AI2111 Smart Science AI-Powered Discoveries for Academic Minds 1
Introduction to AI in Scientific Research
Module 1 – Introduction to AI in
Scientific Research
At the end of this module, you are expected to:
1. Define artificial intelligence;
2. Identify the basic characteristics of AI; and
3. Understand how AI stands out in different industries.
INTRODUCTION
Over the past several decades, there has been a significant surge in study and development of
Artificial Intelligence (AI) after Alan Turing's 1950 foundational article regarding the feasibility of
programming an electrical computer to behave intelligently. Even though there are still many fundamental
issues that need to be resolved, fields like expert systems, natural language processing, speech recognition,
computer vision, robotics, etc. have advanced considerably.
A subfield of computer science called artificial intelligence is focused on the design and research of
computer systems. Artificial intelligence demonstrates intelligence in some form by introducing systems
that are capable of learning new ideas and tasks, reasoning, and making meaningful inferences about their
environment. In addition, artificial intelligence systems are capable of understanding spoken language,
perceiving and comprehending visual scenes, and carrying out other tasks that call for human intellect
(Freitas, 1999).
Global acceptance of AI as a burgeoning technology has occurred. It has been transforming the
industry from a number of industries for a while now. It is an all-encompassing technology used in nearly
every sector of the economy. The influence of AI on service delivery across several industries is covered in
this section.
Nowadays, fully autonomous vehicles are a reality. Tesla is the first automaker to equip a vehicle
with every sensor, camera, and piece of software required for a computer to operate autonomously. The
next big focus for autonomy may be trucks. These vehicles will have a significant influence on
infrastructure and road safety while also saving businesses money by lowering labor expenses.
BASIC CHARACTERISTICS OF ARTIFICIAL INTELLIGENCE
Artificial Intelligence (AI) deals with the development and use of computer systems that can solve
issues that typically require human intelligence. These issues are associated with high levels of complexity
and/or natural jobs (like eyesight or natural language comprehension). Typically, traditional algorithmic
techniques are unable to address them. AI systems mostly work with symbolic data to solve issues rather
than only numerical data, as is customary in computer science. AI heavily utilizes a variety of knowledge
Course Module
AI2111 Smart Science AI-Powered Discoveries for Academic Minds 2
Introduction to AI in Scientific Research
types related to a certain application domain. As a result, the issues of knowledge representation,
acquisition, and application are crucial to AI research and development. Among the most active branches
is the deployment of knowledge-based systems.
One of the main characteristics of AI is its multidisciplinarity. Despite the tight relationship between
computer science and artificial intelligence (AI), concepts from other fields such as languages, logic, and
cognitive sciences are also needed to create AI projects. AI is also used in many other areas, including as
manufacturing, banking & finance, and medical.
The basic characteristics of AI that were previously addressed may be found in a variety of classifications
(Haton, 1993). Now, let’s briefly describe a couple of them.
Theorem-proving: this has been done since AI's inception. Despite the fact that no new theorem has been
proved by a machine, significant results have been achieved. Because it deals with basic elements of
thinking and control, research in this area and the field of games (chess, etc.) is significant.
Natural language processing is a broad phrase that encompasses a number of tasks such as computer-
assisted translation, database or service access, automatic document indexing and retrieval, computer-
generated sentences for human-machine dialogue, etc. Thus, the difficulties to be solved are extremely
challenging, but the potential applications are exciting. To comprehend the language or text being
processed in these tasks, one must possess a vast amount of information and effective contextual reasoning
skills.
Voice recognition and understanding: this area is somewhat comparable to the preceding one, but it is
more challenging due to the inherent determinism in voice signal decoding. Commercial solutions are
currently offered for continuous voice recognition (dictation, transcription, etc.), huge vocabulary, and
isolated word control systems. Sentence meaning comprehension will need the application of advanced
knowledge-based reasoning techniques in the building of real man-machine conversation systems using
speech.
Interpretation of images and vision: among the first systems in the business to be put into operation
were image interpretation and vision applications of sophisticated image processing. Regarding voice
processing, the initial systems relied solely on pattern recognition algorithms to handle basic tasks (such
as identifying typed characters or basic two-dimensional objects). Advanced uses include the steering of a
robot, the analysis of pictures (for inspection or diagnosis reasons), etc. Then, in order to comprehend a
picture or a scenario to some degree, one must once more engage in a knowledge-based reasoning process.
Robotics: the majority of industrial robots on the market now only reproduce motion patterns and tasks
that they were taught in an early stage. Therefore, mechanical engineering was a bigger factor in early
robotics endeavors than cognitive behavior. However, as new generations of robots are being designed, AI
methods are being incorporated into various aspects of their behavior: perception (see the previously
mentioned vision systems for mobile robots), reasoning to determine the best course of action in a given
situation, planning (and dynamic replanning) of actions to reach a predetermined goal, etc.
