, who nowadays are as skilled with words as the best lawyers in Paris. Therefore, our categories of 'intelligence' are:

1. Mathematical or Logical-Mathematical Intelligence (MI): Based on calculation formulas, algorithms, and probabilities.

2. Human Intelligence (HI): This encompasses the 'intelligence of the senses,' which utilizes information provided by the five senses. It guides our actions and reactions necessary for survival, self-protection, and species perpetuation. It is also referred to as 'survival instinct.' For us, it is a form of 'intelligence' because it is not mechanical or automatic; rather, it is inherent to our nature as living beings, involving understanding and acting to preserve our existence.

3. Additionally, we have 'Emotional Intelligence', which is based on our feelings, experiences, and thoughts. This form of intelligence enables us to create and enforce societal rules of coexistence, such as laws, morality, and good manners. Together, these elements constitute HI.

What kind of 'intelligence' is Artificial Intelligence?

Let's be clear: it started with deception. John McCarthy, the inventor of the term 'artificial intelligence,' described it as akin to 'human intelligence' but without its 'biological' constraints. Marvin Lee Minsky, another AI pioneer, straightforwardly stated, "AI must do better in the future at everything people currently do better than it does." Consequently, today AI is presented and 'sold' to us as a sort of magic wand capable of solving any problem and providing answers to any question we may have. This is misleading. We argue that AI is a form of MI, at best a feeble imitation of HI. It has no intention of performing the same tasks as HI, let alone replacing it, except on a superficial level.

Why Artificial Intelligence Couldn't Replace Human Intelligence?

Once we strip away the pure advertising in front of our eyes, we realize that AI is simply a series of algorithms (mathematical calculations). Emotions, feelings, instincts—the essence of 'living'—are and will remain unknown to AI because it lacks a body and the experience of life itself. This fundamental difference with Human Intelligence (HI) arises: AI possesses vast knowledge but lacks understanding of what to do with it. Therefore, AI initially requires human intervention to learn how and for what purposes to utilize this knowledge. Even in its hypothetical second stage, where it might surpass the first, human involvement is crucial for fine-tuning. While AI can potentially create something new, humans predominantly perform the groundwork.

Slaves to Artificial Intelligence? And yet, how...

Let's be clear: behind the increasingly 'spectacular' performances of AI (whether true or not), there lies the hidden labour of tens of thousands, if not more, 'slaves'—people from poor countries paid almost nothing (see the article 'In Madagascar, let's meet the real slaves of AI' on www.franceinfo.fr; others just a few clicks away). The result of their work is an enormous database of 'standard answers', from which AI extracts the one that seems best suited to our requests. And for more complex tasks, the process is even easier and cheaper: computer science students are utilized as part of their educational training. They create the algorithms that enable AI to recognize and utilize emotions and feelings in texts or images (refer to 'Computer science schools: foundation for the generative AI revolution' on www.lemonde.fr; other sources are available online). And how do you think AI companies responded to these revelations? They argue that their programs are merely 'trained' in this manner; once training is complete, they supposedly develop autonomously, producing more refined texts, images, and music without external assistance. They deceive us outright; it's akin to an 'athlete' running and winning a marathon after driving a car up to the 38th kilometer. What happens after this 'training' phase?

Achievements of Artificial Intelligence

It is important to understand: AI generative models (referred to as 'public' in the article 'AI: Dream, Nightmare, or Scam') can sift through a multitude of 'standard answers' to find the one that best fits the given request. Once it locates the answer, AI adds its own generated content to create the impression that it has actively worked for you. This 'extra' is tailored based on the information it has gathered about you—your habits, hobbies, and all the data collected through your internet use such as social networking, online shopping, Netflix preferences, news searches, or email correspondence. As you interact more with it over time, AI learns more about you and becomes better at providing answers that are increasingly tailored to your needs and style. This is how AI operates in the public sphere.

However, the situation differs for AI used in 'private' settings by companies in industries or services.

Different... how?

On one hand, we have technological processes and administrative (or bureaucratic) procedures that are standardized and regulated; they are clear, logical, and leave no room for fantasies or 'slips'. This environment is ideal for AI, where everything hinges on computing power—how many operations machines (computers, servers) can perform simultaneously. For an AI program to excel here, both the software and hardware must be robust enough to store and analyze vast amounts of field-specific knowledge, techniques, processes, and procedures in an ideal scenario. The stronger they are, the better they perform. Here, nothing or almost nothing is accidental or unforeseen; every event or outcome is a logical and predictable continuation of the process or procedure in question. However, don't expect AI to invent something entirely new; at most, it can innovate but within limits—only in the logical extension of existing and proven concepts.

Conversely, in management, uncertainty prevails regarding the capabilities of AI.

Why 'uncertainty'?

