Generative AI Fundamentals
Background
Human memory is
enriched with lot of images or visuals which will easily identifies objects
from memory. Example if you have seen any remote either it can be TV, AC, etc..
We can easily
understand that is Remote. How? Our data stored that object as remote.
Likewise, computers
also retain data that can be processed when needed based on specific commands.
In essence, the functioning of Artificial Intelligence (AI) relies on this
stored data to operate effectively.
In Simple words How
we AI works, with help of data.
AI Landscape
1. Artificial Intelligence - AI
v What is AI? AI is making computer intelligence.
2.
Machine Learning – ML
v
What is ML? We try to learn the patterns and predict the
future.
1. Supervised learning -
Labelled data (Featured Data label)
2. Unsupervised learning
– Unlabeled Data (This is used for Random Analysis).
3.
Deep Learning - DL
v
What is DL? We try to learn the pattern and
predict the future in specific way.
1. Labelled data
(Featured Data label)
2. Unlabeled Data (This
is used for Random Analysis).
4.
Generative AI – Gen AI
v
It used normally Deep learning concept, so it
comes subset of Deep learning.
v
Generative is Unlabeled Data, It can generate
from data repository. This type of model called Foundation model.
v
It can be 2 types.
1. Language model (GPT)
2. Image Model (Mid Journey).
In short, we can say.
§ ML+DL is A system to Predict /Classify/Cluster
§ Gen AI is A system to Generate.
AI vs Generative AI
If we consider simple
formula, Y = F(x)
§
X is Input
§
Y is output
§
F is function to estimate here F will do job for
Predict /Classify/Cluster
But in Generative AI , F is the function to generate
Here we can generate
text, image, audio, video.
Generative AI
Generative Artificial
Intelligence is a kind of AI technology that can create new content, such as
text, images, or any other media when the user provides a specific prompt to
it.
It does this by
learning patterns from existing data and then using this data and knowledge to
generate unique outputs.
Generative AI is
capable of producing highly realistic and complex content that mimics human
creativity. This feature makes it a valuable tool for many industries, such as
gaming, entertainment, and product design.
Examples of Generative AI tools
v
Dall-E => it is used to create image from
text
v
ChatGPT => it is used to create text from
text (conversational model used here)
v Bard => it used
for composition like music,audio.
Generative AI vs Predictive AI
· Generative AI focuses
on creating original
and novel content, while predictive AI aims to forecast future outcomes based
on historical data patterns.
· Generative AI
algorithm focus to create actual content, where predictive AI algorithm uses to
create predictive analysis based on past data which is more used for weather
forecast kind of predictions.
Generative AI categories
2 types
2.
Image
based.
What is LLM
(Large language Model)
·
Trained
on very large data, Generic in nature.
·
Example
ChatGPT
·
Trained
on Millions and billions of parameters.
·
Example: Google
(PlaM), Microsoft (ChatGPT) , Meta have their own model
History
and Future of Generative AI
1.
NLP
(Natural language processing)
2. GANs (generative adversarial networks)
3. GPTs (Generative pre-trained transformers)
1.
NLP (Natural language processing)
It helps machines process and understand the human language
so that they can automatically perform repetitive tasks.
Example: chatbot, speech recognition, text summarization, and
machine translation
2. GANs (generative
adversarial networks)
It has 2 components.
1.
Generative – this component generates the data
based on input
2.
Discriminator – this component makes realistic output.
3.
GPTs (Generative
pre-trained transformers)
It has 2 components.
1.
Encoder
2.
Decoder
We can say example here Chat GPT.
Generative AI Benefits and Risks
Generative AI Benefits
· It can do everything a human can, and the
performance is similar to or better than that of humans.
·
This performs
Automation.
·
We can make Photo Realism,
example using Dall-E tool we can make very realistic images. This helps in
design area.
·
More Productivity,
example in business we can make documentation in short time with help of using
ChatGPT.
Generate AI
Limitations
·
Source
content - Quality of output is based on input content.
·
Bias
·
Incorrect
information
·
New
method to adapt take time.
Generative AI Data
privacy Limitation
·
Regulation
compliance
·
Data
leakage
·
Fake data
creation
·
Data
storage and life span
·
vulnerability
of data
How we can
protect Data privacy in Generative AI
·
Minimize
data collection – required data.
·
Aggregation
and Anonymization
·
Data
Polices
·
Encryption
·
Access
control
·
Auditing
and monitoring
The growing role
of generative AI tools in various industries
·
Video
Game making
·
Automated
design
·
Audio,
Video, Text, Voice creation
·
Text
generated area
·
Metaverse
Area of Generative
AI applicable
·
Healthcare
·
Education
·
Business
and Finance –
operation management, optimize data
·
Media and
entertainment
Generative
AI in health care
1.
Conversational
patient conversation and support - using
GPT
2.
Early
disease detection – image based.
3.
Enhanced
medical training.
4.
Medical
product development and Design
Generative
AI in Education
1.
Personalized learning experience
2.
Training material creation
3.
Learning assessment tools
4.
Personal tutor - one to one guidance
Generative AI in
Business and Finance
1.
Operational management
2.
Targeted ads and social media
3.
Personal finance – planning
4.
Presentation
Generative AI in
Marketing Advertising
1.
Market strategies
2.
Content creation, repurposing, localization
Generative
AI in Media and Entertainment
1.
Content personalization
2.
Immersive experiences
3.
Scalable content creation
4.
Art and media creation
Some
of Generative tools available in market
·
Dall-E
·
ChatGPT
·
GitHub Copilot
·
Google Gemini
Conclusion
Generative AI has emerged as a powerful force in the
technological landscape, enabling content creation and innovation across
numerous domains.
Here this article explained fundamentals of Generative AI.
Comments
Post a Comment