Course Module
AI2111 Smart Science AI-Powered Discoveries for Academic Minds 3
Introduction to AI in Scientific Research
THE CURRENT STATE OF AI-BASED SOFTWARE SYSTEMS
A novel kind of system known as Generative Adversarial Networks (GANs) has emerged as a result of
recent developments in AI. GANs are capable of producing realistic text, audio, or image output. Some fear that
in the future, this technology may replace humans because of its amazing powers. AI is transforming our lives in
many ways, GANs being just one example. More recent instances of artificial intelligence (AI) and its uses in
software systems like GPT3, DALL.E, and virtual reality/augmented reality (VR/AR) are covered in this section.
Advanced algorithms, automated reasoning tools, and fundamental models are only a few of the layers that make
up AI-based software systems. These layers are used by many of the most well-known AI-based systems,
including as DALL.E, AlphaGo, RoBERTa, and GPT3. Prominent models such as DALL.E and GPT3 have
shown remarkable results in natural language processing (NLP) and computer vision.
Transformers is a deep learning method used to build an NLP model known as the GPT3. It was made
available in 2020 and trained on a Common Crawl text corpus. GPT3 uses a large dataset that has been trained in
the English language to provide outputs based on the input entered. The model may be trained to perform any
task that can be thought of, such as creating text and solving arithmetic problems. In addition, GPT3 may be used
to generate text, translate across languages, respond to inquiries about images, and more.
The DALL.E model is an image generator that utilizes variational autoencoders (VAEs), a deep learning
method. In a similar vein, pictures may be generated by DALL.E based on text descriptions by training it with an
image dataset. It was released in 2021 after being trained on datasets like ImageNet. DALL.E may be used to
create pictures that correspond with user-provided captions or URLs. The company that created these models,
OpenAI, is closely associated with the US government and the military-industrial complex (MIC). DeepMind
created AlphaGo, a piece of software designed to compete against itself in the traditional game of Go. Although
the game is similar to chess, it is significantly more complex because to the several moves made in each round
and the simple rules. AlphaGo used reinforcement learning, playing Go against itself often, to hone its abilities
over time and finally master every possible situation in a game of Go with 100% accuracy.
The program RoBERTa, created by Facebook AI Research (FAIR), uses deep learning techniques to
address natural language processing (NLP) problems such as sentence classification and machine translation.
HOW AI STANDS OUT IN DIFFERENT COUNTRIES.
AI is now recognized as a rapidly developing technology on a global scale. For some time now, it has
been changing the industry from various industries. It is a universal technology that is employed in almost every
economic industry. This section discusses how AI is affecting service delivery in a number of sectors.
Fully autonomous cars are a reality these days. As the first manufacturer, Tesla has fitted a car with every
sensor, camera, and piece of software needed for a computer to run on its own. Trucks could be the next major
area of autonomous research. These cars will save businesses money by reducing labor costs, and they will also
have a big impact on road safety and infrastructure.
A few other sectors are incorporating AI as well. AI, for instance, facilitates hedge fund investment
decisions and aids in predicting in the banking industry. Using machine learning and statistical methods,
predictive analytics, often known as forecasting, uses artificial intelligence to estimate future occurrences based
on historical data. For instance, forecasting may be used to estimate stock prices, client demand, and product
Course Module
AI2111 Smart Science AI-Powered Discoveries for Academic Minds 4
Introduction to AI in Scientific Research
sales. Amazon's "Customers who bought this item also bought" product suggestion engine is a well-known
illustration of predictive analytics. It makes product recommendations based on user preferences by utilizing
millions of consumers' historical purchase data.
By collecting data from medical imaging, reports, and health records, artificial intelligence (AI) is
assisting medical professionals in diagnosing illnesses. This facilitates quicker diagnosis, referral of the patient
for additional testing, and prescription writing. AI may also be utilized in the therapeutic process to monitor
patients and notify physicians of any issues. By 2035, AI will save over 7 million lives, predicts Forbes.
Chatbots for customer service and stock management are only two use of AI in retail. In order to increase
productivity, efficiency, and accuracy, many firms are utilizing AI. Furthermore, businesses discover novel
applications of AI to streamline consumer and staff lives, from customer support to product design.
Course Module
AI2111 Smart Science AI-Powered Discoveries for Academic Minds 5
Introduction to AI in Scientific Research
References and Supplementary Materials
1. Haton, J.P. (2006). A brief introduction to artificial intelligence. Loria.
2. Ghosh, M. & Arunachalam, T. (2021). Artificial intelligence for information management: A
healthcare perspective. Research Gate. 10.1007/978-981-16-0415-7_2
3. https://www.red-gate.com/simple-talk/development/data-science-
development/introduction-to-artificial-intelligence/#heading-2
Course Module
