Project

Teach machine to come up with factual, objective news summaries (text, audio, and visual) and optimize their entry points on home and assistant.

Role: Content Strategist

Process

• Identify subjective and offensive words for a block list.

• It isn’t until you read 5 summaries of the same story from 5 different reputable sources that you realize the slant in the news we read. Strict rules had to be created in order to obtain objectivity (or as close to it as we could get).

• Audit computer-generated summaries for guideline adherence, sentence structure, plagiarism, and readability.

• Work with engineers to teach machine nuance in both writing and reading.

• Explore different ways summaries could be reused within Assistant (i.e. in “people also ask” and queries for related people or places).

• While the machine can spit out a simple summary, it runs into trouble on more complex stories and has difficulty with vernacular. An exploratory phase was implemented to see if localized teams were needed in order for the product to be a success.