Automation and artificial intelligence (AI) have made inroads into almost every business – from banking, insurance, legal, and construction to travel, real estate, food, and media and broadcasting. The media industry is one of the sectors where AI is making the biggest impact. Take podcast transcription, for instance. Besides the time-tested option of using transcription outsourcing companies to manually convert podcasts or other audio/visual content into text, journalists can leverage AI to get the job done. Robots are even writing news stories. But will algorithms get it right? Is AI a boon or bane for journalism?
Automated News Writing a Growing Trend
According to a www.journalism.co.uk report, AI can help journalists do certain tasks better. It can free journalists up, giving them more time to follow stories and create better content. The report explains how Sky News worked with machine learning platform Graymeta featuring facial recognition technology to spot and identify guests arriving at the UK Royal Wedding 2018. The drawback is that such tools are not perfected and may make mistakes.
An article published by Knowledge@Wharton reported that a 2018 report by the Reuters Institute for the Study of Journalism on a global survey of nearly 200 top editors, CEOs and digital leaders showed that nearly three-quarters are already using AI. The article lists several expert opinions and experiences on the benefits of AI for media:
- Allows newsrooms operate more cost effectively by generating a large volume of stories much more quickly than humans
- Can mine and utilize data to generate thousands of stories a day in multiple languages
- Helps with time-consuming transcription of recordings
- Can churn out headlines and bullet-pointed stories very quickly
- As AI can generate multiple versions of basically the same story, it is a good option for financial reporting or earning reports where stories don’t change much
- Take over routine reporting so that journalists have time to chase more interesting stories
- Works well for stories that use highly structured data such as sports news
- AI can fill in templates with local data and help journalists produce hundreds of regional stories a week
- Provides automatic translation of news from any language
However, as the use of AI in journalism increases, there is also growing concern about its adverse impacts. The Knowledge@Wharton article quotes Bloomberg News editor-in-Chief John Micklethwait as saying, “We have … a new specter haunting our industry – artificial intelligence”. According to him, journalists are afraid that they will be replaced by robots, which will affect the quality of writing and lead to the proliferation of fake news.
Wharton marketing professor Pinar Yildirim points out that a big part of journalism is still about pluralism, personal opinions, being able to interpret facts from different angle, all areas where AI fails. Humans have the edge when it comes to judgment and interpretation of facts, events and news stories.
Automated Transcription
Journalists are also using automated transcription tools to convert audio/visual content into text format quickly. Transcription is the first step in putting the story together from raw data. Scanning transcripts helps journalists find key written segments quickly and easily determine what to include in the final story. Including transcripts with digital media stories support search engine optimization (SEO) by making the content visible in search results. Including a transcript and subtitles with content will help people engage with it when they cannot listen to a podcast or watch a video. Subtitling allows people to scroll through their social media feeds without audio.
A www.multimedia.journalism.berkeley.edu blog described the users’ experience with the automated transcription tool Trint. According to them, Trint has a user-friendly interface and comes with integrated keyboard shortcuts for easy navigation and convenient features like rewinding 5 seconds or highlighting/striking through. It also offered an impressive level of accuracy and could speed up workflow. The report concludes that while the software worked well with clear, isolated audio recorded by a professional microphone, it had certain limitations:
- Overlapping dialogue and ambient sound affected transcription accuracy
- The average user can find the custom embedding platform confusing
- It may be an option for those who don’t need a perfectly documented version of the audio
Other typical errors in automated transcripts include difficulty identifying speaker turns when there are multiple speakers, mistakes in names of people and places, non-recognition of punctuations, such as hyphens, quotations, and exclamation marks, as well as proper capitalizations and paragraph breaks. Transcription software also does not have language inference systems, with the result that it cannot make corrections based on context. Accuracy also depends on actors such as the speaker’s accent or and high quality audio.
With these limitations, it is obvious that manual transcription services are still highly relevant for media professionals. Outsourcing audio and podcast transcription will result in a high-quality, accurate text version of the files. Expert transcriptionists pay great attention to detail and can handle large files, multiple speakers, accents, and even audio challenges. The best option for journalists is to use transcripts of interviews, video content, and raw footage as first drafts. These can be sent to a reliable audio transcription service provider for editing.
We Need Human Analysis of Facts and Events
AI and automated transcription can help journalists produce voluminous content that might be what newsrooms are looking for. However, as experts point out, media should focus on creating value over quantity. The human element is necessary to produce stories that are “more investigative, more original, more creative in a way that would actually meet consumers’ needs… (Lewis Seth Lewis, chair in emerging media at the University of Oregon School of Journalism and Communication- Knowledge@Wharton).
