Artificial intelligence

As artificial intelligence (AI) continues to reshape industries, understanding the benefits

of AI technology, and the distinctions between technologies, becomes increasingly
crucial for making informed business decisions and navigating the rapidly evolving
digital landscape. From the foundations of machine learning and deep learning to the
complexities of natural language processing, the AI technology landscape encompasses
a spectrum of cutting-edge tools that are re-defining how organizations engage with
intelligent systems.
Demystifying the AI technology landscape and having a familiarity with these
technologies enables organizations to identify the most suitable solution for their
specific needs and ensures effective implementation and integration into existing
workflows.

AI technologies can be best understood as a kind of nesting of layered technologies that

all fall under the overarching category of IPA.
Intelligent process automation
IPA is a grouping of technologies used to manage and automate digital processes and is
designed to assist human workers by augmenting human labor and performing tasks
that are typically repetitive processes. IPA includes the following technologies:
 Robotic process automation (RPA): Seeks to automate tasks within existing
processes
 Digital process automation (DPA): Uses low-code tools to automate processes
that can span multiple applications with a focus on automating or partially
automating tasks involved in a variety of business practices that typically require
some form of human interaction
 Business process automation (BPA): Seeks to automate and streamline multi-
step processes
  AI: Systems with the ability to learn, perform and reason like a human
Moving into the AI realm, IPA begins to make decisions based on the provided data and
identified trends to augment human decisions or even make decisions independently.
IPA utilizes AI to drive innovation, optimize and streamline workforce efficiency.
Artificial intelligence
AI is an overarching term that encompasses the field of developing computer systems
to perform tasks that require human intelligence, such as reasoning, decision-making
and pattern learning and includes any technology that falls within the scope of ML and
automation.
Additional subfields of AI include:
 Natural language processing (NLP): Seeks to enable machines to understand,
interpret and respond to human language in a valuable way
 Large language model: Trained on extensive datasets to understand and
generate human-like text and can perform various language-related tasks
 Computer vision: Seeks to enable computers to derive information from
images, videos and other data
 Deep learning: ML in which multi-layer artificial neural networks adapt and
learn from vast amounts of data
With the continued advancements in autonomous vehicles and the ability to tackle
complex, nuanced challenges, artificial intelligence continues to evolve, inching closer
to human-like cognitive abilities.
Machine learning
Machine Learning (ML) enables AI systems to learn from data by recognizing patterns
and making predictions or recommendations based on statistical analysis. It can adapt
over time with new data, but its learning is generally guided by specific features and
rules defined by developers.
There are four types of ML:
 Supervised learning: In supervised learning, models are trained on labeled data.
Each training example is paired with an output label. The algorithm iteratively
makes predictions on the training data and is corrected by the teacher, allowing
the model to learn over time.
 Unsupervised learning: Machine learning models that operate without the need
for labeled data. Unlike supervised learning where models are trained on a
labeled dataset, unsupervised learning algorithms work with datasets that have
not been annotated or categorized. Unsupervised learning is crucial in scenarios
where obtaining labeled data is impractical or too expensive. It provides a
 method to extract meaningful information and insights from raw, unstructured
data, paving the way for deeper understanding and better decision-making.
 Semi-Supervised Learning: Semi-supervised learning falls between supervised
learning and unsupervised learning. In semi-supervised learning, the algorithm
is trained on a dataset that contains both labeled and unlabeled data. Generally,
a small amount of data is labeled while a large amount of data is unlabeled.
 Reinforcement learning: When the model learns from its own actions and
feedback from the environment, such as a robot navigating a maze or a game
agent playing chess, before using an algorithm to find a policy that maximizes a
reward function with the goal of learning optimal behavior for complex and
dynamic situations.
ML is widely used in various applications, such as speech recognition, image
recognition, spam filtering, fraud detection and self-driving cars.
Deep learning
Deep learning is a subset of ML that uses artificial neural networks to enable digital
systems to learn and make decisions based on unstructured data. A neural network is a
series of algorithms that can learn from input data to discern features, such as
distinguishing characteristics among various images, to make independent decisions
without explicit programming.
ML enables AI systems to learn from data by recognizing patterns and making
predictions or recommendations based on statistical analysis. It can adapt over time with
new data, but its learning is generally guided by specific features and rules defined by
developers.
On the other hand, deep learning employs neural networks with multiple layers (hence
why it's often referred to as deep) to analyze various factors of data. Unlike traditional
ML, deep learning autonomously extracts features from raw data, eliminating the need
for manual feature extraction. It learns from data in a way that is somewhat analogous
to human learning, through a hierarchy of concepts where each layer of the network
extracts and refines features from the input data.
Generative models, like those used in generative adversarial networks (GANs), are a
notable application of deep learning. They learn to generate new data that resembles the
training data and can be used to improve decision-making capabilities over time as they
are exposed to more data. However, it's not just generative AI that benefits from deep
learning; other forms of AI, like those used in image and speech recognition, natural
language processing and autonomous systems also leverage deep learning to enhance
their performance and capabilities over time.
Generative AI
Generative AI refers to a category of algorithms that can generate new data that
resembles a given set of training data. While it's not exclusively about language models,
it indeed encompasses them. These algorithms can create a variety of content types
including images, text, computer code or audio based on the patterns they learn from
the input data, aiding in accelerating the creative process.
Large language models like ChatGPT are instances of Generative AI applied to text
generation. They are trained on vast datasets to produce human-like text based on the
input prompts they receive. These models can be fine-tuned to perform a myriad of
tasks, enhancing their versatility and utility across different domains.
Generative AI has evolved into an advanced search and content generation tool
integrated across various industries to augment daily tasks and produce new content.
With 55% of organizations piloting or in production mode with generative AI, the
growth of AI in the market highlights the need for risk management as the creation of
new AI-driven content accompanies inherent risks.
As organizations continue to embrace AI-driven solutions, understanding the
differences in capabilities across available technologies is critical for organization’s
looking to harness its transformative power to drive innovation, increase efficiency,
unlock new possibilities and stay ahead of competitors in the rapidly evolving digital
landscape.
I. Main idea

What is the main idea of the passage?

