This content is an AI-powered exam review covering fundamental concepts of Generative AI, including model types, data processing, Google Cloud's AI offerings, techniques for improving model output, and business strategies for successful AI implementation, with a strong emphasis on responsible AI practices.
Mind Map
انقر للتوسيع
انقر لاستعراض خريطة الذهن التفاعلية الكاملة
Welcome to this exam review from Birdsy,
your AI powered study partner. Be sure
to like and subscribe for more exam prep
videos. Let's get started.
>> Let's start with a review of the
fundamentals of generative AI.
Core generative AI concepts,
terminology, and model types.
What is a key characteristic of large
language models? Is it A. They process
and generate humanlike text. B they
primarily focus on image generation. C
they are used exclusively for data
encryption or D they operate only in
multimodal environments.
The answer is A they process and
generate humanlike text.
LLMs are designed to understand and
generate humanlike text by processing
large amounts of language data, making
them capable of tasks like translation
and summarization.
In generative AI, what does the term
diffusion primarily refer to? Is it A
the spread of AI models across networks?
B, the distribution of training data, C,
the blending of multiple AI models, or
D, the transformation of noise into a
coherent image.
The answer is D, the transformation of
noise into a coherent image.
Why this question may seem a bit tricky?
Diffusion and generative AI refers to a
process of gradually transforming noise
into a coherent image. Not to be
confused with data distribution or
network diffusion.
Sarah is developing a multimodal AI
system that processes both text and
images. Which model type should she
consider? Is it A large language models,
B diffusion models, C multimodal models,
The answer is C multimodal models.
Multimodal models are designed to handle
and integrate multiple types of data
such as text and images, making them
suitable for Sarah's project.
Which of the following is a common
application of diffusion models? Is it A
text summarization?
B image generation, C speech
recognition, or Data encryption.
The answer is B image generation. Diffusion
Diffusion
models are often used for generating
highquality images by iteratively
refining noise into a coherent image.
John wants to create a chatbot that can
understand and generate humanlike
responses. Which model should he use? Is
it a large language models, b diffusion
models, c multimodal models, or d
convolutional neural networks?
The answer is a large language models.
LLMs are ideal for chatbots as they are
trained to understand and generate
humanlike text making them suitable for
conversational AI.
What is a primary function of large
language models in generative AI? Is it
A generating realistic images, B
understanding and generating humanlike
text, C processing audio signals,
or D predicting stock market trends.
The answer is B understanding and
generating humanlike text.
LLMs are designed to understand and
generate humanlike text based on input
data, making them essential for tasks
like text generation and language translation.
translation.
In the context of generative AI, what
does diffusion most accurately describe?
Is it A the spread of AI technologies
across industries? B the gradual
improvement of AI models over time. C a
process of generating data by reversing
a noise process or D the integration of
AI with existing systems.
The answer is C a process of generating
data by reversing a noise process.
Why this question may seem a bit tricky?
Diffusion can be confused with other
processes but in AI it refers to a
method of generating data by reversing a
noise process.
Have you noticed the code in the bottom
corner? Next time you're stumped, take
that code to birdsy.ai for immediate
help and a free study session with
Birdsy, your personal AI study partner.
Plus, you'll get access to hundreds of
additional questions, helping you move
closer to passing your exam with
confidence. The next time you're feeling
Emma is building a system that can
generate both text and images from a
single input. Which model type should
she use? Is it A large language models,
B diffusion models, C convolutional
neural networks, or D multimodal models?
The answer is D multimodal models.
Multimodal models are designed to handle
and generate multiple types of data such
as text and images from a single input.
Which of the following is a
characteristic of diffusion models in
AI? Is it A generating data by reversing
a noise process?
B primarily used for text generation,
C optimized for audio signal processing
or D used for predicting numerical data trends.
trends.
The answer is A generating data by
reversing a noise process.
Diffusion models are characterized by
their ability to generate data by
reversing a noise process, making them
suitable for creating realistic images.
Carlos is tasked with creating a chatbot
that can understand and generate
responses in multiple languages. Which
model type should he consider? Is it A
diffusion models, B large language
models, C recurrent neural networks, or
D generative adversarial networks?
The answer is B large language models.
LLMs are well suited for multilingual
text understanding and generation,
making them ideal for creating chat bots
that operate in multiple languages.
