Machines, not humans, will become the biggest consumers of news in 2018. This shift will be driven by the growing impact of smart devices and the internet of things in the information ecosystem.

When two machines speak to each other, they speak in a language of categorization — a taxonomy. Words are organized into categories, which then trigger an output. In machine-to-machine journalism, this output (a story or other piece of news content) from one machine is then interpreted by a second machine before it reaches its final audience.

For example, one machine can create stories out of data, send that story to a second machine, which then personalizes that story and only then disseminates it to a human. But machine-to-machine journalism can also achieve tangible outcomes beyond informing individual news consumers. For example, an automated financial story about stocks can change another machine’s investment patterns, or an AI lawyer can assess how libelous an automated article can be across different jurisdictions.

Journalists will need to examine how these machines speak to each other and how they develop connection with one another — perhaps even emotional ones.

But replicating human judgment in language is not an inevitable technological advance. Machine-to-machine journalism will be confronted with the challenge of replicating a kind of “journalistic intuition” merely through data points. This is particularly difficult, because humans have a very complex and adaptive way of assessing value judgments in one another’s speech. In other words, when two people speak, each has an intrinsic barometer that assesses the relative value of the words the other party is using against a standard that you perceive him or her to have. For example, when someone who tends to exaggerate describes something as “revolutionary,” the word has a different relative value than when somebody uses it who is rarely impressed. And as relationships develop, humans gain an increasingly nuanced understanding into one another’s speech value judgments. It remains a question whether a journalistic intuition can be developed using existing data points as an input.

The proliferation of smart devices in 2018 will quickly generate demand for “smart content” — news tailored to a specific individual or use case, and personalized with a specific editorial style. And the way that two or more smart machines can speak to one another through stories presents new opportunities and challenges — it can generate higher engagement but could also lead to information asymmetries.

Today, most people find information via search or social, and while these two channels are radically different in functionality, they have one thing in common —  any given article surfaced through these platforms is exactly the same for everyone in the world.

For example, an article entitled “Legionnaires’ disease outbreak looms over southern California” reads the same to me, a 31-year-old male living in New York City, and for a mother of two in her early 40’s residing in San Diego.

Content today is one-size-fits-all. And why wouldn’t it be? A journalist writes a story hoping to reach as many people as possible.

Search and social helped tailor information choices to individuals to a degree by leveraging content recommendation technology. But what happens when the content itself can be created, processed, and distributed through algorithms (the smart machines) that analyze readers’ locations, social media posts and other publicly available data? News consumers could then be served content through a smart device, tailored to their personality, mood, and socioeconomic status, among other things — infinite versions of the same story. True personalization drives higher consumption, but without editorial guidelines and well defined journalistic standards, could also create siloed views of the same reality.

When machine-to-machine journalism becomes a prevalent way to produce, process, and distribute content, then those algorithms will start falling under the purview of journalists. Just as it’s important to verify a source’s reliability, it will become crucial to consider the reliability of smart machines.