How Generative AI works
Data Sourcing in Generative AI
Innovation and Transformation with Generative.AI and how it works
Data Sourcing in Generative AI
- Data Scraping from the Internet
- Extracting data from websites and databases
- Structuring and verifying the data
- Data Processing, creating Parameters
- Gen-AI learns data patterns, behaviors, and relationships.
- Creates PARAMETERS that network the above
- Parameters are the final data. The original content is in
no longer used in the AI.
- The Artificial INTELLIGENCE is neural
- Patterns, behaviors, and relationships. like the brain
From Parameters to Text Output
Innovation and Transformation with Generative.AI and how it works
From Parameters to Text to Mind
1) Parameters – The AI’s Knowledge Base
- The training results in Trillions of parameters.
- Parameters are highly compressed.
- GPT-4 has 1,76 Trillion parameters.
2) Generating Text
- AI doesn’t retrieve content from a traditional database.
- It uses parameters to generate text on-the-fly.
- Builds relations to ~600,000 words (0.00003% of data).
- Which we can read – and vice versa
Implication
- The “source” of the AI for its process is the prompt
Traditional ERP vs. Gen-AI ERP
Innovation and Transformation with Generative.AI and how it works
Enterprise – Traditional vs. Gen-AI
1) Traditional Order Processing
- Linear Process:
Order entry → Inventory check → Payment → Shipping. - Manual intervention often required.
- Based on pre-set rules.
2) Gen-AI Order Processing
- Dynamic Lateral Process:
AI can predict inventory needs, plans purchases,
optimizes payments, includes fraud detection,
and perfects shipping routes and timelines. - Minimal human intervention.
- Constantly learns and improves from new data.