# Intro to AI FAC alumni event --- ## What is AI? :thinking_face: Artificial intelligence, machine learning, deep learning... ---  --- Artificial intelligence is the simulation of human intelligence processes by machines, especially computer systems --- Using algorithms to allow machines to solve problems which were previously solved only by humans --- Machine learning and Deep Learning are subsets of artificial intelligence pertaining to the way in which these machines learn --- ## Machine Learning Algorithms with the ability to learn without being explicitly programmed. --- Here's a non-glamourous example: chatbots. --- ## Deep Learning Algorithms inspired by the brain's structure. Able to o deal with more complicated “cognition” tasks such as image recognition, text-classification, self-driving cars, and more. ---  --- ## ML models Machine learning models are used to make predictions --- Low-medium impact - What is the chance of rain tomorrow? - How many people will buy this product? - Is this email spam or not? --- High impact - How likely are you to default on your loan? - How likely are you to commit a crime or re-offend? - How employable are you? --- ## How do these models learn? Broadly speaking, the predictions can be summarised into two types: supervised and un-supervised learning --- Supervised Learning: Where algorithms learn patterns from existing data and apply them on new data --- Unsupervised Learning: Where algorithms discover general patterns in the data --- ## Unsupervised vs. supervised learning Supervised learning uses labeled datasets, whereas unsupervised learning uses unlabeled datasets. “labeled” means that the data is already tagged with the right answer by humans. --- In supervised learning with labelled data sets we have the opportunity to objectively evaluate if the model has given a correct answer or not. --- In unsupervised learning, the model studies unlabelled data points and surfaces patterns or groupings. It is not told exactly what to look for --- ## What is algorithmic bias? The systematic discrimination that can occur in machine learning algorithms due to the presence of bias in the data used to train them --- Bias can be introduced into an algorithm in a variety of ways, including through the selection of data used for training and the choice of features used in the model. --- This can result in the algorithm making decisions that discriminate against certain groups of people, based on race, gender, or other characteristics. --- How did we get here, Anni? --- ## Resources [Weapons of Math Destruction - book](https://en.wikipedia.org/wiki/Weapons_of_Math_Destruction) [AI definition and uses - short video](https://www.youtube.com/watch?v=nASDYRkbQIY) [Coded Bias - Netflix film](https://www.netflix.com/gb/title/81328723) **Articles** [Data demystified](https://www.datacamp.com/blog/data-demystified-the-difference-between-data-science-machine-learning-deep-learning-and-artificial-intelligence?utm_source=linkedin&utm_medium=organic_social&utm_campaign=221128_1-differences_2-mix_3-all_4-na_5-bl_6-data-demyst_7-na_8-ogsl-li_9-na_10-bfcm22_11-n/a) [Machine learning with real life examples](https://medium.com/analytics-vidhya/machine-learning-with-real-world-examples-3e79877d08b3#:~:text=An%20example%20of%20unsupervised%20machine,eggs%20tend%20to%20buy%20bacon.) [Supervised vs. Unsupervised learning - 1](https://www.springboard.com/blog/data-science/lp-machine-learning-unsupervised-learning-supervised-learning/) [Supervised vs. Unsupervised learning - 2](https://www.alteryx.com/glossary/supervised-vs-unsupervised-learning#:~:text=Supervised%20and%20unsupervised%20learning%20have,tagged%20with%20the%20right%20answer.&text=A%20classification%20problem%20uses%20algorithms%20to%20classify%20data%20into%20particular%20segments.) [The Negative Feedback Loop: Technology Needs To Know When It Gets Things Wrong](https://youthedata.com/2018/05/23/the-negative-feedback-loop-technology-needs-to-know-when-it-gets-things-wrong/)