What Are We Actually Making More Time For?

We have reached a point where technology can produce more information in seconds than a person could reasonably process in hours, and somehow, we have decided that the most impressive part is how quickly it can write. I find that interesting because producing information and understanding information are not necessarily the same thing. The more I think about artificial intelligence entering journalism, the more I wonder whether we're beginning to confuse the two. We keep asking whether AI can do a journalist's job, but shouldn't we first be asking what a journalist's job actually is? If journalism is simply taking information and turning it into readable paragraphs, then AI is becoming pretty good at that. But if journalism involves recognizing which information matters, questioning what we're being told, investigating what isn't immediately obvious, and helping people understand the world beyond the facts placed in front of them, that's an entirely different conversation.

In Journalism on Autopilot, Haisten Willis describes sports reporter Curt Conrad, who uses automated journalism to generate basic game coverage. Instead of spending hours organizing scores, statistics, and game summaries, Conrad has more time for interviews, photography, and in-depth reporting. I actually love this use of technology, not because a computer can write a sports recap, but because of what that allows a human being to spend time doing instead. We talk about AI in terms of how much time it saves, almost as if saving time automatically means we're making progress. But what happens to that time afterward? If a newsroom uses AI to produce 800 stories instead of 80, that looks like an incredible improvement when we're measuring productivity. But what if that same newsroom used the technology to produce its routine coverage while giving reporters more time to investigate issues that otherwise wouldn't receive attention? Which outcome actually makes the public better informed? I don't think we can answer that by counting articles.

Willis also explains how the Associated Press dramatically increased its production of quarterly earnings reports through automation. I see genuine value in that. There are plenty of situations where organizing verified, structured information doesn't require a journalist to manually write every sentence. AI can also help reporters identify unusual patterns, monitor developing events, and process information that would otherwise take hours to examine. But there's something about this that I think we overlook. We could have an endless supply of articles explaining what happened according to the data we already possess without actually discovering anything new. Increasing the amount of information available doesn't necessarily increase the amount of knowledge we have about the world. And for an industry whose purpose is to inform the public, I think that distinction deserves more attention than the number of stories a computer can generate.

One example that stood out to me in Rob Tornoe's Newsrooms Should Carefully Consider AI Innovation and Accuracy involved a chatbot that invented an accusation of plagiarism against a real person. It didn't just make up the accusation. It created the appearance of an article and citation to support it, even though neither existed. What interested me about that example wasn't simply that the technology produced false information. It was how easily it could create the appearance of evidence. Think about how we decide whether something is believable. Someone makes a claim, references an article, provides a date or publication, and we assume there's a process behind those details. We assume someone found the information, checked it, and established that the evidence supported the conclusion. But generative AI can imitate what verified information looks like without actually verifying anything. It can reproduce the final appearance of research without the research ever happening.

To me, that creates a problem much bigger than a computer occasionally getting something wrong. We've developed technology that can provide an explanation before we've established whether the explanation is true. Usually, we investigate something because we don't have an answer. We gather information, recognize patterns, question inconsistencies, and eventually develop an explanation. But what happens when the explanation comes first? Are we still as motivated to investigate, or do we accept something that sounds reasonable and move on because the uncertainty has already been resolved for us? I think that's especially important in journalism, where a believable explanation can influence how people understand real events and real people. The danger isn't only that AI might give us incorrect answers. It's that we might become increasingly comfortable accepting answers without understanding how they were established.

Willis describes another example involving the Los Angeles Times and its automated earthquake reporting system, Quakebot. In 2017, the system published a report about a magnitude 6.8 earthquake in California. The earthquake really happened, but it happened in 1925. I actually find that example more interesting than if the system had completely invented an earthquake. The underlying event was real. The problem was that the information was presented in the wrong context. That makes me think about how easily we confuse factual accuracy with understanding. Imagine reading that a city increased its spending by 30 percent. That number could indicate irresponsible financial management, an emergency response, or an overdue investment in something the community needs. The statistic itself cannot explain which situation we're looking at. Somebody has to investigate the circumstances surrounding it. Journalism can't simply be the accurate arrangement of information because even accurate information can create a misleading picture when the context is missing.

Another point Tornoe raises is that AI systems can be less reliable when asked about local political figures because there is less existing information available about them. That immediately made me think about smaller communities. Consider a school board making a decision that affects local families, a city changing its housing policies, or an issue residents have been discussing for months that nobody has formally investigated. Those stories could directly affect people's lives, yet there may be very little existing information for an AI system to work with. AI can summarize what has already been documented, organize public records, and help reporters identify patterns, but it cannot independently establish what happened during a conversation nobody recorded. It cannot replace the process of asking a public official a question that nobody has thought to ask yet. Someone has to go find out.

