How the news ecosystem might look like in the age of generative AI

Public uptake will be one of the most important driving forces shaping the AI-mediated information ecosystem, and, by extension, journalism and news media’s place in it.

It’s also a force that is sometimes overlooked in discussions that tend to focus on start-ups, incumbent technology companies, and others on the supply side.

But recent examples including blockchain/Web3, the “Metaverse”, voice operated smart speakers, and augmented/virtual reality all illustrate that incumbent tech companies and investors backing new entrants can invest billions in new technologies marketed as transformative. But for the supply to be transformative, it needs to meet an effective demand.

AI pragmatism on the public demand side

On the public demand side – business-to-business (B2B) and business-to-government (B2G) are separate issues – years of research on how people engage with other digital technologies suggest the most likely overall approach from the public at large might be what we could call “AI pragmatism,” defined by a combination of three things:

• abstract concern about the possible societal impact of AI
• generalised scepticism towards many of the institutions using AI
• a practical appreciation of many of the tools that in part rely on it, as long as the benefits for them personally, and perhaps society, are clear to them.

This is, broadly speaking, how much of the public has approached the commercial internet, search engines, and social media. Just as journalists, civil society, and researchers are documenting many problems with generative AI, they have over the years documented many problems with older forms of digital media. 

Most people don’t love any of these things, and aren’t blind to the imperfections, or to the self-interest of those behind the offers, but they demonstrably use them anyway – arguably because they, on balance, find them useful and worthwhile (unlike VR, etc.).

For reality-based journalists and news media thinking about what generative AI might mean for them, it is really important to recognise two key things.

First, while much of the public are probably going to be quite sceptical of the use of generative AI for many kinds of news, they are, if past research on people’s perception of editors versus algorithms as ways of getting news, also quite sceptical of many current offers. The fact that people see a lot of crap on the internet does not automatically make them value your content more.

Second, much of the public are routinely using, at vast scale, digital offers that they don’t necessarily particularly trust, and countless advertisers continue to invest in the same offers despite whatever misgivings they have. The fact that people are concerned about the downsides of something does not necessarily mean they won’t use it if the upsides are clear and tangible.

These two things mean that even if the “trust gap” between news in general and news found on platforms recurs for news in general and generative AI, it isn’t necessarily going to provide journalism and news media with much of a bulwark against new competitive challenges. Some individual publishers do and will stand out, but not everyone and everything.

Public uptake will in part be driven by whether people find discrete generative AI offers (ChatGPT, etc.) and new bespoke generative AI platforms (e.g. GPT Store) useful, but probably most importantly by integration of generative AI into already widely used products – cloud computing offers (e.g. Azure OpenAI service), productivity software (e.g. Microsoft Copilot), search engines (e.g. Google’s Search Generative Experiences, Microsoft offering its Copilot through Bing), browsers (e.g. Google’s Chrome offering AI-generated summaries), social media (e.g. Snap’s My AI), and perhaps hardware (e.g. suggestion that Apple may integrate Google’s Gemini into future generations of iPhones).

Excepting cases where the companies involved in the latter want to conspicuously flaunt the term, generative AI may well become normalised to the point of being naturalised or even well-nigh invisible, no more noticeable or noteworthy to most of us than current use of e.g. machine-learned ranking in search engines or dynamic neural networks components in social recommendations are.

AI experimentalism and AI incrementalism on the supply side

On the supply side, leaving aside advertisers and all of us in our capacity as audiences, key actors in the information ecosystem can be stylized in a simplified form as:

• platforms who provide tools and technologies
• promoters who, in whole or in part, communicate for a living (whether individual “influencers”, political actors, or organisational PR and marketing)
• public self-expression by citizens, intrinsically-motivated communication-cum-“user-generated content”, sometimes individual, sometimes more organised)
• and publishers, content provision, including news, generally for profit in addition to the various principled motivations also involved. (The heterogeneous population of developers offering a wide range of different kinds of digital content and services are not strictly speaking publishers, but they are effectively often competing with these for attention, advertising, and users’ money.)

There are going to be ups and downs, a lot of variety, and, as always, a range of scenarios are possible. But because generative AI presents an opportunity for them, it is likely that a critical mass of platforms and promoters, and many forms of public self-expression, will embrace it. 

There will be numerous examples of disappointing results and outright failures, and many things will fade away. At the same time, disruptive innovations are likely to arise from “AI experimentalism” from incumbent platforms who feel threatened, new entrants, and surprising forms of emergent properties and user-driven innovation. Usage, as the line goes, is oxygen for ideas, and there will be a lot of usage.

