Blog: Big Data & AI for Media — The San Francisco Edition
Microsoft hosted the Big Data & AI for Media conference put together by World Newsmedia Network discussing some of the work the industry has been doing to leverage data. I had the joy of moderating the conference and wanted to share my top takeaways (in order of conference, not neccessarily importance).
- Media has a responsibility: to give the right information, without bias, to our audiences. There are large ethical issues we need to think about but the larger danger could be allowing unintentional small bias’ to manifest when AI, a self learning tool, is used.
- Data sets must be representative. If not we are working on false assumptions which at best could perpetuate current problems, as worst could have dire consequences.
- Teams do better when they are interdisciplinary and diverse. Time and time again we see that if we reflect the world, or at least the audience we are looking at serving, the results are better, stronger and more accurate.
- When we share data internally making it highly relevant to the audience, be it journalists, sales teams or clients, makes a dramatic impact on how it is welcomed and accepted.
- There are plenty of great tools around for managing and understanding data. Right now the questions deosn’t seem to be what can we do? More what should we do? Just because we can track a users activity and personal data from contacts to current location and much in-between, should we? The line hasn’t been drawn, and is unlikely to be by regulatory bodies any time soon.
- Whilst we have good tools, they are not perfect and there are still a lot of issues to solve around data and AI. We need a layer of HI (human intelligence). Recent examples include the problems Facebook faced with trying to take down all the footage of the NZ shootings, and YouTube’s ‘match’ of 911 with the fire at Notre Dame.
- Data should lead, or be a key part of, product and content development — from what I have heard, very few companies look at content itself, what are the factors that can be identified when people turn off/click away? Vertical Networks showed us that there is an opportunity for us to learn more by analysing format, execution and packaging.
- Media companies have a lot of first party data and some are putting this to excellent use, especially when layered with additional data.
- Maybe we need to re-evaluate sales process entirely. Both the San Francisco Chronicle and Sacramento Bee have pulled data to the heart of their sales processes shifting sales to client objectives, not to selling space. This is disrupting the agency model as it means working directly with clients.
- It can take time to build and sign off a big data and AI strategy. Sometimes that’s what’s needed… but big ideas can also start small and produce results to be built on.
- Historically we’ve thought of data as mostly text based but this is expanding, both in data available and analysis of, image, video and voice.
- And one thing that came through in all presentations: all successful strategies keep the customer at the core, what will improve their experience?