Data types, embeddings, tokenization,
fine-tuning, and model life cycle at a
high level. What is the primary purpose
of tokenization in natural language
processing? Is it A to encrypt data for
security purposes? B to split text into
smaller units for analysis.
C to convert text into numerical data
directly or D to summarize text by
extracting key phrases.
The answer is B to split text into
smaller units for analysis.
Tokenization is the process of breaking
down text into smaller units such as
words or phrases to facilitate analysis
by AI models.
In the context of embeddings, which
statement is true about their
dimensionality? Is it A. Embeddings
always have a dimensionality of 100? B.
Embeddings must match the dimensionality
of the input data. C. Embeddings can
have varying dimensions based on the
model or D. Embeddings are always two-dimensional.
two-dimensional.
The answer is C. Embeddings can have
varying dimensions based on the model.
Why this question may seem a bit tricky.
Embeddings can have various dimensions,
but they are not fixed to a specific
number like 100 or 300. It depends on
the model and application.
Alex is fine-tuning a pre-trained
language model for a specific task. What
is a crucial step Alex should take
during this process? Is it A. Adjust the
model's parameters using task specific
data, B increase the model's size by
adding more layers, C reduce the model's
complexity by removing layers, or D use
the model without any modifications.
The answer is A. Adjust the model's
parameters using task specific data.
Finetuning involves adjusting the
model's parameters on a task specific
data set to improve performance on that task.
task.
What is a common use of embeddings in
generative AI? Is it A to encrypt data
for secure transmission? B to generate
random text sequences. C to visualize
data in two dimensions or D to represent
words in a continuous vector space.
The answer is D to represent words in a
continuous vector space.
Embeddings are used to represent words
or phrases in a continuous vector space,
capturing semantic meanings and relationships.
relationships.
Jaime is developing a chatbot using a
generative AI model. What should Jaime
focus on to ensure the model understands
user input effectively? Is it A
increasing the model's training data
size, B reducing the model's response
time, C implementing effective
tokenization techniques, or D enhancing
the MBI's graphical user interface.
The answer is C implementing effective
tokenization techniques.
Tokenization is crucial for breaking
down user input into manageable parts
for the model to process effectively.
What is the primary role of embeddings
in generative AI models? Is it A to
increase the model's computational
speed, B to represent data in a
continuous vector space? C to reduce the
size of the training data set or D to
ensure data privacy during model training.
training.
The answer is B. To represent data in a
continuous vector space,
combings convert categorical data into
numerical form, capturing semantic
relationships and enabling models to
process and understand the data more effectively
effectively
in the context of model fine-tuning.
What is a common misconception about
adjusting learning rates? Is it A a
lower learning rate always results in
better model accuracy? B learning rate
should remain constant throughout
training, C a higher learning rate
always speeds up training. Or D learning
rates have no impact on model performance.
performance.
The answer is C. A higher learning rate
always speeds up training.
Why this question may seem a bit tricky?
It's easy to assume that a higher
learning rate always speeds up training,
but it can lead to instability and poor convergence.
convergence.
Maria is developing a generative AI
model and needs to ensure it effectively
processes user input. What should she
focus on to improve the model's
understanding of language? Is it A
implementing effective tokenization strategies?
strategies?
B, increasing the model's parameter
count. C, reducing the training data set
size, or D, focusing solely on model architecture.
architecture.
The answer is A implementing effective
tokenization strategies.
Tokenization is crucial for breaking
down text into manageable pieces,
allowing the model to process and
understand language effectively.
finding this exam review helpful? If you
haven't already, like and subscribe for
more videos that'll get you ready to pass
pass
Google Cloud's generative AI offerings.
Key Google Cloud models and platforms.
What is the primary purpose of Google
Cloud's model garden? Is it A to host
user generated content, B to provide a
repository of pre-trained models, C to
manage cloud storage, or D to offer
cloud-based gaming services.
The answer is B to provide a repository
of pre-trained models.
Model Garden is designed to provide a
repository of pre-trained models that
users can access and deploy,
facilitating easier integration of AI
into applications.
Which Google Cloud Platform is
specifically designed for building,
deploying, and scaling machine learning
models? Is it A Google Kubernetes
Engine, B Google Cloud Functions, C
Vertex AI, or D Google Cloud Storage?
The answer is C. Vertex AI.
Why this question may seem a bit tricky.
