I’ll start by saying that I am very interested in anything that helps me to do my job faster or smarter. It’s also worth saying that I think a lot of tools that are described as being powered by AI are in fact simply automation tools. However, since they also help with doing work faster, I’m on board.
My first forays into using AI tools were pretty basic; things like social listening tools that use natural language processing to try to determine sentiment (pretty poorly at the time). But things have advanced since then and my eyes have been opened to a wealth of tools and applications of AI, particularly in journalism right now, that make me excited for future uses.
In my opinion the journalism industry seems far more advanced in its creation and application of AI than the PR industry. For this reason, most of the examples are journalistic applications however I believe, as I’ll come onto, that these also have use-cases for PR professionals.
One of the forces behind the ability for the journalism industry to invest in AI applications, especially within Europe, seems to be Google’s Digital News Innovation Fund (DNI).
The Digital News Innovation Fund (DNI Fund) is a European programme that’s part of the Google News Initiative, an effort to help journalism thrive in the digital age.
The DNI has funded a wealth of really interesting projects, some of which I’ll highlight in more detail. I’ll be honest running through them was a rabbit hole mission that I spent way too long on, so be warned in case the same happens to you!
Current applications
Fighting Fake News
The DNI has funded several projects which are aimed at ‘Battling misinformation’, a couple of interesting ones are below.
Full Fact is the UK’s independent fact checking charity which received two rounds of DNI funding. The first round was to support Full Fact’s automated fact checking abilities allowing them to fact check content at a much larger scale.
The second round is explained by its Head of Automated Fact Checking:
“The project will help us develop a set of standards for sharing fact checks and evaluate automated fact checking to ensure that the systems provide results in a consistently balanced way,” Mevan Babakar
From what I can see Full Fact started out mainly fact checking live broadcasts but now also investigates general news and, since January have announced they will now be fact checking Facebook posts
Other projects that fall into the ‘Battling misinformation’ category are around verifying user generated content (UGC).
The Associated Press (AP) received a grant for the project AP Verify which utilises AI to verify UGC based on various criteria to see whether it is legitimate content worth including in a story or there are factors at play to indicate otherwise.
Verifeye Media is a similar project that received funding, journalists can upload photo or video content straight from their smartphone and receive a confidence indicator for whether the content is safe to use or requires further review.
Local news resurgence
Several DNI funded projects are aimed at helping journalists to tell hyper-localised stories, one example is Reporters and Data and Robots (RADAR).
RADAR is a partnership between the Press Association (PA) and Urbs Media that enables regional news titles access to stories jointly written by journalists and AI.
This uses open source data to generate localised stories; a journalist will write a template of the article and the AI uses natural language generation (NLG) to then generate individually localised stories based on the template and data set. RADAR enables around 250 stories to be generated from just one journalist-written template.
Driving efficiency
A non-DNI funded project this time, The Washington Post created its AI technology Heliograf which has been dubbed ‘robot reporting’ and writes short form articles and social media messages mainly reporting on sports.
The Washington Post are certainly not alone in this though. The AP in the UK developed Wordsmith which has hugely increased their output vs manual efforts:
On another track, a DNI funded project Trint uses AI to transcribe and translate audio or video content. Some of the main benefits of this are:
- Saving time – it can transcribe a 45-minute audio clip in less time than it takes to play
- Saving money – it charges a flat fee far lower than most manual transcription services
- The output is all time-coded meaning you can search within the transcript to find the right parts of the video/audio, click to listen to them, and edit as you wish.
This I see being hugely useful not just for journalists but also for PRs and all other content creators who are working with audio and video. Especially relevant with the explosion of Podcast popularity now.
Finally, news.bridge takes this even further than Trint allowing you not just to transcribe and translate your audio, but it’s AI will also produce voiceovers, subtitles and summaries of transcriptions and translate all of this into target languages for you.
Finding and making the news
Most of the examples in the previous section were utilising AI to produce fairly formulaic content based on data you find yourself and feed in, as opposed to necessarily helping you to find the story. However, there are some examples of publishers that using AI to do exactly that.
Reuters launched Lynx Insight which aims to suggest story ideas to journalists as oppose to write the copy, although it can also help with that. These snippets are taken from a Wired article.
Forbes announced a similar tool last year; Bertie.
BuzzFeed also notably used machine learning to create a story identifying spy planes.
Building bots
Its not just writing the content or suggesting topics that AI is being used for though, it’s also helping to create different formats for content.
The BBC’s bot builder has enabled journalists to transform long form content into more engaging and easier to consume content through in-article chat bots.
It’s not just chatbots either though, Chinese media outlet Xinhua went a step further and has created AI news anchors!
A few days ago, they now introduced the world’s first female AI news anchor as well:
Democratic accountability
As well as the tools I mentioned aimed at battling misinformation which can often be linked to politics and public bodies, the DNI has also funded projects such as Alveteli which has a suite of tools aimed at holding public representatives to account.
What’s Google’s play?
If you’ve got to this point you might be thinking; what’s in it for Google? They’re investing all this money in all these projects so there must be something in it for them. As nice as it would be to think they are doing it simply to help the journalism industry, I believe they may benefit in other ways. A couple of examples being.
YouTube
As mentioned, a number of the projects involve video, which Google is very keen to push for YouTube’s sake and as such also has GNI YouTube Innovation funding with 87 projects funded across 23 countries. The obvious benefit for Google here being the more content being uploaded to YouTube, the more views they get and the more advertising revenue they can generate.
