Page 240 - Kaleidoscope Academic Conference Proceedings 2024

2024 ITU Kaleidoscope Academic Conference

Copyright safeguards a broad range of creative expressions. Most of the AI we encounter today falls under the category
This includes the written word (novels, poems, articles, of weak AI, also known as narrow AI or artificial narrow
etc.), plays and dramatic works (scripts, choreography, intelligence (ANI). This type of AI is like a specialist,
etc.), musical compositions, and artistic creations trained to excel at specific tasks. Narrow AI powers
(paintings, photographs, sculptures, and more). It even impressive applications like virtual assistants (Siri, Alexa),
extends to films and sound recordings. medical diagnostic tools (IBM Watson), and even self-
driving cars.
In India, computer software and databases are also
protected under literary work. Computer software (source On the other hand, strong AI is the realm of science fiction
code) is protected in the PDF format and original databases for now. It encompasses two theoretical concepts:
are protected. A database is essentially a collection of
information that’s organized and stored. While the Indian Artificial General Intelligence (AGI): Imagine a machine
Copyright Act doesn’t have a specific definition for with human-level intelligence – capable of independent
“database” or “computer database,” it does recognize thought, problem-solving, learning, and planning for the
compilations, which include databases, as a form of literary future. This is AGI.
work and therefore subject to copyright protection. The
European Commission Directive on Databases Copyright Artificial Super Intelligence (ASI): This takes things a step
offers a broader definition of a database than what you further, envisioning machines that surpass human
might find in some national laws. The Directive defines a intelligence altogether. Think HAL 9000 from “2001: A
database as any collection of information, regardless of Space Odyssey” – that’s the fictional realm of ASI.
whether it’s creative works, simple data, or other materials.
The key aspect is that this information must be organized While strong AI isn’t a reality yet, researchers continue to
systematically, allowing for easy access electronically or explore its potential.
through other means. This definition even includes the
tools needed to navigate the database itself, such as indexes 3.1. Machine learning and deep learning
and thesauri.
Machine learning and deep learning are both subfields of
The Indian Copyright Act assigns authorship based on the AI, with deep learning being a more specialized branch of
type of work created. For literary & dramatic works, the machine learning. Both leverage a technology called neural
person who creates the work, like a writer or playwright, is networks to learn from vast amounts of data. These neural
considered the author. For musical works, the composer networks are essentially computer programs inspired by the
who writes the music holds authorship. For artistic works, human brain’s structure. They consist of interconnected
the artist who creates the visual work, such as a painting or layers that analyze data and make predictions based on
sculpture, is the author. For photographs, the photographer what they find.
who takes the picture is considered the author. For films &
sound recordings, the producer who oversees the creation is The key difference between machine learning and deep
designated as the author. For computer-generated works, learning lies in the type of neural networks used and the
creative outputs produced by computers in the realm of level of human involvement. Traditional machine learning
literature, drama, music, or art, the law assigns authorship algorithms use simpler neural networks with fewer layers.
to the person who commissioned the work. These algorithms typically rely on supervised learning,
where human experts pre-label and organize the data for the
3. ARTIFICIAL INTELLIGENCE AI to learn from.

AI is a branch of computer science focused on developing Deep learning takes things a step further by utilizing deep
machines that can mimic human intelligence. These neural networks – structures with many more hidden layers
machines can tackle problems and solve them in ways that compared to traditional machine learning models. This
traditionally require human intervention. AI can work complexity allows deep learning to perform unsupervised
independently or be combined with other technologies like learning. Here, the AI can analyze massive amounts of raw,
sensors or robotics to achieve even more complex tasks. unlabeled data to identify patterns and features on its own,
without needing human intervention to categorize
Think of digital assistants on your phone, GPS navigation everything beforehand. In essence, deep learning unlocks
systems, or even self-driving cars – these are all real-world the potential of machine learning by enabling it to process
examples of AI at work. Machine learning and deep and learn from much larger and more complex datasets.
learning, often mentioned alongside AI, are subfields within
computer science that deal with creating algorithms 4. AI-GENERATED CONTENT
inspired by the human brain. These algorithms can “learn”
from vast amounts of data, allowing them to improve their AI-generated content encompasses various creative outputs
ability to categorize information or make predictions over – written text, videos, even computer code – produced by
time. machines known as generative AI tools. These tools are
trained on massive datasets, enabling them to craft relevant

– 196 –