AI in practice: Immediate Media

Q: What is the current state of play with regards to AI usage?

A: Like many, we’re working through the challenge of implementation, helped by four years of company-wide innovation and learning. Experimentation days, hackathons, our innovation fund, training programmes and open access to tools have built a culture where AI thinking is championed widely, rather than confined to a few selected teams. As a result, we have identified clear priorities and defined where we will not use generative AI at all.

Our mission is impactful but always responsible application, and underpinning it are our usage guidelines, reviewed at least every quarter by a governance group from across the business, alongside external statements so audiences know how and when we engage generative AI in our processes, products and services.

That’s not to say implementation is a breeze. Like the rest of our industry, we’re squinting into the big topic spotlights of licensing, the steady decline in search referrals, and the near-impossible business of cost modelling, alongside big tech teases about screen-free devices (doughnut-shaped assistants?).

This year, we’ve been setting up large-scale projects that will help us move quickly and sure-footedly. We’re structuring our owned IP across digital, print, audio and video so it’s accessible by man and machine, with rights, metadata and version history attached. On top of that, we’re building a content studio for our creators where AI will be part of certain workflows, covering research, creation, maintenance, distribution, insights and planning.

Parts of this are already live. For example, our drafting tool pairs AI with only our own published archive and returns first drafts with clickable citations, so an editor can see where every line came from, and it’s in regular use by several editorial teams. Newer, and still bedding in, is house style and fact checking support for writers and sub editors, so human editors can spend time enhancing content rather than fixing the same errors repeatedly. In addition, we’re rolling out AI to do the necessary but unglamorous work of creating the ‘furniture’ in our CMS, the metadata, internal linking, footers and alt text.

Beyond content, a newly formed acceleration hub gives teams access to experts to move projects from experimentation to reality, and work is already underway with our advertising team. We’re updating our data platform so trusted data is easier to access, particularly as we shift from pageviews to engagement as the real indicator of success. And, based on internal surveying to understand need, we’re relaunching our peer-to-peer community programmes and AI skills training for 2027.

Q: What are the best opportunities for future development?

A: We’ll keep using AI to give our creators more time back to create. One example is digital archive maintenance, where agents monitoring relevant topics could flag the pages needing a refresh. This would save our teams hours and help us keep all content up to date, rather than only the most popular pages.

Soon, AI will help our teams and our audiences find what they want far more easily through natural language search, with returned answers structured to give the most useful experience. Some of the big hypotheses in the industry suggest the days of creating a flat article for just one type of reader are coming to an end. We should be able to meet audiences where they are, with information pitched to how much they already know about a topic, and context used to shape the most relevant response. Even the choice as to whether to read, watch or listen will be theirs, whether that’s on our platforms or off them on new devices.

A combination of automation and expert curation will also allow for much richer experiences. Whether that is recipes and meal plans that can easily adapt to specific diets and health goals, geo-targeted notifications when in places of historical significance, or second-screen viewing companions, there are a whole raft of new features, products and experiences it’s now possible for us to credibly pursue.

The future is very much unwritten, and we’ll be watching closely for where the pivots need to happen in our thinking. Creative potential, community and audience experience will continue to be at the heart of all our decisions when applying AI.

Q: What are your three key learnings?

1. Cut the jargon. You can show off everything you know about AI and lose a lot of people along the way. Build with the teams doing the job rather than for them. Not everyone will want to come, and that’s fine, but silo the thinking and you’ll struggle to get meaningful adoption.
2. Finding use cases is just the start. We surfaced plenty of individual uses, but the opportunity stayed with the individual. Transformation happens when you can safely enable AI in the systems where teams are already working and connect it to the platforms they publish to. AI is usually part of a fix rather than the whole fix, and for more deterministic work, might not be the answer at all.
3. Keep evolving your thinking and be honest. Be specific about what you will and will not do with AI, and don’t dress up the problems. Be willing to try things and either scale or fail fast. We have stopped several experiments that weren’t a good fit and kept the learnings for future projects.