Media Makers Meet/AI and the media: Q&A with Rich Fairbairn, Glide Publishing Platform, UK. HomeAbout UsAgendaSpeakersSponsorsMeetVenueBlogContact usRegistration now closed Author: Cobus HeylPosted: 1st Feb 2023 (Updated: 9th Feb 2023) Blog navigationPrevious Blog: AI & the Media: Q&A with Dan Pacheco, Peter A. Horvitz Endowed Chair in Journalism Innovation, USANext Blog: Afternoon of Day 1 and pre-brunch morning on Day 2 The Di5ru.pt blog In preparation for an in-depth report on AI and the media, we spoke to several experts in the field via email. One such expert is Rich Fairbairn, Director at Glide Publishing Platform. How do you see AI’s impact on content operations this year? What are the things publishers should think of right now? Before we get into details it’s worth setting the scene a little, and clarifying a point for the pedants among us. Most of the AI people are talking about is not “true AI”, but more like Machine Learning and similar ‘educatable’ software. But we’re all calling it AI because, well, that’s what everyone is claiming it is in their marketing, their news reports, and everything else. It’s fair to say that outside of proper computational science circles, the rest of us just see it as a catch all term for ‘a leap forward in what software can do when it can learn patterns’. Even if you aren’t yet convinced that such AI could have much impact on or for your business now, it is at a tipping point for wider use in daily life for all your readers and audiences and many of your suppliers. Whether it’s packaged as chatbots, helpdesks, recommendations, better search, better ads, better healthcare, documents that write themselves, cars that drive themselves, whatever – there are going to be countless tools and services, often quite mundane, getting AI-like assistance. So why does that matter? Until now, it’s been easy to see anything badged AI as the preserve of massive businesses with oceans of money, data science teams, and years to plan. No more: we’re now in the era of cheap accessible AI that does boring stuff and can be used easily. Look how effectively ChatGPT has demystified AI for content businesses simply because it’s doing something we can understand: it writes things. In seconds, for literally fractions of cents. Boom! Just like that “AI in publishing” went from mostly intangible to something completely real and accessible. With that in mind, it’s now much easier to consider how AI could be useful in publishing. You don’t need to think of it as reserved for the highest-concept projects and budgets suitable for moon landings: you can start with tasks which are small and fiddly but vital, at low cost and commitment. If you used AI to, for example, only do better SEO page descriptions, you would be removing a fiddly task for a person and probably see an instant improvement in page performance to boot. Your writer can bring way more value elsewhere. Increment that to, let’s say, better article metadata, better image metadata, better choices of related articles, better paywall pricing offers – all things which can improve returns instantly and free up disproportionate human time, which is what we all want. We want AI to free up people and magnify their effectiveness. How do you see it develop in the shorter to longer term? What does that mean for newsrooms? At a high level we see two strands developing in the near term: New publishing start-ups, with non-industry funding, going all-in on AI to do things current publishers can’t do quickly or won’t risk doing. Think fully automated sites and services, new brands, with barely any human intervention. The existing industry players adopting AI tools in their current business models and brands, in bite-size chunks to do specific things like those mentioned earlier: efficiency not earthquake. The first strand could most visibly be of entire new sites or services that put AI content generation at their core, with people pulling the AI levers rather than intervening much on individual pieces of content: point your news bots at topics, have them prompt a ChatGPT-like service with a request, and out pops an article. Then what? Model around advertising and affiliate revenue and gobble up traffic by being quick and comprehensive, or find highly-targeted subscription and B2B channels where speed of information matters much more than grammatical flair or particularly special insight – results, events, and data-driven types of products. These disruptive entrants could be similar to what Axios or The Athletic were, coming from fresh waters and growing very rapidly, but it could just as likely be Facebook, or Google, or Twitter. Who knows, if they feel the cash-for-content deals being forced by legislators are too punitive, I’d be surprised if they do not look to their own versions of ChatGPT to procedurally generate news and answers and cut out publishers. Their battle – and in fact probably the AI battle in this space – will be for those providers to convince people their AI content is more trustworthy and accurate than other AIs, or has sufficient human oversight to be above criticism. In fact, ironically given how the question of trusting AI is a legitimate question, look for a drive by some to eventually claim their AI news is more trustworthy than that created by humans. What are some of the pitfalls you’ll warn publishers about? Why? Vendor overload, and a blizzard of “AI for everything” on offer from firms with no previous experience in the publishing space treating you just like any random business. Also, watch for an overreach of ambition or achievability from within your own organisations. In the same way we tell publishers they should not be CMS companies, you should remind yourself you are not an AI company. You are just looking for a tool that does a specific thing: the AI label is almost irrelevant. Importantly, it should be a way of improving something that you already understand well, can measure, and know how to improve. If you do not know what AI is going to solve for you, then you probably don’t need what you’re being offered. Elsewhere, expect this space to be extremely fast moving with lots of new entrants. That being the case, prioritise being able to pick up or put down these new tools quickly. If they are huge undertakings to integrate with your business or negatively impact your day to day operations, then they become significantly bigger risks. Like any new sector, there will be first-to-markets, and best-to-markets – there will be fierce competition from providers and vendors in price, performance, and easier adoption. Moreover, beware of extreme difficulty