Artificial Intelligence

• It's a board field for the development of systems capable of performing tasks that typically require human intelligence
• Perception
• Reasoning
• Learning
• Problem solving
• Decision-making
• It's an umbrella-term for various techniques

Timeline

• 1950s
• Alan Turing proposing the Turing Test > A machine has intelligence if it's able to have a conversation with a human and fool the human into thinking that the human is speaking to another human

• John MacCarthy coins the term "Artificial Intelligence"

• 1970s

• Expert systems created

• MYCIN: AI rule-based system to detect bacteria

• 1990s

• Machine Learning & Data Mining

• 1997 IBM's Deep Blue defeats the world chess champion Garry Kasparov

• 2010s

• Deep Learning

• 2016 Google's AlphaGo defeating Go champion Lee Sedol

• 2020s

• Virtual assistants
• Autonomous vehicles
• Healthcare diagnostics
• Fraud detection
• Discussions on ethics and regulations

AI Components

1. Data Layer
   • Vast amount of data
2. ML Framework and Algorithm Layer
   • Use cases, requirements
3. Model Layer
   • Structure, parameters and functions
4. Application Layer
   • Interface to serve the model

Model workflow

1. Train a model with a training dataset
2. AI model is created
3. The model is used to evaluate a new data

AI vs. Machine Learning

• Not all AI systems are Machine Learning system, although the vast majority of them is
• For example the MYCIN expert system did not "learn from data" but instead it was built with a collection of over 500 explicit rules (yes/no textual questions) programmed by humans