Q1) How does @inject_eu work? #AIChat
— Nick Tang (@nickhtang) January 11, 2019
https://platform.twitter.com/widgets.js
A1 So, INJECT is a new digital tool to support journalists to write more original journalism, more quickly than now. It combines NLP, information retrieval, entity extraction and recommender tech to present journalists with content to inspire new angles on stories #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1 followup: Does someone sort through the information INJECT gathers to make sure that there’s no bias in the article? #AIChat
— Nick Tang (@nickhtang) January 11, 2019
https://platform.twitter.com/widgets.js
A1. The team maintains an editorial board to manipulate news creatively from different better researched sources #AIChat. But creative ideas can come from all sources, so more sources are better…
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1 followup: It is hard to find journalistic content without ANY bias. Since we are all (still) humans, everything we write is based on our upbringing etc
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
INJECT could help journalists to overcome their own biases – in helping the to see "opposing" things #AIchat
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
A1 The heart of INJECT is a set of creative search algorithms that operationalize expertise of experienced journalists in strategies. User choices to explore, for example numbers and evidence, or human angles, or comedy, invoke different creative search algorithms #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1. Unlike robot-journalism and other #AI uses, INJECT does not seek to replace any key element of a journalist’s work. Instead, it uses different technologies to enhance/empower journalists operating in less and less time, and fewer resources. Journalists control the AI #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1 Our INJECT team are currently designing new algorithms to manipulate the creative news index (now over 10million articles) to discover new, evolving, breaking and under-reported angles on topics. These new features should be available by the summer #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1 In background, 24/7, INJECT generates creative indexes of news stories. Using RSS, it reads and makes sense of about 15000 stories published each day from 300+ titles in 6 languages, and builds a rich index from which to generate creative guidance for journalists #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A1 Furthermore, to support hyper-local and specialist journalism, INJECT’s algorithms can read and make sense of local news and other types of archive, so that INJECT’s creativity support can exploit these information assets #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
Many journalists are afraid of algorithms and #AI – but INJECT's mission is to create AI tools that support journalists #AIchat
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
As we’re talking Q1, here’s Q2 for Professor @NeilMaiden and @the_claus How does @inject_eu help journalists and their newsrooms? #AIChat
— Nick Tang (@nickhtang) January 11, 2019
https://platform.twitter.com/widgets.js
A2: #AIChat. INJECT is different from other search engines. Engines like Google look to return the best-fit that you asked for. INJECT takes what you ask for, and gives you something a little different. Many people struggle with idea, at first. I think of it as ‘Googlisation’
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2 Journalists also use INJECT in unpredicted by creative ways. For example some freelancers use it to prepare for interviewing people, others to discover new stories and outlines to pitch to editors #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2) NYT or WaPo have great archives – and if you use it right it can be a goldmine. But smaller news outlets often don't have to resources to do that. INJECT could provide help here #AIchat
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
A2 #AIChat By ‘smart’ archive, we mean that we attach machine-readable semantics to published articles. We can extract content, tag it with inferred meanings, and manipulate these INJECT inferences in the tool’s reasoning engines
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2 #AIChat. Yes, Claus raises a key issue. Many media organisations are seeking to make smarter use of their #news archives, and journalists also want to be inspired by what they themselves have written before
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2. Our studies with INJECT reveal that INJECT works better to support more original feature articles. So our recent focus has been to support news and media businesses focused on quality #journalism. #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
Follow up Q in response to A2: It's great that you are surfacing smaller publications!! Given the problems we've had with fake news lately, how do you ensure the news (source) is credible? #AIchat
— Mia Dand (@MiaD) January 11, 2019
https://platform.twitter.com/widgets.js
A2 #AIChat. Mia – we have chosen a set of over 300 publication titles in many countries across 3 languages, to reduce the risk of bias, echo chambers etc. But this approach is not foolproof. So INJECT seeks to empower journalists to avoid ill-researched sources
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2: Adding to that: Even big new organisations get things wrong. So far there is no AI to solve this #AIchat
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
A2: Plus: With INJECT also smaller publications can make us of AI and machine learning. With our tool they can turn their "dumb" archive into "smart" archive #AIChat
— Claus Hesseling (@the_claus) January 11, 2019
https://platform.twitter.com/widgets.js
A2 Our studies of INJECT newsrooms have shown that its use improved the novelty of articles written by journalists over a period of months. Journalists generated not only new angles but also new content and new forms of content #AIChat
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
A2) and an essential one to keep in mind. Even with the AI software, it’s still up to the reporter to make sense of it all. In order to create a well balanced article. #AIChat
— Nick Tang (@nickhtang) January 11, 2019
https://platform.twitter.com/widgets.js
Well-said @nickhtang I am a big fan of quality journalism and any technology that helps surface content from the smaller publications that get overshadowed by the big outlets…!!! #AIchat
— Mia Dand (@MiaD) January 11, 2019
https://platform.twitter.com/widgets.js
#AIChat One wider observation is that journalists seek more original news stories. This requires #creative thinking of different forms. In this sense INJECT is different for uses of machine learning tools in news
— Neil Maiden (@NeilMaiden) January 11, 2019
https://platform.twitter.com/widgets.js
#AIChat Thanks to Nick and Mia. You can find out more about our INJECT project and product at https://t.co/AmHC0aW6Zx
— Neil Maiden (@NeilMaiden) January 11, 2019