Vertex AI is specifically designed for
machine learning tasks while other
options might seem related but serve
different purposes.
A company wants to integrate a
pre-trained language model into their
customer service chatbot. Which Google
Cloud offering should they consider
using? Is it A Model Garden, B Google
Cloud Run, C Google Big Query, or D
Google Cloud Functions?
The answer is A. Model Garden.
Model Garden provides access to
pre-trained models, making it suitable
for integrating into applications like chatbots.
chatbots.
What is the main feature of Google
Cloud's Vertex AI? Is it A data storage
management? B web hosting services, C
cloud-based gaming, or D building,
deploying, and scaling machine learning models.
models.
The answer is D. Building, deploying,
and scaling machine learning models.
Vert.Ex AI is primarily used for
building, deploying, and scaling machine
learning models, offering a
comprehensive suite of tools for these tasks.
tasks.
A developer is tasked with deploying a
scalable AI model for realtime data
analysis. Which Google Cloud Platform
should they use? Is it A. Google Cloud
Storage, B. Google App Engine, C Vertex
AI, or D Google Cloud Functions.
The answer is C. Vertex AI.
Vertex AI is ideal for deploying
scalable AI models, offering tools for
realtime data analysis and model management.
management.
What is the primary function of Google
Cloud's Vertex AI? Is it A to provide
cloud storage solutions, B to build,
deploy and scale machine learning
models, C to manage network security or
D to offer data analytic services.
The answer is B to build, deploy and
scale machine learning models.
Vert.ex AI is designed to help users
build, deploy, and scale machine
learning models efficiently, integrating
various Google Cloud services.
Which Google Cloud offering is best
suited for accessing a variety of
pre-trained models? Is it A Vertex AI, B
Gemini, C Model Garden, or D Cloud Functions?
Functions?
The answer is C. Model garden.
Why this question may seem a bit tricky?
The term pre-trained models might lead
one to think of vertex AI, but model
garden is specifically designed for
accessing a variety of pre-trained models.
models.
A retail company wants to implement a
recommendation system using Google
Cloud's AI offerings. Which platform
should they use to efficiently build and
deploy their model? Is it Avertex AI, Bquery,
Bquery,
C Cloud Storage,
or D App Engine?
The answer is A. Vert.Ex AI.
Vert.Ex AI is ideal for building and
deploying machine learning models,
making it suitable for implementing a
recommendation system.
Which Google Cloud model family is known
for its advanced language understanding capabilities?
capabilities?
Is it A TensorFlow,
B AutoML,
C BERT, or D Gemini?
The answer is D Gemini.
Why this question may seem a bit tricky.
The term language understanding might
suggest multiple models, but Gemini is
specifically known for its advanced
capabilities in this area.
APIs, services, and solution patterns on
Google Cloud.
What is the primary purpose of Google
Cloud's rag solution pattern?
Is it A to store large data sets efficiently,
efficiently,
B to improve the speed of data
processing, C to enhance content
generation with relevant information? Or
D to provide real time analytics.
The answer is C. To enhance content
generation with relevant information.
RAG is designed to enhance the quality
of generated content by retrieving
relevant information from external
sources, ensuring more accurate and
contextually relevant outputs.
Which Google Cloud service is primarily
used for ground truthing in AI models?
Is it a AI platform,
B data labeling service, C big query
or D cloud storage?
The answer is B data labeling service.
Why this question may seem a bit tricky?
Ground truthing involves labeling data
accurately, which is a key function of
Google Cloud's data labeling service,
not AI platform or Big Query.
A company wants to integrate a
generative AI model with their existing
database to provide realtime customer
support. Which Google cloud solution
pattern should they consider? Is it A
retrieval augmented generation?
B data labeling service, C cloud
functions, or D vertex AI.
The answer is A retrieval augmented generation.
generation.
Rag is suitable here as it can retrieve
relevant data from the database to
enhance the AI models responses making
it ideal for real-time customer support.
Which Google cloud service is used to
deploy machine learning models at scale?
Is it A big query,
B cloud storage, C data labeling
service, or d Vertex AI?
The answer is D. Vertex AI.
Vert.Ex AI is designed to build, deploy,
and scale machine learning models
efficiently on Google Cloud.
A developer is tasked with creating a
plug-in that allows a generative AI
model to access external APIs for
additional data. Which Google Cloud
feature should they leverage? Is it A.
