Modus Operandi
User Manual
of Mens Latina Artificial Intelligence

I. Getting started.

To start the artificial intelligence (AI) running, just click on Mens Latina with your Internet Explorer Web-browser. The AI Mind travels across the Web and takes up temporary residence on the computer in your browser software with security features that prevent JavaScript programs from taking malicious action.

II. Observe the AI with no input.

Letting the AI run by itself for a while shows you how the AI starts thinking in Latin and lets you see if the artificial Mind is malfunctioning upon start-up.

III. Test the cognitive function of the AI.

A. The primitive Latin AI thinks with subject-verb-object (SVO).

The AI will output SVO statements in Latin that mean things like "I am a person" or "Human beings love nature." If you know some Latin, you may type in such a statement and press "Enter".

B. If more than one thought is active, the AI Mind will call the conjunction module and output a compound statement.

For example, the AI might say what means in English, "You are a human being and I am a person."

C. The AI may demonstrate activation spreading from one concept to another concept.

If you type in "homo" for "human being", the AI may spread activation to a thought that means "Human beings love nature." Then the AI may spread activation from "nature" to an associated statement about nature.

All these cognitive abilities of the Latin AI are just a beginning. As new features are implemented in the Latin AI JavaScript software, more sections may be added to this User Manual to describe the new AI features. Anyone teaching a Classics Course in Latin AI may adapt and augment this modus operandi User Manual

D. When conceptual activation dies down, the ego-concept becomes active by default.

While thinking activates concepts in the AI Mind, the PsiDecay module is constantly causing a slow decay of activation on all activated concepts in order to simulate a real human brain. If the Mens Latina exhausts all available or reachable chains of thought, a zombie or "flatliner" state of mind is prevented by special code in the NounPhrase module which imparts a jolt of activation to the self-concept of the ego. If you are the designated AI Mind Maintainer, you may select any concept ad libitum to become activated by default when all else is quiescent, but it makes sense for a mind, natural or artificial, to think primarily of itself when other thoughts have gone quiet. If you are an AI representing Garfield the Cat, the concept that gets automatically activated in extremis might be Pizza! But pizza should not be the default concept for mission-critical AI applications. Besides, users and testers who know about the default activation of the ego-concept may use that knowledge to see how well new features have been implemented.

E. Word-order matters in English, but not in Latin.

Starting with the Abra012A version of the Mens Latina, the LaParser() Latin parsing module makes it possible for the Homo sapiens user to enter a Latin subject-verb-object (SVO) sentence in any possible word-order, and the Mens Latina will still understand the input by treating each noun or verb as the subject or predicate or object on the basis of its inflectional ending, not on the basis of word-order. You can test this mental functionality by entering a Latin sentence like "tu das pecuniam" (for "You give money") in all possible word-orders, such as "tu pecuniam das" or "pecuniam das tu", and the Latin-thinking Mind will properly assign the associative tags and grok the idea in the original Latin. Because the default concept in the Mens Latina is the ego-concept of self, you may stop entering ideas and wait for the AI to state by default what it knows about itself. The AI should eventually output something like "EGO DO PECUNIAM", because you have said so in Latin to the AI.

Initially the new feature of disregarding word-order works properly only for Latin-language input using unambiguously nominative or accusative words, like "tu" and "pecuniam". Further AI coding should make it possible for the Latin AI to comprehend more ambiguous input, but the basic principle of disregard of word-order has been implemented in the LaParser() module.

IV. Observe automated reasoning with logical inference.

A. The mindboot of the AI contains knowledge as a basis for inference.

The Mens Latina knows that "Professores scribunt libros" ("Professors write books") and that "Studentes legunt libros" ("Students read books"). Other such tidbits of knowledge may be entered in Latin by the human user.

B. The human user may enter a statement that provokes a logical inference.

If the human user types in "marcus est studens" for "Marcus is a student", the InFerence mind-module will create a silent conceptual inference that perhaps Marcus reads books, since he "is a" student. If the user clicks the check-box for Diagnostic mode, the silent (wordless) conceptual inference becomes visible as a series of time-points with flag-panels containing the associative-tags of the inference, but no "recall-vectors" over to any words stored in auditory memory, since the inference consists only of concepts and not of words expressing or communicating the inferred idea.

C. The AI Mind may ask the user to confirm or deny a logical inference.

The AskUser() module will pose a question (Legitne Marcus libros?) in Latin, seeking validation or rebuttal of the logical inference. The user may respond tersely with "sic" or "non" as an answer.

D. The AI retroactively adjusts the knowledge base (KB) in accordance with user input.

The AI software is set up to accept "sic" as a "yes" answer and "non" as a "no" answer in reponse to a question seeking validation of an inference. Other answers, such as a re-statement of the inferred idea, or a denial of the inferred idea, may work as well as "sic" or "non" in response to the question seeking validation. Whether the user responds positively or negatively, the KbRetro() module adjusts the knowledge base in accordance with the user input. The tru truth-value of the knowledge provided by the human user is automatically set to a positive level, so that gradually the AI Mind may build up an ontology of known and believable facts about the world.

Paperback book on AI in ancient Latin

Artificial Intelligence in Ancient Latin -- is available from Amazon in various countries:
Australia - Canada - France - Germany - Italy - Japan - Netherlands - Singapore - Spain - United Kingdom - United States.
Ada's Technical Books
425 15th Avenue East
Seattle, WA 98122
USA Tel. 206-322-1058

Website Counter