Year: 2020
Type: report
Abstract: Artificial Intelligence (AI)-for the purposes of this report we define it as an umbrella term for a range of algorithm-based technologies that solve complex tasks by carrying out functions that previously required human thinking.Data -this term is principally used in this report in relation to 'training data'.This is the initial set of data used by AI systems through which they learn to base subsequent decisions or predictions.However, we acknowledge for some types of AI system additional data is collected through system use.Machine Learning (ML) -is a subset of artificial intelligence.Machine learning algorithms build mathematical models based on sample data in order to make predictions or decisions. Deep Learning (DL)-is a subset of ML that uses neural networks capable of learning unsupervised from data that is unstructured or unlabeled. Reinforcement Learning (RL) -after initial training the system continues to refine its decisions/predictions based on new inputs. Neural Networks -describes algorithms whose topology is inspired by the human brain.Supervised Learning -Supervised learning is when the training data contains both inputs and an output and the algorithm learns the mapping function between input and output.Unsupervised Learnin g -Unsupervised learning is when the training data contains inputs with no corresponding output.Bias -in this report we are principally concerned with algorithmic bias.This describes systematic and repeatable errors in AI systems that create unfair outcomes such as privileging one group over others.As AI is more widely adopted it will impact all aspects of our life including healthcare (spotting early signs of illness and diagnose disease), governance (to target interventions
Authors: Joseph Lindley, Paul Coulton
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3041229905