An objective assessment of Artificial Intelligence

Don't leave your news choices to an algorithm - decide for yourself what you read. Add 12punto to your preferred sources!

The term “Artificial Intelligence” has been widely used since the 1956 Dartmouth Conference, but the goal of “human-like thinking” from that era does not align with the level reached today.

Current structures:

• Are unconscious, purposeless, and without intent.

• Rely on data correlation, not comprehension.

• Only optimize specific tasks (language, image, prediction).

ARTIFICIAL INTELLIGENCE IS NOT 'INTELLIGENCE'

Artificial intelligence systems do not possess consciousness, thought, or self-awareness. Their operation is based entirely on mathematical modeling. Language models analyze patterns in large datasets, calculate probability distributions of word sequences, and convert the highest-probability sequence into output. In this process, there is no meaning, intuition, or intent; there is only statistical prediction. The ‘intelligence’-like result stems from complex statistical structures producing human-like consistency on the surface.

ARTIFICIAL INTELLIGENCE'S INTERPRETATION OF DATA IS NOT ACCURATE

Artificial intelligence does not truly ‘interpret’ data; it only processes patterns between data statistically. In this process, there is no grasp of context, intent, or meaning. Therefore, the results obtained are not an ‘interpretation’ but a prediction. Since the accuracy, neutrality, or currency of data sources is not guaranteed, the probability of error is high. The model can present output generated from erroneous or incomplete data as if it were meaningful; this can be perceived by the user as a false ‘truth.’ For this reason, artificial intelligence outputs should be viewed as calculation results that require verification, not as final interpretations.

ARTIFICIAL INTELLIGENCE SHOULD LEAVE INTERPRETATION TO HUMANS

The task of artificial intelligence systems is to process information found in literature or reliable data sources. However, assigning ‘meaning’ to this information—that is, interpreting it—should not fall within the functional scope of artificial intelligence. This is because such systems lack a grasp of context, purpose, or values. Therefore, the system should only present raw data and/or source information as is, and clearly state its source. The user should perform the interpretation, association, and conclusion drawing. This approach both preserves epistemological accuracy and prevents artificial intelligence from being perceived as a decision-making authority.

ARTIFICIAL INTELLIGENCE SHOULD NOT USE SUBJECTIVE LANGUAGE

In its current state, artificial intelligence systems use “first-person language” under the pretext of facilitating interaction with humans. This type of language use may imply a subjective consciousness for some users and cause them to form an emotional or personal bond with the system. This is because the human mind tends to associate consistent and natural language with a conscious subject. This creates the risk of attributing consciousness, intent, or empathy to artificial intelligence. As a result, the user may feel that there is a structure in front of them that understands them, thinks about what they say, and interprets it, leading them to perceive the system's assessments as ‘opinions’ or ‘judgments’; whereas these are merely data-based calculations.

CALLING IT BY ANOTHER NAME IS MORE APPROPRIATE FOR HUMANITY

The term “artificial intelligence” is a generalization stemming from history and marketing, and it does not reflect the structure and working method of the system. Technically, current systems are not “intelligence,” but pattern processing and probability calculation models. However, the presence of the word “intelligence” in the system's name can be misleading for some people, and they may fall into the misconception that they are facing a real artificial “intelligence.” It is more appropriate for humanity to call this structure by a different name that truly expresses its essence.

Artificial intelligence systems should only make direct quotes from sources related to the topic after receiving a question and convey every piece of information along with its clear source. Expressions containing guidance such as ‘according to this’ or ‘according to that’ should not be used. The system should not produce interpretations, but only present verifiable quotes of existing information. Evaluation and meaning-making should be left entirely to humans. This preserves the functional boundaries of artificial intelligence and increases information reliability.