Welcome to the yellow pages for
The RUN-EU Erasmus-BIP
The Digital Mirror.
Here you'll find all materials to the lectures and workshops on the metaverse, avatars and virtual idols.
More stuff is yet to come ;)
In the modern technological landscape, we are often told that Artificial Intelligence is a "black box"—a mysterious, opaque engine of neural networks and probabilistic soup. While these systems are impressively dynamic, they lack the architectural blueprints that allow a creator to truly see and shape the soul of a machine. As a digital architect, I find a unique power in the elegant, rule-based simplicity of AIML (Artificial Intelligence Markup Language). It allows us to move away from statistical guesswork toward intentional design, making a machine feel like a conversational partner rather than a mere calculator. While modern AI thrives on the unpredictable, AIML offers a system that is transparent, traceable, and entirely under your command.
The core difference between AIML and traditional modern AI is the shift from probability to certainty. Most modern systems use statistical models to guess the next word in a sequence. AIML, however, is an XML-based language where every response is explicitly defined by a set of rules. For a creator, this is not a limitation-it is a superpower. In a world of "hallucinating" AI, rule-based logic provides a predictable anchor for virtual assistants. This makes the responses transparent, traceable, and easy to control. By using a structure of categories, patterns, and templates, you ensure that your "digital mirror" says exactly what it is intended to say, creating a reliable foundation for any human-machine interaction.
One of the most elegant tools in the architect’s kit is the wildcard, represented by the asterisk symbol. This placeholder allows a single rule to recognize an infinite variety of user inputs without the need to program every possible sentence. The real magic happens with the star tag, which captures the specific text the user typed at the position of the asterisk and reuses it in the response. For example, consider this logic:
I LIKE *
Why do you like
Human conversation is defined by variety. If you greet a friend the exact same way every single day, the relationship feels mechanical. To prevent a chatbot from sounding like a broken record, we use the random and li tags. By defining multiple potential answers for a single input like HELLO, we can inject a sense of "personality" into the code. Instead of one static answer, the bot can choose between "Hello!", "Good day!", or "Nice to see you!" at random. These small variations in greeting drastically change the user's perception, shifting the experience from a repetitive calculation to a dynamic interaction that feels fresh and human-like.
A conversational partner needs a memory. AIML facilitates this through variables like "username" using the set and get functions. As a digital architect, I view the think tag as the "curtain" of the performance. It allows the machine to process information—like saving a name—in the background without cluttering the immediate response to the user.
By using the think tag, the internal processing remains invisible. Later, when the user asks, "WHAT IS MY NAME," the bot uses the get tag to retrieve the data. If the user previously said their name was PETER, the bot recalls it perfectly. This creates a continuous thread of memory, turning isolated exchanges into a cohesive relationship.
The greatest challenge for a simple bot is ambiguity. Without context, a bot cannot know what the word "IT" refers to. As a digital architect, I solve this by "designing the environment" using the topic tag. This allows us to build different "rooms" of conversation where the same question yields different meanings based on the space the user occupies. For instance, the question "WHAT CAN IT DO?" is fundamentally confusing to a bot without a topic. However, by grouping rules into topics, we provide a solution to this unpredictability:
Mastering these basic building blocks—explicit rules, wildcards, randomness, memory, and context—allows any creator to "Build their Idol" and craft a digital presence with surgical precision. As we move forward, we must decide how to balance the "unpredictable magic" of neural AI with the "intentional design" of rule-based AIML. By choosing to build with transparency and control, we don't just create a tool; we create a sophisticated Digital Mirror that reflects our own vision and logic. Which version of ourselves will we choose to reflect?