Day 1 AI vs ML vs DL
AI vs ML vs DL — And Why It Matters for Security
Today, I explored the differences between Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) — terms often used interchangeably, but each has its own level of depth and scope.
🔁 Analogy to Simplify It
AI is like a Country
It’s the broadest concept — enabling machines to mimic human intelligence.
Example: A robot that can clean your room by understanding voice commands and navigating space.
ML is like a State within that Country
A subset of AI where machines learn from data to make predictions/decisions without being explicitly programmed.
Example: Image recognition, recommendation systems.
DL is like a Province within that State
A specialized type of Machine Learning that mimics how the human brain works — using layers of “neurons” to learn from complex data like images, sounds, or videos.
Example: Self-driving cars that understand images, voice, etc.
They're not separate technologies — they're nested within each other.
🔒 Looking Through a Security Lens
These technologies don’t just power amazing innovations — they also open up new attack surfaces:
AI – Input Manipulation If someone gives confusing or malicious voice commands, the robot may malfunction or perform incorrect actions.
ML – Adversarial Attacks A slightly altered image could fool the model into misclassifying (e.g., a cat image classified as a dog).
DL – Poisoned Training Data If attackers inject bad data, the model may learn incorrect behavior (e.g., ignore stop signs). 👉 Interesting: DL models are often “black boxes,” making failures hard to diagnose or fix.
As these systems make critical decisions, understanding and securing them is no longer optional — it’s essential.
I'm starting from scratch — but by Day 100, I aim to become a Subject Matter Expert in AI/ML Security.
📢 Follow along, share your favorite resources, or drop ideas below. I’m all ears!
📚 Resources
Day 1/100 — Beginning My #100DaysOfAISec Journey
Last updated