Because Human Intelligence (HI) is involved to a certain extent. Furthermore, there isn't yet enough information available to create the standardized response 'blocks' that AI needs to be 'trained' on. This is why companies like OpenAI, Google, and Apple are entering the race to acquire them, aiming for even more refined versions of Siri. This explains the frantic race to develop the most efficient 'voice assistants'—to accompany CEOs, directors, and executives (the ultimate target audience). Through their use, AI companies gain access to information on how decisions are made within an enterprise. Once they have this data, they can replicate the model used for AI in the public sphere. However, merely creating the database won't suffice. The 'training' of AI will require more qualified individuals than the simple 'click workers' from the third world or computer science students. It will require people who comprehend how decisions are made and operations are managed within a company, at all levels. Currently, there's uncertainty whether AI can soon take on the role of an assistant, particularly in tasks involving understanding and interpreting the nuances of human language and behavior—such as summarizing the details of a sensitive negotiation. Nonetheless, AI companies and experts assure us that they are diligently working to find solutions.

What solutions?

Other, even more sophisticated programs aim to replicate the human brain and its internal connections. These programs mathematically encode Human Intelligence (HI), with the goal of creating AI capable of thinking and acting like humans, understanding gestures, actions, attitudes, emotions, feelings, and reflexes. In this regard, programs such as Natural Language Processing (NLP), which enables machines to understand, use, and generate human language, and Large Language Models (LLMs), a more advanced form of NLP, have been developed. It's certain that more similar programs will follow soon. However, these programs are limited to 'basic' tasks. When it comes to more profound aspects of human nature, the claim that they can be imitated or reproduced is false.

What do you mean?

We claim that AI could explain, anticipate, and replicate human behavior as usual. Yet, we still doubt whether AI can always be correctly 'trained' to account for local specifics, such as race, sex, or religion. However, AI cannot understand or predict when a person decides to suddenly change their behaviour, become unpredictable, or base decisions on intuition or a 'sixth sense' instead of logical reasoning. Simply put, AI, being Machine Intelligence (MI), cannot replicate Human Intelligence (HI).

The Intelligence of Man

We understand 'intelligence' in our own way, not necessarily aligned with dictionary or scientific definitions. For us, intelligence combines MI with specific elements of HI and practical adaptability ('getting by' within societal rules and laws). This is something AI cannot currently achieve and may never achieve, as it involves unpredictability and illogical behaviour. Yet, companies promote and 'sell' AI under the promise that it will soon replicate these human capabilities.

Why do they do that?

To make money off us, of course. AI companies and investors are in it for profit; they aren't motivated purely by altruism. To achieve this goal, they aim to make us reliant on AI and then charge for access to their services. This tactic didn't quite work with social media—we love using it, but as soon as we're asked to pay, we switch to a free alternative (as seen with 'X' after Elon Musk took control of Twitter). Consequently, some may believe today that AI will eventually accomplish everything, making it impossible for us to work or live without it. They might even think AI will one day dominate the world in place of humans, dictating our existence. However, it's crucial to realize that AI only operates based on what we teach it and responds only to specific requests. It exists in both reality and illusion simultaneously.

Reality and illusion

We believe that AI will only be able to replace humans in terms of Machine Intelligence (MI). The changes attributed to AI entering our lives have been minimal so far, but they are set to increase dramatically in the coming months and years. Its capabilities will expand significantly due to advancements in both powerful hardware and software (with associated energy consumption, which will be discussed in a future article of this feature), and also because it will increasingly rely on more qualified 'trainers' (primarily in terms of human behaviors and actions). By combining these elements, AI will indeed provide increasingly better responses to human needs and will autonomously improve (still within the realm of MI, i.e., what is most likely or logical to occur). From this perspective, AI meets the criteria of 'intelligence'. However, when we move beyond this realm of mathematical logic, its 'intelligence' reaches its limit. This is because AI cannot comprehend or replicate human actions and thoughts that deviate from the 'most probable' path. Concepts such as 'arrangement', 'shortcut', and intuition operate on a separate track from AI. In comparison to these, its 'intelligence' is illusory—a mechanism designed to persuade us to use it more frequently and, above all, to pay for it. Yes, let's utilize AI, but let's remember that ultimately, it remains just a tool!

The necessary tool

One that is perhaps becoming increasingly essential. Its importance depends on how much we value it, but we must not forget that AI can also be used against us. It is not infallible; it can be easily overcome when we choose to act differently than it expects or deems 'logical'. Therefore, our advice is to use it to our advantage as much as possible.

And the future?

Let's be clear: everyone is diligently working to develop AI, and there are no plans to halt these efforts. The Americans and, to a lesser extent, the Europeans currently lead in this field and will likely continue to do so. Some individuals in states hostile to the West—scientists, business leaders, and propagandists—create panic to frighten us with the idea that AI will surpass humanity, relegating us to second place and seizing power. Their proposed solution is to slow down or halt AI development, but this sentiment predominantly exists in the West. Meanwhile, countries in the Global South are intensifying their efforts. The outcome of choosing this path remains uncertain, albeit it seems unlikely to occur.

Most likely scenario: 65 – 70%

Certain countries in the Global South will likely surpass us technologically with the aid of strong AI, asserting their dominance over the rest of the world, completely and ruthlessly.

Less likely scenario: 25 – 30%

Countries in the Global South continue to develop AI, but without advancing sufficiently to dominate the West.

Unlikely scenario: 0 – 5%

Countries in the Global South fail to replace the creative capacities of the West. Progress halts, and AI remains merely a project, something whose fulfillment will be postponed to another time.