recruiting in the space if you choose to do so. All the more reason to search for easy to use tools that use AI, rather than being tricked into thinking you are an AI company. Thinking about Glide specifically, how do you think about AI integrations with the platform? For us this breaks down to how we use AI within GPP, and also how we allow customers to use other 3rd party AI tools alongside GPP. In terms of AI inside GPP – the closer you look, the more possibilities you see. The whole point of GPP is to remove friction from content businesses, so they can focus on what they are best at. Within that development objective, AI assistance for key tasks and outcomes is a natural addition so it’s not a fundamental rethinking of anything from our perspective. Our responsibility is to think 2-3 years down the line on behalf of publishers, who should expect us and providers like us to be hard at work on exploiting these new AI tools. And we are, just as lots of our customers are already using AI and ML tech. That applies equally to integrating with 3rd party tools. One enormous advantage a headless CMS or SaaS like ours has compared to traditional CMS is that all these new AI products are being built on the most modern principles of microservices, APIs, and interoperable formats: that’s what GP has been architected to exploit, so it’s a fantastic match for us. In layman’s terms, if your CMS is an old clunker, adding AI integrations is going to be of limited benefit. And more generally, what are some things publishers will see change with how content is managed from a technology perspective? I think this is really a question about where and how AI tools sit alongside human workflows. Certainly for the next 5-10 years we would expect the shape of traditional newsrooms and content operations to be much the same as now, but with AI tools giving help to authors and producers along each stage of the content journey, and improving website experiences for audiences and commercial teams. They should be freeing up time and adding value. For publishers worried about development costs and difficulties, I actually think they’ll have much less to do to implement all this than they fear because it will in fact be their vendors and partners who are using AI to improve their supplied services and products. Sit back and wait for it all to come to you may not be a bad strategy (if you have good suppliers that is!) That means us from the CMS side, smoothing off friction in content creation and leveraging intelligence within the platform to do much more for users and commercial and SEO teams. It means people like Getty doing tons more within image management and search and metadata. It means paywall partners helping craft more intelligent offers and subscription bundling. It means better commercial deals, better ad targeting, better data on your audiences and so on. Countless little things that add up. Of course there will be those all-new publishing operations created from scratch in an AI-first model whose workflows and structure don’t look anything like a current publisher. It would be naive to think that there will not be a rise in this kind of content business. The hard part to predict is how big or influential they will be: they will stand or fall on what they offer their audiences and grabbing market share. Once they are established and accepted, there will be no going back from them. Lastly, what would you advise people concerned about their jobs becoming obsolete? What should they do to remain relevant? I guess this is the 800lb gorilla sitting on an elephant in the corner of the room: will AI in publishing sweep away publishing’s people? I don’t think we are in a position to talk any more deeply than this than the likes of the World Economic Forum and Harvard and others who do studies on AI. The general consensus is that more jobs will be created than will be replaced, and I tend to believe that will be the case for publishing too. AI needs managing, it needs training, it needs improving, it needs redirecting, it needs overruling, it needs fixing, all things that will be turned over to people to do. And the better you are at those things, the more likely you are to find a better space for yourself within an AI-driven publishing future. And all those things need an understanding of content and audiences, and specifically your content and your audiences. What I think will be swept away are tasks – boring repetitive tasks that can be automated and removed from people’s responsibility. Instead, they will contribute higher level thinking and effort to the overall level of content they make, and to their business’s decision making. Why should a highly-qualified journalist be spending large amounts of time finessing SEO for multiple channels when a bot could do it in milliseconds? For a writer, is their time best spent on captioning that 7th picture in a gallery, or instead being able to send a breaking news story much quicker? Why should a producer be burrowing into picture libraries for a slightly better image to avoid using the same picture again and again, when a bot could pre-fetch 10 that are perfect and never been used before, and ensure the metadata is complete too? So across the spectrum of content business roles, at a high level what AI will be doing is allowing them to compete better and with less low-value drag from the kind of tasks we have all secretly wanted to be able to automate anyway: the dull but necessary. And what about the content itself? For sure, photographers and artists are seriously considering what image- and video-generating AI will mean for them, and it makes sense that writers will too. I think the landscape will be reshaped, but ultimately people still have this unique ability to know what other people will find interesting, and so far no-one has an AI which can beat humans for that attribute. If you look at TikTok, which could easily be described as an AI-driven business, their AI and algorithms calculate (better than anyone else!) what things their users want to see: the AIs still rely on the human creators, and still rely on humans consuming the content to create the signals of what is popular. The AI just measures what us humans like and gives us more of it, very quickly. What I’d say is that yes there will be AI content, but there will also be more space for non-AI content because it will be better, more interesting, more useful, and have a value as a result. *Thanks to Glide Publishing Platform for the support enabling us to develop the AI and Media report, and for always staying impartial in terms of how we develop our reports.