by Andrew Cochran

AUTOMATED JOURNALISM – News stories are prepared by machines acting within predetermined parameters. Examples are (a) repetitive editorial tasks, such as converting corporate earnings or sports scores into stories, (b) flagging breaking news, often by finding anomalies in feeds or data sources, or (c) identifying variances from previous statements or results.

AUGMENTED NEWSROOM – Intelligent machines routinely assist journalistic work.

AUTOSCRIBE – A computer system that generates stories from data.

CHATBOT – An algorithm that simulates one end of a conversation or exchange. If the material is about news, the algorithm might be called a newsbot.

DEEPFAKE – Synthetic video, audio, or text generated by an AI system intended to deceive users. For example, comments by a politician that they never said.

HYBRID WORKFLOWS – Humans and automated systems jointly contribute to an outcome. For example, algorithms gather editorial materials that are then selected by human judgment. See augmented newsroom above.

NATURAL LANGUAGE PROCESSING (NLP) – A computer operates on everyday words or speech. For example:

  • smart speakers that ‘listen’ to a request and then ‘speak’ a reply
  • transcription services that rapidly convert speech into text
  • translation services that rapidly adapt text or speech from one language to another

Sub-fields are Natural Language Understanding (NLU) and Natural Language Generation (NLG).

STRUCTURED DATA – Facts are formatted, making them easier to be processed by algorithms. For example:

  • Data classified by their type, such as date, headline, summary, language, etc., together often called a database
  • Data labelled by its content, such as dog, cat, car, polyp, eye, sneeze, laugh, etc., together often called a training set

In contrast, unstructured data has no apparent organization, such as words in a narrative.