AI basics
This is the ability of computer systems to perform tasks that would typically require human intelligence
What is AI?
This company built an AI to screen resumes that ended up being biased.
What is Amazon?
When a model learns easy patterns in training but has low performance in real world
What is shortcut learning?
In the formula, this term represents the raw data you provide, such as the number of yellow pixels in an image.
What is input?
This popular type of machine learning algorithm is modeled specifically after the structure of the human brain.
What is an Artificial Neural Network?
Siri and Alexa are everyday examples of this kind of AI application.
What is virtual assistance?
This Microsoft chatbot was released in 2016 and learned from Twitter users.
When a model performs poorly in both training and testing
What is underfitting?
If a certain feature is very important to the prediction, the computer will assign it a high value of this.
What is weight?
These are the interconnected layers within a neural network that process and analyze complex data, like neurons.
What are nodes?
These three fields work together to build machines that have the ability to learn and adapt.
What is math, computer science, and statistics?
This moral came out of the Amazon bias incident.
AI doesn’t have a moral compass, it's just a mirror
When a model performs well on training data but poorly on new data, oftentimes due to too little or too similar data.
What is overfitting?
This part of the AI equation acts as an offset based on what the model learned during its training.
What is bias?
This specific type of AI creates original content like realistic video, audio, or long-form text rather than just identifying it.
What is generative AI?
Self-driving cars and medical diagnostics are examples that fall into this category of AI applications.
What are advanced AI applications?
What medical purpose has AI been used for?
Recognizing certain forms of cancer.
A model accurately classifies cows only when the image depicts a cow in a grassy field. Why might the model be doing this, and what needs to be improved in our training set?
background correlations, needs better data diversity
A classifier outputs: Cat: 0.78, Dog: 0.20, Car: 0.02. What does the 0.78 represent?
What is the model’s estimated probability for “Cat”?
To test a "Rock, Paper, Scissors" model's accuracy, you might introduce these, such as wearing gloves or accessories.
What are complexities?
Netflix and YouTube use AI for this purpose.
What are recommendations?
This real-world example of AI automation is commonly found in grocery stores and retail shops, allowing customers to process their own purchases.
What are self-checkout kiosks?
This happens when the training and testing data sets are too similar, resulting in high accuracy in the lab but low performance in the real world
What is data leakage?
To get the final output, you multiply the Input by the Weight and then add this third component.
What is bias?
What does the “GPT” of models like ChatGPT stand for?
Generative Pre-Trained Transformer