Big Query, B. data labeling service, C
cloud functions, or D Vertex AI.
The answer is C, Cloud Functions.
Cloud Functions can be used to create
plugins that interact with external APIs
providing the necessary data to the AI model.
model.
What is the primary function of Google
Cloud's Vertex AI matching engine? Is it
A deploying machine learning models,
B performing largecale similarity
matching, C managing data sets for
training or D automating hyperparameter tuning.
tuning.
The answer is B performing large scale
similarity matching.
Vert.ex AI matching engine is designed
to perform large-scale similarity
matching which is crucial for tasks like
recommendation systems and image search.
Which Google Cloud service is best
suited for integrating generative AI
models with external APIs? Is it API
gateway, B cloud run, C cloud functions
or D pub sub?
The answer is C. Cloud Functions.
Why this question may seem a bit tricky?
The term integrating might lead one to
think of API gateway, but cloud
functions is more suited for executing
code in response to API calls.
A retail company wants to use Google
Cloud to enhance their product
recommendation system with generative
AI. Which solution pattern should they
consider to ensure the AI model uses the
most relevant data?
Is it A retrieval augmented generation,
B grounding,
C vertex AI autoML, or D big query ML?
The answer is A retrieval augmented generation.
generation.
Retrieval augmented generation is ideal
for enhancing recommendation systems by
ensuring the AI model accesses the most
relevant data.
Use case alignment. Which service model
solves which type of business problem?
Which Google Cloud service is best
suited for generating humanlike text
responses in a customer service chatbot?
Is it A cloud vision API, B dialogue
flow, C or D cloud storage?
The answer is B dialogue flow.
Google Cloud's dialogue flow is designed
for building conversational interfaces,
making it ideal for chat bots that
require humanlike text responses.
Which Google Cloud service would you use
to analyze customer sentiment from
social media posts? Is it A Dialogue Flow,
Flow,
B Cloud Speech to Text, C Natural
Language API, or D Cloud Translation API?
API?
The answer is C natural language API.
Why this question may seem a bit tricky?
Natural language API is often confused
with services like dialogue flow, but it
specifically analyzes text for
sentiment, making it the right choice here.
here.
A retail company wants to personalize
product recommendations for its online
shoppers based on their browsing
history. Which Google Cloud service
should they use? Is it A recommendations
AI, B cloud autoML, C cloud functions or
D cloud pub sub?
The answer is A recommendations AI.
Recommendations AI is specifically
designed to provide personalized product
recommendations based on user behavior
and browsing history.
Which Google Cloud service would you
choose to transcribe audio recordings
into text for analysis? Is it A Dialogue
Flow, B Cloud Vision API,
C Natural Language API, or D Cloud
Speech to Text?
The answer is D. Cloud Speech to Text.
Why this question may seem a bit tricky?
Cloud speech to text is often confused
with natural language API, but it
specifically transcribes audio to text.
A media company wants to automatically
tag and categorize large volumes of
video content. Which Google Cloud
service should they use? Is it A cloud autoML,
autoML,
B video intelligence API, C cloud storage
storage
or DQ?
The answer is B video intelligence API.
Video intelligence API is designed to
analyze video content, making it ideal
for tagging and categorizing videos.
>> Finding this exam review helpful? If you
haven't already, like and subscribe for
more videos that'll get you ready to pass.
pass.
>> Techniques to improve model output,
prompt engineering, and tuning
strategies to overcome limitations.
What is a common strategy to reduce
hallucinations in AI model outputs? Is
it A increasing the model's temperature
setting? B, providing more specific and
detailed prompts.
C, using a larger data set for training,
or D, decreasing the model's learning rate.
rate.
The answer is B, providing more specific
and detailed prompts.
Using more specific and detailed prompts
helps guide the model towards generating
accurate and relevant outputs, reducing
the chance of hallucinations.
Which approach is least effective in
mitigating bias in AI models? Is it A
implementing fairnessaware algorithms,
B conducting bias audits on training
data, C regularly updating the model
with new data
or D using diverse data sources.
The answer is C regularly updating the
model with new data.
Why this question may seem a bit tricky?
Regularly updating the model with new
data is important, but without
addressing bias in the data itself, it
may not effectively mitigate bias.