This is where I see a possibility that's much more concerning than the usual argument about robots replacing journalists. What if we eventually become incredibly efficient at distributing information about the world while becoming less capable of discovering what's actually happening in it? We could have thousands of articles, summaries, notifications, and updates appearing constantly. Everything could look incredibly informative. But if fewer people are conducting original reporting, where is all that information coming from? At some point, we could just be creating more and more content from an increasingly limited foundation of firsthand knowledge. That's not necessarily a more informed society. It might just be a society with a lot more to read.

I don't think the solution is to reject AI. That would mean ignoring genuinely valuable technology because it introduces risks. Tornoe discusses how journalists can use AI-assisted tools to extract information from scanned documents, brainstorm story angles, and improve editing. Willis describes systems that can recognize unusual patterns in data or alert reporters to potential breaking news. I think these applications could make journalism significantly better. Imagine an investigative reporter being able to search through thousands of pages of public records in a fraction of the time it would normally take. That's not eliminating journalism. That's giving someone more capacity to do it. But I think we need to distinguish between technology that helps journalists discover questions they might not have thought to ask and technology that encourages them to believe the questions have already been answered. A system that identifies an unusual pattern in financial records could be the beginning of an important investigation. A system that confidently explains that pattern without enough evidence could be the end of one before it even starts. The technology might be similar, but the consequences are completely different.

That's also why I don't think the right balance between AI-generated content and human reporting can be measured in percentages. I wouldn't necessarily object to a newsroom using AI to generate a large portion of its routine coverage if the information was reliable, the system was appropriately checked, and the organization remained accountable for what it published. What matters more to me is who established the information in the first place. Who asked the questions? Who verified the claims? Who recognized that something didn't add up? And when the explanation turned out to be more complicated than expected, who continued investigating? AI can assist with those processes, but generating a plausible answer is not the same as completing them. I also think readers deserve to know when substantial portions of news content have been generated by AI. Willis raises the issue of transparency in automated reporting, and I agree that disclosure matters. However, a label doesn't make an inaccurate article harmless. If a newsroom publishes something, it should be responsible for what readers are being told regardless of whether the sentences were written by a person or a computer.

There's another part of this discussion that I don't think gets enough attention. News organizations don't just tell us what's happening. Through their decisions about coverage, they also influence which events and issues receive public attention. If AI makes certain kinds of stories significantly easier and cheaper to produce, wouldn't those stories be more likely to appear? Not necessarily because they're more important, but because they're easier. Sports scores, earnings reports, election results, and other structured information fit automated reporting particularly well. But complicated stories involving conflicting experiences, institutional decisions, or people whose perspectives haven't been documented require a different kind of work. I wonder whether increased automation could eventually shape not just how news is produced, but what the public gets to know about. A newsroom might produce more content than ever before while gradually dedicating less attention to stories that are difficult to turn into predictable formats. We'd have created technology intended to make journalism more efficient, only to potentially narrow the boundaries of what the public understands.

Willis includes different perspectives on whether automation will give journalists more freedom to pursue meaningful stories or eventually reduce the number of people employed to report them. I think both outcomes are possible, and that's exactly why the technology itself cannot determine whether this is progress. If news organizations use AI to create more opportunities for investigations, community reporting, and original work, that's an improvement. If they use it to justify producing more content with fewer people asking questions, I don't see how the public necessarily benefits. The difference isn't really about how advanced the technology becomes. It's about what news organizations choose to value.

I keep coming back to this assumption we have about technology: that if something makes a process faster, it must be making the process better. But why do we assume that? Sometimes the time it takes to do something is part of what makes the result valuable. A journalist spending weeks trying to understand conflicting information isn't necessarily being inefficient. They might be doing exactly what responsible journalism requires. Not knowing the answer immediately isn't always a problem that needs to be solved. Sometimes uncertainty is the reason we investigate something in the first place. And that's what concerns me about becoming so accustomed to technology that produces immediate, confident explanations. We might start treating the time required to question something as an inconvenience rather than recognizing that questioning is how we establish whether something is true.

I don't want journalism to become a race to generate the most believable explanation of reality. I want it to help people understand reality, especially when reality turns out to be more complicated than the first explanation they were given. So yes, I believe AI belongs in journalism. It can make reporters more capable, expand access to information, and eliminate tasks that consume time without necessarily requiring human interpretation. But I don't think the future of journalism should be decided by how much content a newsroom can produce. It should be decided by how much more the public can actually understand because of the work being done.

Maybe that's the question we should have been asking about AI all along. If technology can give journalists back so much time, what are we actually making more time for? Because if the answer is simply producing more information, I'm not convinced we've accomplished much. But if the answer is asking better questions, investigating what other people overlook, challenging explanations that seem too easy, and giving journalists the freedom to pursue stories that haven't already been told, then I think AI could be one of the best things to happen to journalism. I just don't want us to become so impressed by technology's ability to give us answers that we forget why we started asking questions in the first place.