AI Basics
In 2016, this AI system defeated world champion Lee Sedol. Can you name the program?
AlphaGO
Who invented the term "AI"?
John McCarthy
The three types of machine learning
supervised learning, unsupervised, reinforcement
Large language models (LLMs) are trained on these
What is a training token
A way of making computer, robot, or software think and act like human.What is Artificial Intelligence (AI)?
What is Artificial Intelligence (AI)?
Who is considered the father of AI?
Alan Turing
This type of learning happens when a computer is given both data and the correct answers, so it can learn from its mistakes.
Supervised learning
This is the process of breaking down text into smaller pieces so a training model can understand it
What is tokenization
A system that can do just one (or a few) defined things as well/ better than humans.
What is Narrow AI?
This time period was characterized by reduced funding and interest in AI research
What is the AI Winter
In this type of learning, a computer looks at data without being told what the right answers are — it finds hidden patterns all on its own.
What is unsupervised learning
Model consists of layers of nodes inspired by the human brain's neural structure
This is the ability to perceive or infer information and retain it as knowledge
What is Intelligence?
At this conference, AI was established as an academic field
Dartmouth conference
This type of learning is like training a dog — the computer tries different actions and learns from rewards or punishments to figure out what works best.
What is reinforcement learning
This structure in a neuron receives incoming signals from other neurons, functioning similarly to the input layer in a neural network
What is the function of dendrites?
Expensive to develop, not enough programmers, very few products on the Healthcare market
What are the disadvantages of AI in Healthcare?
When was the first chatbot, ELIZA created?
1965
These systems can recognize images, sounds, or patterns in data by mimicking how the human brain works.
What are neural networks
True or false - the purpose of a neural network is to predict the next token given a series of input tokens
What is TRUE
Deep fakes, disinformation, "jailbreaking" prompts to bypass moderators,
Intentional Misuse of LLMs
During the AI Winter, where were funds being deferred?
Space exploration
This is when a computer program is given feedback (like points or penalties) as it tries to figure out how to succeed at a task.
What is reinforcement learning
In this layer of a neural network, the actual "thinking" occurs.
what is the hidden layer
This is a machine learning system trained on large volumes of text
Chat GPT
Considered the first use of AI in HC, this system predicted antibiotics for bacterial infections
Mycin
Netflix, Siri, Alexa are all examples of this
Machine Learning Models
We refer to this when neither an LLM or its designers can explain outputs
The Black Box Problem