A developer notices their AI model is
generating biased outputs when analyzing
customer feedback. What should they
prioritize to address this issue? Is it
A. reviewing and adjusting the training
data for biases. B increasing the
model's complexity. C reducing the
model's parameters or D implementing a
more aggressive learning rate.
The answer is A reviewing and adjusting
the training data for biases.
Prioritizing a review and adjustment of
the training data ensures that biases
are identified and corrected at the source.
source.
What is a key benefit of prompt
engineering in AI models? Is it A
increases the model's processing speed,
B reduces the model's computational cost,
cost,
C enhances the model's training
efficiency, or D improves the accuracy
and relevance of outputs.
The answer is D improves the accuracy
and relevance of outputs.
Prompt engineering helps guide the model
to produce more accurate and
contextually relevant outputs by
providing clear and specific instructions.
instructions.
In a project to develop a multimodal AI
system, the team struggles with
integrating text and image data. What
strategy should they use to overcome
modality limits? Is it A separate models
for each data type? B, a unified model
architecture for all data types. C,
increasing the data set size for each modality,
modality,
or D, focusing on one modality at a time.
time.
The answer is B, a unified model
architecture for all data types.
Using a unified model architecture
allows for seamless integration of
different data types, addressing
modality limits effectively.
A team is developing a chatbot that
frequently generates irrelevant
responses. What strategy should they
implement to improve the relevance of
the chatbot's output? Is it A increase
the model's training data? B refine the
prompts used in interactions,
C reduce the model's complexity, or D
switch to a different AI framework.
The answer is B. Refine the prompts used
in interactions.
In this scenario, refining the prompts
can guide the model to produce more
contextually appropriate responses
addressing the issue of irrelevance.
Rounding retrieval augmented generation,
evaluation metrics and monitoring model quality.
quality.
What is the primary purpose of retrieval
augmented generation in AI models? Is it
A to increase the speed of model training,
training,
B to enhance the accuracy and relevance
of model outputs, C to reduce the
computational cost of AI models, or D to
simplify the model architecture.
The answer is B to enhance the accuracy
and relevance of model outputs.
Rag combines retrieval of relevant
documents with generation capabilities
to enhance the accuracy and relevance of
AI model outputs.
Which metric is most suitable for
evaluating the grounding of AI model
outputs? Is it A latency,
B throughput, C factual consistency,
or D model size?
The answer is C. Factual consistency.
Why this question may seem a bit tricky?
Grounding evaluation often involves
human judgment, but factual consistency
metrics like precision are used to
assess how well outputs align with
source data.
A company uses rag to improve customer
service chat bots. They notice
inconsistent responses. What should they
evaluate to address this issue?
Is it A retrieval accuracy, b model
training speed, c user interface design,
or d server uptime?
The answer is a retrieval accuracy.
Evaluating retrieval accuracy ensures
that the most relevant documents are
used, which can improve response consistency.
consistency.
During a model quality review, a team
finds that their AI models outputs are
not grounded in the provided data. What
is a likely cause? Is it A excessive
model parameters, B high latency, C
overfitting, or D inadequate retrieval mechanisms.
mechanisms.
The answer is D. Inadequate retrieval mechanisms.
mechanisms.
Inadequate retrieval mechanisms can lead
to outputs that are not well grounded in
the source data.
Business strategies for a successful Gen
AI solution.
Identifying high value use cases,
building business cases, and measuring ROI.
ROI.
What is the primary goal when
identifying high value use cases for a
Gen AI solution? Is it A to implement
the latest technology trends?
B to align with strategic business objectives,
objectives,
C to reduce operational costs
immediately or D to increase employee engagement.
engagement.
The answer is B to align with strategic
business objectives.
The primary goal is to align the use
case with strategic business objectives,
ensuring it delivers significant value
and supports the organization's goals.
Why might a business case for a Gen AI
solution fail to convince stakeholders?
Is it A it includes too many financial projections?
projections?
B it focuses on long-term strategic
goals, C it lacks clear ROI demonstration,
demonstration,
or D, it emphasizes competitive advantage.
advantage.
The answer is C, it lacks clear ROI demonstration.
demonstration.
Why this question may seem a bit tricky?
Stakeholders need clear quantifiable
benefits. Focusing solely on technical
details without demonstrating ROI can
lead to a lack of buyin.
A company is considering a Gen AI
solution to improve customer service.
