As I mentioned in a recent blog about how AI is more the purview of the CEO than the product team (for now), my Product Advisory Council was less than inspiring on the topic of AI. As the discussion unfolded, it changed from “We’re not doing too much with AI” to “Oh, we’re using it here. And here. Oh and here.”
There are a number of uses that are flying under the radar as they are specific to specific products or product problems people are working on. Sometimes these are best worked on in isolated cases instead of a holistic major AI strategy. And when I say AI, I include machine learning as the two are often conflated.
Let’s break some of these down:
Content creation: I’m sure many journalists are using generative AI for discovery and ideas, maybe even to investigate data. And, of course, there are companies such as United Robots creating content from scratch where there is reliable, repeatable structured data (housing, crime, and sports seem to be the top three areas).
Article summaries: Doesn’t have to be fully AI, but it can help a journalist or editor create a summary in seconds that can then be tweaked by a human. You can see my previous article on how Artifact has a fun version of this here, and I will soon be going a bit deeper on summaries with an expert in the field, so stay tuned.
Recommendations: A lot of people have been experimenting with personalisation under next read or even on the home page. This is becoming more and more sophisticated. And as the name implies, machine learning gets better the more data it has and the more it learns.
Text to speech: This has been around for a long time and is now so sophisticated that in major languages, you can buy trained voices off the shelf. Many companies are even integrating it into their major products so consumers can choose to read or listen. Join us on August 9 for a free to member Webinar on this. Apple’s new iOS (currently in public beta) lets you clone your own voice. Once it’s out in full release, there will be mountains of data to train it on. And yes, anyone will be able to do this and (as I will write about more in future, this is going to massively change our approach to audio).
Speech to text: As we create more audio-first products, we can also use the technology “the other way around.” Currently, most audio isn’t searchable and has to be manually summarised if at all. If we use text-to-speech technologies, we can apply many of the same efficiencies to audio that we do text: summaries, search, giving the consumer the ability to choose.
Translation: Very simply by using translation tools, we can make our content available to more people. When I spoke to Australian broadcaster SBS recently, they told me part of their mission was to appeal to all Australians — and that means multiple languages, both native and its history of building on immigration. New technologies enable them to do this quickly and cheaply (two traits public companies love ;).
These are things we can use now. Longer term, all of these can be built on. Imagine what we can do if generative AI is built into CMS? Imagine what we could do with repeatable structured data? Maybe modular journalism (which I wrote about here) could help with that. And another subject I am passionate about: personalisation. We can go beyond thinking about content topics to format and delivery.
Of course there is much to build out here, and we at INMA are looking at how we can pull together case studies on AI so that we can learn as a group. If you are working on any of this, I’d love to hear from you. Either to brainstorm, hear learnings, or get a case study published. Hit me up at Jodie.hopperton@INMA.org.
If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.