Notifications

Log in to see your notifications.

Artificial Intelligence

14 min read

What is artificial intelligence?

Artificial intelligence (AI) is the ability of a computer or machine to do tasks that normally need human intelligence, such as understanding speech, recognising faces, translating languages, making decisions and creating text or images. It is used in phone assistants, spell-checkers, recommendation of videos, medical diagnosis, farming advice and self-driving vehicles.

TermMeaning
Machine learning (ML)AI that learns patterns from data instead of following only fixed rules
Deep learningML using many-layered neural networks
Generative AIAI that creates new text, images, music or code
Natural language processingAI that understands and produces human language
Computer visionAI that “sees” images and video
RoboticsMachines that sense and act in the world

How does it learn?

How a machine learning system is builtData is collected and cleaned. A learning algorithm finds patterns in the training data and produces a model. The model is tested on new data it has not seen. Once it is good enough it is used to make predictions, and its results are checked and the model improved over time.DataTrainModelTestPredictfeedback: improve the data and the model

Measuring how good it is

Accuracy is one way to measure a model.
Worked exampleAccuracy of a plant-disease model

A model checked 200 leaves and was right on 180. Find the accuracy.

  1. Divide.

Answer: The accuracy is 90%. It still makes mistakes on one leaf in ten.

Benefits, risks and ethics

BenefitsRisks and issues
Saves time, does repetitive workMistakes and false or invented answers (“hallucinations”)
Helps in health, farming, education and languageBias if the training data are unfair
Works 24 hours a dayPrivacy: data about people
Gives personal help and learning supportJob changes; misuse such as deepfakes and cheating
Analyses huge amounts of dataEnergy use; dependence on technology

Use AI responsibly: check its answers against reliable sources, do not share private information, give credit and be honest about what you wrote yourself, and remember that humans are responsible for the decisions.

Models usually improve as they learn from more examples.
Try itModel accuracy

Choose the numbers of correct and wrong predictions.

Check yourself Machine learning means a computer
Check yourself Bias in an AI system can come from
Check yourself A responsible user of AI should

Key terms

Artificial intelligence
Machines performing tasks that need human-like intelligence.
Machine learning
Learning patterns from data.
Training data
Examples used to teach a model.
Bias
Unfairness in a model’s results.

Summary

  • AI performs tasks that normally need human intelligence.
  • Machine learning finds patterns in training data; test the model on new data.
  • Accuracy = correct ÷ all predictions × 100.
  • Use AI responsibly: check, protect privacy and be honest.
Digital Marketing