Meanwhile, because generative AI presents a threat to their inherited business models, a critical mass of publishers are likely to take the same defensive approach they have for decades generally taken to digital technologies that challenge their existing business. Given the pressures and incentives they face, it is not surprising that many are primarily trying to harness AI for incremental purposes, essentially trying to use it to cut costs and realise more value from current material through greater use of AI for personalisation and re-versioning of content. Let’s call this “AI incrementalism.”

The likely implications are two-fold.

First, many publishers will produce more content more cheaply in a world where the bulk of news as we know it is already, from the point of view of much of the public, demonstrably largely commoditised, generic, and highly substitutable, and therefore of little value in terms of willingness to pay attention, let alone pay. If publishers primarily use AI to produce more of the same more cheaply, they will further reduce the already limited commercial value of all but the most effectively differentiated news content.

Second, others’ use of AI will mean that publishers continue to lose ground when it comes to most of what industry consultants call “user needs” and academics call “uses and gratifications”. With the partial exception of keeping people up to date on current affairs, publishers play a less and less prominent role in virtually every other use case, ranging from entertainment to education to engagement, all of which are served by a multitude of other competitors, some of whom will aggressively experiment with using generative AI.

What might come next, and what it means for journalists and publishers

Short term, it seems unlikely that generative AI will have anything like the transformative impact the current hype suggests. It looks a lot like a bubble, and bubbles eventually burst.

This will superficially seem to reward AI incrementalism.

But the burst will leave something behind – as the dot.com bust did – and longer-term investments will continue, technologies will improve, and user-uptake will increase, generating new ideas and insights, so the impact will be profound.

When that impact arrives, publishers who are AI incrementalists may find they have grown more efficient at delivering something that audiences and advertisers increasingly do not value.

If the public at large will be AI pragmatists, if AI incrementalism is short-term reasonable but long-term risky, and if AI experimentalism in its most flamboyant forms is beyond the resources of most individual publishers, how might journalists and news media focused on the longer term move forward?

Aside from the question of how to cover AI, I think they need to face three key challenges:

• First, to identify what, from members of the public’s point of view, sets them apart from an already abundant supply of content and information now being supercharged by growing use of generative AI by many other actors.
• Second, to consider whether generative AI is yet another chance for publishers to lean into an increasingly personalised, participatory, and personable information ecosystem that news media continue to mostly serve in one-size-fits-all, one-way, and often relatively impersonal ways.
• Third, to resist the temptation to use AI primarily as a “faster horse” for doing more of the same at a lower cost in yet another series of attempts to cut their way to success. There is limited public demand for much of the news that is currently being pumped out. It seems unlikely that there will be greater demand for more of the same pumped out more cheaply in an even more competitive marketplace for attention.

None of these challenges are new. AI just exacerbates their existential nature.

It’s important to be clear – the decline of most publishers’ commercial revenues and societal relevance will only be further accelerated by generative AI. Despite the very real opportunities, journalists have plenty of reasons to be worried.

The rise of the internet means publishers have largely lost control of distribution. The rise of platforms means they have largely lost control of discovery. The rise of so-called Web 2.0 began to challenge their control of content, and generative AI will supercharge that challenge. 

Legacy business models are already in managed decline, and using AI as yet another way to cut costs will not revert that decline. Many publishers, and thousands of hard-working journalists, are likely to be caught up in yet another wrenching revolution of the unfinished digital revolution.

But AI may be helpful for those publishers who are able and willing to define and double down on what makes them different, who are genuinely interested in meeting people where they are, and who can resist the temptation to further commodify the journalism they offer. 

Most of us want to understand the world beyond personal experience. Original reporting is often a necessary part of enabling us to do so, as is the willingness to take responsibility for exercising editorial judgement, assessing competing claims and conflicting pieces of evidence in often fast-evolving, contentious and opaque situations. AI, like other digital technologies, can help journalists do this better, and can help publishers give people a better experience when they engage with journalism.

I think that’s where the long-term opportunity lies – at the intersection between the timeless journalistic aspiration to seek truth and report it, the constantly evolving set of tools and technologies that can help journalists do that and can help people engage with journalism, and the enduring public desire to make sense of the world and what happens in it.

This is a longer version of the scenario I submitted to the Open Society AI in Journalism Futures 2024 convening in April, which challenged would-be participants to respond – in 300 words - to this prompt: “What might an AI-mediated information ecosystem look like 5 to 15 years from now? What are the key driving forces that get us there? How will media be produced and consumed? How will this alter markets? What new power structures might emerge? How might audiences change?”