Last week, we were invited to speak at ConTech in London, a conference on ‘transforming content through data science, AI and emerging technologies’. The focus on case studies was high, and the conference creates as such an opportunity to learn what’s out there and how to start. The following statement by Freddie Quek from Solera expresses this well: we have many AI solutions, but what is the problem? We took this opportunity to discover the AI solutions out there and which problems they’re trying to solve. It allows you to get to know the market and the many perspectives people have on it.
Finally in the “Harnessing” session we have Freddie Quek @FreddieQ, VP Software Engineering at @SoleraInc asking what intelligence is required to make sense of the ocean of data we are creating? This is the stuff we need to know! #NeedleInAHayStack #DataScience #ConTech2018 pic.twitter.com/NAljsqsUQh
— ConTech (@ConTechLive) November 30, 2018
One of the concerns brought forward is the one of biased content and “fake news”. AdVerif.ai talked about how their algorithms are trying to address this. In the end, the models will only be biased, if what you’re feeding them is biased. So any bias should always be addressed in the data itself in the first place. As such, the well-known principle “garbage-in, garbage-out” still applies. Concerning this topic, the question of whether we should be able to explain the models also came up. If we recommend specific articles to readers, should we be able to inform them about ‘the why’? As with everything, it all depends on context, on the particular application and on how people are informed.
And finally in the other room we have a real insight into what #AI + #NLP have to do with content creation automation by Sabine Louet, the CEO of the fantastic @SciencePOD 🤓👌🏻 Is there a new generation redefining #NLP + even where we are most using our own “natural language”? pic.twitter.com/TxSQzZuAOm
— ConTech (@ConTechLive) November 30, 2018
Another concern which came through, was the fear for automatic content generation and selection and how it will affect the jobs of the editorial team. Well, don’t worry, according to Sabine Louët from SciencePOD. The jobs will stay, although they might change. Content selection, summarization and even generation can help journalists so that they can more quickly process information or help free up time to focus on the most interesting and valuable content. This was also expressed by our AI Panel at the Digital Growth Summit, where United Robots and Syllabs spoke about the articles their robots write. Similarly, we try to improve the way content is distributed by personalizing newsletters for readers of The Times and The Sunday Times. At ConTech, we explained how “James, your digital butler” is able to learn reader habits.
And in the other room is the session on appraising the opportunity for automated content creation + distribution starting with Jasmein Lismont from the brill @TwipeMobile on the “JAMES, your digital butler” collab with @thetimes + @thesundaytimes 💡👌🏻 pic.twitter.com/NjzudYWlOM
— ConTech (@ConTechLive) November 30, 2018
Still many things to learn, but also already many successful projects and startups on how AI can disrupt and transform the news industry.
Thanks to Jasmien Lismont, Junior Project Manager and Data Scientist at Twipe, for this article on her experience at ConTech 2018.