What should they prioritize to build a
strong business case? Is it A
identifying quantifiable benefits? B,
highlighting the novelty of the
technology, C focusing on potential cost
savings only or D emphasizing employee satisfaction.
satisfaction.
The answer is A identifying quantifiable benefits.
benefits.
Prioritizing the identification of
quantifiable benefits ensures the
business case is compelling and
demonstrates clear value to stakeholders.
stakeholders.
Which metric is most effective for
measuring the ROI of a Gen AI solution?
Is it A employee satisfaction scores, B
customer engagement rates, C operational
efficiency improvements, or D financial
gains versus investment costs?
The answer is D. Financial gains versus
investment costs.
Why this question may seem a bit tricky?
While all metrics are useful, ROI is
best measured by comparing the financial
gains to the investment costs.
In a scenario where a company wants to
implement a Gen AI solution to enhance
product recommendations, what should be
the first step in identifying high value
use cases? Is it A analyzing competitor strategies?
strategies?
B, investing in the latest AI technology.
technology.
C, understanding customer needs or D
training employees on AI tools.
The answer is C understanding customer needs.
needs.
Understanding customer needs ensures the
solution addresses real pain points and
delivers value forming the basis for a
high value use case.
Responsible AI, ethical, legal,
security, governance, and organizational
change management.
What is a primary goal of responsible AI
in business strategies? Is it Asuring
ethical and transparent AI use? B
maximizing profit through AI, C
automating all business processes, or D
eliminating human oversight in AI systems.
systems.
The answer is a ensuring ethical and
transparent AI use.
Responsible AI aims to ensure that AI
systems are developed and used in ways
that are ethical, transparent, and
accountable, minimizing harm and
maximizing benefits.
Which aspect of responsible AI is often
misunderstood as solely a technical
challenge? Is it a ethical considerations,
considerations,
b governance, c security measures or d
organizational change management?
The answer is B governance.
Why this question may seem a bit tricky?
Governance is often seen as a technical
issue, but it also involves ethical and
organizational considerations making it
more complex than just a technical challenge.
challenge.
A company is deploying a Gen AI solution
and wants to ensure it aligns with legal
standards. What should they prioritize?
Is it A maximizing AI efficiency,
B implementing advanced algorithms,
C ensuring compliance with data
protection laws or D reducing
operational costs.
The answer is C ensuring compliance with
data protection laws.
Ensuring compliance with data protection
laws is crucial to align with legal
standards as these laws govern how data
is used and protected.
During an AI project review, a team
realizes their AI model may
inadvertently discriminate against a
minority group. What is the most
responsible action to take? Is it A
proceed with deployment and monitor outcomes?
outcomes?
B, ignore the issue as it affects a
small group. C, consult legal advisers
for potential liabilities,
or D, pause deployment to address and
mitigate bias.
The answer is D. Pause deployment to
address and mitigate bias.
Pausing deployment to address bias
ensures that the AI system does not
perpetuate discrimination. Aligning with
ethical AI practices,
deployment, scaling, and change
management considerations.
A company is deploying a gen AI solution
and needs to ensure smooth integration
with existing systems. What should be
the primary focus to achieve this? Is it
A training employees on new AI tools
underscore B revising company policies
to include AI ethics?
C ensuring platform compatibility with
existing systems
or D increasing budget allocation for AI projects.
projects.
The answer is C ensuring platform
compatibility with existing systems.
Integration with existing systems
requires a focus on platform
compatibility to ensure seamless
Thanks for watching. If you found this
exam review helpful, be sure to
subscribe for more real world practice
and topic focused study videos and visit
birdsy.ai to start your free trial of
انقر على أي نص أو طابع زمني للانتقال إلى تلك اللحظة في الفيديو
مشاركة:
معظم النصوص تصبح جاهزة في أقل من 5 ثوانٍ
نسخ بنقرة واحدة125+ لغةالبحث في المحتوىالانتقال إلى الطوابع الزمنية
الصق رابط YouTube
أدخل رابط أي فيديو YouTube للحصول على نصه الكامل
نموذج استخراج النص
معظم النصوص تصبح جاهزة في أقل من 5 ثوانٍ
احصل على إضافة Chrome
احصل على النصوص فوراً دون مغادرة YouTube. ثبّت إضافة Chrome للوصول بنقرة واحدة إلى نص أي فيديو مباشرةً من صفحة المشاهدة.
