Peering into the crystal ball of AI, the media, and the future

New report lays out the pathways of possibility in the relationship between audiences, the media, and content in an AI world.

by Rich Fairbairn

Published: 17:54, 29 August 2024

Last updated: 19:20, 29 August 2024

The problem with trying to navigate the emerging reality of AI in news and media is that no-one knows what direction things will go. 

It’s all happening so fast!

If you combine the speed of AI tech development, with the rapid swings of public perception of AI content and what it could be used for, with the hydra of opinions within media and publishing about AI tech and IP rights, and the global catch-up by lawmakers putting the ‘late’ in regulate, then you can understand why we are being swamped by the sort of change which took decades to play out for web and social media eras.

To help shed some light, and with wonderful timing for your beach reading, is probably the most thorough attempt yet to ask tough questions about the future of information, journalism, and media in the era of AI. And it comes to us recommended by publishing industry futurologist Simon Regan-Edwards, so it must be good!

Spoiler alert: it is.

The AI in Journalism Futures report saw researchers from Open Society Foundations work with nearly 900 respondents involved with journalism and “civic information”, and interested in the development of AI, to distill their thoughts and ideas about journalism, media, and AI.

The objective was to chart pathways in the thickening AI/media fog, via online and in-person events and workshops where a wide range of suggested scenarios were explored and assessed to derive some widely agreed likelihoods.

That word “likelihoods” is important, because like everything else to do with AI and media, no-one really knows yet. And like many forward-looking reports which cannot reach conclusions, there is a good chance it all becomes true in one way or another.

Five ways to see the future

The testing and probing of dozens of user-suggested scenarios saw them averaged over time into five overarching umbrella scenarios, each with a wide range of possibilities encompassed.

They all look eminently possible and – not surprising – have both negative and positive ways of being viewed.

The five themes in the report shape up as:

Machine in the Middle: We end up with AI-powered news gathering and personalisation tools combining to create source-to-consumer channels that do not depend on human input.

This is probably the most debated and explored hypothesis, as it proffers believable scenarios which vary from zero-human-oversight information bubbles, news orgs as information brokers because they have humans, personal AI news agents and presenters, and AI-to-journalist tools and systems which springload the effect journalists can have.

Power Flows to Those Who Know Your Needs: If AI makes it trivial for any person to be served any information in any format on any device, knowing what to actually produce for each consumer will become far more important and powerful. In other words, know thy reader is more important than ever.

The debates around this span the negative – if you think social and search platforms being even better positioned to scoop up all traffic is a negative – to really positive, where the people best positioned to know what ‘data’ is vital to good content are those who make the effort to go and speak to other people and to create said content.

Omniscience for Me, Noise for You: Social and demographic splits are magnified by access to particular AI-driven spaces and platforms. Aka, if you worry about people being in information bubbles now, wait until you get a loada this.

One of the more esoteric discussion scenarios centred around the power of AI – just as the internet at large – to serve different people in different ways. Essentially, if you have the right AI, you are at a massive advantage to those who do not.

Critically, the “right” AI is subjective to the user: the right AI for you might be the antithesis of right to someone else, just as can be the case with social media spaces, or being on social media at all.

As I have suggested before, I think this is where the content deals being done between platforms and media owners will become vital, not just for the use of the content itself to stop AIs descending into fugue but for the marketable badge of authenticity particular media brand partners will give to AIs. What do you think would be the bragging value of an AI “exclusively authorised by Wikipedia” for example? (Setting aside the assumption that plenty of AIs have pillaged Wikipedia already.)

AI with Its Own Agency and Power: Envisioning AI systems which operate in information ecosystems with minimal human oversight.

The thrust here is not sentient Terminator AIs making their own decisions, but rather about the potential direction organisations and systems could go in which humans design the information patterns and set the rules, and AIs execute the plan under increasingly distant oversight without anyone needing to do anything. One could argue that this is how modern social media and search engines already work.

AI on a Leash: Many of the potential opportunities and potential harms from deploying AI within the information ecosystem remain unrealised because of restrictions intentionally placed on its use either by state regulators or directly by information consumers.

Perhaps the most depressing scenario: If we all distrust AI, and regulators mirror that distrust, then is AI already a dead dog in terms of having any power to truly benefit us?

This discussion suggests that the early missteps and excesses of the people behind AI could well undermine its long-term potential to do anything worthwhile. It may end up being neutered and 'safer', but its potential will be wasted.

For me this is where the major owners and arbiters of AI have to show their working and what guardrails are in place. If they fight amongst themselves at the expense of public trust, their future dreams could well be over before they begin.

My tapdance over the peaks of the report do no justice to its finer details, and I do urge you to arm yourself with its discussions, if not its inevitable non-conclusions.

Go to the Open Society Foundations report homepage here, or download the PDF directly here. No sign-up needed.