Assistant
Google has been heavily pushing its assistant and voice search over the last couple of years and a lot of the DNI projects also support this.
The projects that centre around translation and creation of localised voiceovers are certainly beneficial in this area.
The place for humans
As you’ll see from most of the projects I’ve mentioned, these are not centred around AI replacing human jobs, but complementing the jobs and reducing the repetitive or unnecessary tasks humans were previously performing that can be performed by machines to free up human time to be better spent.
There are certainly some areas that desperately require human monitoring and/or intervention for the foreseeable future.
Ethics
The first of those would be ethics and I think we have been able to see there are some staggering issues around AI and ethics. We definitely still need humans to decide the ethical considerations of the applications of this technology.
One of the most important elements of this is the transparency to consumers of how and when AI is being used.
Last year you may remember this story where a flawed algorithm led the UK government to deport thousands of students incorrectly. The reason for this was a BBC Panorama study that looked at how many students were being able to get study visas illegally, essentially by getting other people to fake their English proficiency test. The government then employed a firm to review these applications and identify ‘invalid’ results which they then served deportation notices to. The problem was the way in which the firm reviewed the test was using voice recognition software which was flawed and incorrectly marked tests as invalid (i.e. faked) when they weren’t. This meant thousands of students were incorrectly marked for deportation when they should not have been and sparked mass legal action against the government.
Editing and refining
As you can see from a lot of the examples that I’ve shown, the content that’s being generated using NLG is fairly basic. Human journalists are still needed for anything more in-depth than simply reporting on facts or figures.
So, the AI might be coming up with the story ideas; analysing data to provide the angles that are interesting, but it still takes a real human journalist and to craft that into a story that is compelling and engaging to read. I think even if we see some of these practices coming over into the PR world, again, it will still take a PR person to really find the right hook and to make it appeal to humans.
The Future?
So, what does the future hold for AI in the context of journalism and PR?
Better AI writing assistants
I think the AI writing assistants that we’ve already seen will continue to improve and I think we’ll start to see them coming through in other CMS, not just in journalism. I see assistants in CMS helping with things like; finding multimedia to insert, finding sources and facts that we might want to include in content, being able to click a button to factcheck in the background etc. I think that’s going to continue and we’re just going to see more and more features being included in CMS platforms to help people produce content more efficiently.
Improving accessibility
I think we’re also going to see AI helping to improve accessibility in a lot more ways; I mean that both in terms of making content more accessible for people with disabilities and impairments, but also helping people in different countries to access content.
I think we’ll start to see not just recommendations from CMS/content systems on how to improve accessibility, but the ability for the system to just improve it for you.
I’ve mentioned the projects around automated transcription and translation services and I see that as really exciting for PRs and Content Marketers to have the ability to increase the reach of any piece of content or campaign much wider than previously possible due to budget constraints.
Fighting fake news and filter bubbles
I think the fighting of fake news and filter bubbles is only going to become more and more and important and relevant in the years to come. Especially as we have potentially a general election coming up in the next year or so and certainly the US presidential election in 2020. All the work combating fake news and filter bubbles is going to become more and more important and we’re seeing a lot of investment going into those areas, so I think it’s safe to say that we’ll see some exciting developments happening there.
Creating stories from data
I think we’re going to see a lot more tools being created and applications of AI that help us to find and create stories. Rather than us finding datasets ourselves and feeding them through and it suggesting stories, I think we’ll see AI being used to profile audiences better, understand what they’re interested in and then serve up potential stories based on that, without us having to specify and feed it data sets.
Consumer scepticism
I also think that we’re going to see a lot more consumer scepticism though, I think things like the Cambridge Analytica scandal last year and all the data breaches and scandals that have can you come out of the news with Facebook and Google have made people very wary of how they share their data. So, I think we’ll see people being a lot more reluctant to share their data which may in some cases slow down AI development and applications if the data is not available to feed the machine.
More data training/hiring
Finally, I think one of the other things it’s going to be so important for the future is going to be more training for journalists and PR professionals on how to find data and transform it into interesting visualisations and stories.
I also think in terms of journalism, but also the PR industry, we’ll see more hires of people that have the capabilities to use data and to develop AI technologies and data visualisation to support the roles that they’re doing. I think that’s one area where PR is certainly lacking behind journalism right now.
What about PR?
The CIPR has the AI in PR panel who do some really great research into AI applications within PR. They conducted a great study looking at how AI is currently being used in PR and predicting what that my look like in five years.
The two diagrams below are taken from that report looking at the skills needed in PR and mapping against how AI can help.
As someone working in PR a lot of what I’ll be doing is keeping an eye on the tools that are being created and used by journalists and seeing if any of those come out of beta and are available to the PR industry too as there’ll be a lot of helpful cases for those.
For example, having technology that can help us to understand our client’s target audience, the stories that they’ll be interested in and the tools to gather information to generate those stories and angles is hugely valuable as a PR.
If the tools that currently exist aren’t made available more widely, I think we’re at a stage where those operating in the PR industry will have to start creating them themselves. We now know the technology is ready and capable to doing what we need it to do so replicating that for PRs will be key.
The Google DNI publishes a list of open source projects that can utilised so that is a really great place to start for developing our own tools.
This post ended up very long so congratulations if you actually made it to the end! I’d love to hear your thoughts on how you’re currently seeing AI being used in journalism and PR and where you think the future’s going too.
