Please enable JavaScript.
Coggle requires JavaScript to display documents.
Generative AI, Agentic AI, predictive models, Plus Deep Learning, General…
Generative AI
types
LLMs
SLMs
LMMs
SMMs
frontier
foundation
emergence
transformer
machine translation
text
image
video
sound
new significant capability
approach
open source
Open AI ChatGPT
closed source
Open AI ChatGPT
Google Gemini
Anthropic Claude
supervised
unsupervised
re-inforcement
prediction of content based on input
the prediction of new content
Agentic AI
benefits
scalability
increased adaptability
efficient
autonomy and adaptability
no down time
goal-oriented
cost reductions
higher productivity
execute parallel processes
minimal human support
updated perspectives
learn
industries
supply chain
finance
insurance
healthcare
traffic control
share trading
cyber security
autonomous vehicles
Applications
pharmaceuticals and science
deepmind
protein folding / drug discovery
materials science
energy fusion
mathematics
Financial Services
customer experience
fraud detection
algorithmic trading
risk management
insure tech
regulatory compliance
pci-dss
credit scoring
cybersecurity
Telecommunications
personalised services
NBA
new applications
virtual assistants
AR
contact centres
chatbots
virtual assistants
network operations
performance loading
network optimisations
network configuration
performance monitoring
security
ddos attacks
software engineering
creative industries
arts
music
film
Attributes
proactivity
adaptability
collaboration
specialisation
Objective Driven
Perceive
Sensors
Databases
APIs
Reason
RAG
Act
Integrates
APIs
MCP
agent 2 agent comms
A2A
Learn
challenges
hallucinations
mt: prompt engineering
mt: controlled data
stochastic
untrue answers
mit: temp adjustment
mit: adjust p and k parameters
misinformation
disinformation
mit: RAG
mit: fine tuning
mit: train your own / fine tuning
supervised fine tuning
reinforcement learning from human feedback (RLHF)
mit: compare and challenge outputs
via prompts
inputting to other AI models
copyright
security vulnerabilities
jailbreaking
data poisoning
prompt injection
data breach
performance on unknown data
Approach
data gathering
processing
LLM
understand
mult-agents
horizontal
narrow skills
lateral communication
vertical
hieararchical
higher levels more power
use LLMs
lower levels
data collection
predictive models
approach
classifying
clustering
regression
input driven
suits
unsupervised
supervised
used for
predicting next data point
classify an object
loan approvals
medical diagnosis
Plus Deep Learning
General AI