A) The passage discusses the historical development of artificial intelligence
technologies.
B) The passage explains the various components of AI technologies and their
significance in transforming business processes.
C) The passage focuses solely on the ethical implications of using AI in organizations.
D) The passage highlights the limitations of AI technologies in modern applications.

Which statement best captures the essence of the passage?

A) AI technologies are primarily used for entertainment purposes.
B) Understanding AI technologies is essential for organizations to enhance efficiency
and drive innovation in a rapidly changing digital landscape.
C) The passage emphasizes the need for organizations to avoid using AI technologies.
D) AI technologies are only relevant to large corporations and not small businesses.

II. Questions
Answer the following questions
1. Why is it important for organizations to understand the distinctions between
different AI technologies?
2. What does IPA stand for, and what is its primary purpose in organizations?
3. List the four technologies included under IPA as mentioned in the text.
 4. What role does AI play within the context of IPA?
5. What is the overarching definition of AI according to the passage?
6. What are the four subfields of AI mentioned in the text?
7. How does Natural Language Processing (NLP) function, and what is its primary
goal?
8. What does a large language model do?
9. Describe how computer vision contributes to the capabilities of AI.
10. What is deep learning?
11. How does reinforcement learning differ from supervised and unsupervised
learning?
12. How does generative AI contribute to the creative process according to the
passage?

III. Understand the passage

Decide whether the following statements are true or false (T/F) by referring to the
information in the text. Then make the necessary changes so that the false statements
become true.
1. Intelligent Process Automation (IPA) is solely focused on replacing human
workers in various industries.
2. Robotic Process Automation (RPA) is a component of IPA that aims to automate
tasks within existing processes.
3. Deep learning requires manual feature extraction from data, making it similar to
traditional machine learning techniques.
4. Generative AI encompasses a variety of algorithms capable of creating new data
types, including images, text, and audio.
5. The passage states that machine learning (ML) can only operate with labeled
data, making it less flexible than deep learning.
6. Natural Language Processing (NLP) is a subfield of AI that focuses on enabling
machines to understand and respond to human language
7. The passage indicates that 55% of organizations are currently avoiding the
implementation of generative AI technologies.
8. Reinforcement learning involves learning from feedback based on the model's
actions and interactions with the environment.
9. The passage suggests that understanding the differences in AI technologies is not
critical for organizations looking to innovate and improve efficiency.

IV. Understanding words

Refer back to the text and find synonyms for the following words
1. Learning
2. Estimation
3. Advantages
4. Enhancing, increasing
 5. Novelty
6. Multi-layer learning
Refer back to the text and find antonyms for the following words
7. Human Intelligence
8. Manual Operation
9. Similarities
10. Limitations
11. Static
12. Inflexible
V. Content review
Use the information in the text to complete the following summary
1. IPA aims to assist human workers by ……………. repetitive tasks, thereby
augmenting human labor rather than replacing it.
2. Robotic Process Automation (RPA) focuses on automating specific tasks
within existing ………………., streamlining operations without the need for
significant changes to the underlying systems.
3. DPA employs …………. tools to automate processes that span multiple
applications, emphasizing the automation of tasks that typically require
human interaction.
4. BPA seeks to automate and optimize ……………… processes, enhancing
overall workflow efficiency.
5. AI is defined as the field dedicated to developing computer systems capable
of performing tasks that require human-like intelligence, such as reasoning,
……………., and pattern recognition.
6. Natural Language Processing (NLP) focuses on enabling machines to
understand, interpret, and respond to human …………… in meaningful
ways.
7. Large Language Models are trained on extensive datasets to generate
……………. text and perform various language-related tasks.
8. Computer Vision seeks to enable ……………. to extract information from
images, videos, and other visual data.
9. Deep learning utilizes multi-layer ………………………… to learn from vast
amounts of unstructured data, allowing for autonomous feature extraction.
10. There are four types of …………………..: supervised learning, unsupervised
learning, …………………………….., and reinforcement learning, each
with distinct methodologies and applications.
11. Generative AI is defined as a category of algorithms that can create
………………………, including images, text, and audio, based on learned
patterns from training data.
VI. Vietnamese- English translation
Translate the following sentences into English, basing on the information in the
text.
1. Trí tuệ nhân tạo đóng vai trò quan trọng trong nhiều lĩnh vực và có ảnh hưởng
sâu rộng đến cuộc sống hàng ngày cũng như các ngành công nghiệp.
…………………………………………………………………………………
…………………………………………………………………………………
…………………………………………………………………………………
2. Học máy (Machine Learning - ML) được thảo luận như một thành phần chính
của trí tuệ nhân tạo (AI), với các loại hình của nó - học có giám sát, học không
giám sát, học bán giám sát và học tăng cường - cùng với các ứng dụng của chúng.
Học sâu (Deep Learning), một phân nhánh của ML, được đặc trưng bởi việc sử
dụng mạng nơ-ron để phân tích dữ liệu phi cấu trúc.
.…………………………………………………………………………………
…………………………………………………………………………………
…………………………………………………………………………………
VII. Discussion
Talking about Artificial Intelligence
- Subfields of AI
- Applications of AI
- The Future of AI: Opportunities and Challenges