I’ve sat through more AI vendor pitches than I can count over the past two years. Enterprise platforms promising to “transform newsroom workflows.” Startups claiming to “democratize journalism through artificial intelligence.” Very slick presentations, very compelling demos.

And I keep asking the same question: what problem does this actually solve?

The answers are usually vague. “Efficiency.” “Scale.” “Future-proofing.” Words that sound important but don’t connect to any specific pain point journalists actually experience.

Here’s my increasingly unpopular opinion: most AI tools being marketed to media companies are solutions looking for problems. They’re built by technologists who’ve never worked in a newsroom, solving theoretical challenges that look impressive on a roadmap but irrelevant to actual journalism.

The Pitch vs. The Reality

Let me give you a specific example. I recently demoed a tool that uses AI to “automatically generate story ideas based on trending topics and audience interests.”

The demo was impressive. Dashboards showing trending keywords. AI-generated story pitches. Engagement predictions for each potential story.

But here’s the thing: no newsroom I’ve ever worked in struggles to generate story ideas. The problem is almost always the opposite—too many potential stories, not enough resources to cover them. Journalists don’t need AI to tell them what’s newsworthy. They need more reporters, more time, more editorial support.

This tool solves a non-problem while creating new ones. How much does it cost? How much time does it take to implement and maintain? What happens when the AI suggests chasing trending garbage instead of important stories? Who’s accountable when the “audience interest” optimization produces a newsfeed indistinguishable from any other?

The Problems AI Vendors Ignore

You know what actual problems newsrooms face? Here’s a partial list:

Understaffing. Most outlets are running skeleton crews, asking reporters to cover beats that used to be divided among three or four people. AI doesn’t fix this unless it can actually do the reporting—which it can’t.

Documentation and institutional knowledge loss. Every time a veteran journalist leaves, years of source relationships and beat expertise walk out the door. I’ve seen AI tools that claim to help here, but none that actually work.

The economics of original reporting. Investigation is expensive and often doesn’t pay for itself in traffic. Aggregation is cheap and performs well. AI makes aggregation even cheaper, which doesn’t solve the underlying incentive problem.

Reader trust erosion. People increasingly don’t trust news media. AI-generated content, however well-intentioned, doesn’t help with this. It might make it worse.

Platform dependency. Publishers’ fate is tied to decisions made by Google, Facebook, Apple, and now AI companies. This is a structural vulnerability that no tool addresses.

The vendors aren’t building tools for these problems because they’re hard and the solutions aren’t scalable SaaS products. It’s easier to sell “AI story generator” than to confront the actual crisis in journalism.

The Tools That Actually Help

I don’t want to be entirely negative. Some AI applications are genuinely useful in newsrooms.

Transcription is the obvious winner. Tools like Otter.ai or Trint save hours of tedious work. The accuracy has gotten good enough that most journalists trust them for interview transcription.

Translation tools help international coverage. Google Translate isn’t perfect, but it lets reporters work with sources in languages they don’t speak and access documents that would otherwise be inaccessible.

Structured data extraction can speed up investigations. When you’re dealing with thousands of documents, AI can help identify relevant records and pull out key information.

Notice what these have in common: they do specific, limited tasks that journalists genuinely don’t want to do, freeing time for actual journalism. They’re tools, not transformations. They make human journalists more effective rather than claiming to replace human judgment.

The Sales Tactics That Should Worry You

When an AI vendor can’t clearly articulate the problem they solve, they fall back on fear.

“Your competitors are adopting AI. You’ll be left behind.”

“The future of journalism is AI-native. You need to adapt or die.”

“Young journalists expect AI tools. You won’t be able to recruit without them.”

This is marketing, not analysis. The vendors have no idea whether your competitors’ AI initiatives are working. (They’re probably not.) “The future” is genuinely uncertain; nobody knows what AI-native journalism will look like. And in my experience, young journalists care more about working on meaningful stories than having fancy tools.

Fear-based selling targets executives who don’t have time to evaluate tools properly but worry about being the person who didn’t act. It’s effective marketing and terrible decision-making.

How to Evaluate AI Tools Honestly

If you’re in a position to evaluate AI tools for a newsroom, here’s my suggested framework:

Start with the problem. Can you articulate a specific workflow problem this tool addresses? “We need to transcribe interviews faster” is a real problem. “We need to be more AI-enabled” is not.

Ask for evidence. Not demos—demos are choreographed. Ask for case studies from similar newsrooms. Ask for data on actual outcomes. Ask to speak with current customers.

Calculate the full cost. License fees are just the beginning. Factor in implementation time, training time, ongoing maintenance, and the opportunity cost of not investing those resources elsewhere.

Consider what happens if it doesn’t work. How dependent will you become on this tool? Can you walk away if it underperforms? Vendor lock-in is a real risk.

Pilot before committing. Any vendor confident in their product should offer a meaningful pilot period. Reluctance to pilot is a red flag.

The Bet I’d Make

Here’s my prediction: most of the AI journalism tools being hyped today will be defunct or irrelevant within three years. The companies will run out of money, pivot to other industries, or get acquihired for their technology.

The tools that survive will be the boring ones. Transcription services. Research assistants. Data processing utilities. Not industry transformers—industry incrementalists.

Meanwhile, the fundamental challenges facing journalism—trust, economics, staffing, platform dependency—will still be there. They’ll require human solutions: better management, smarter business models, stronger reader relationships.

I’m not anti-AI. I use AI tools constantly in my own work. But I’ve grown deeply skeptical of AI as a category solution for journalism’s problems. The solutions are being built by people who don’t understand the problems. And newsrooms, desperate for any lifeline, are spending money they don’t have on tools that won’t help.

We deserve better. But we’ll have to demand it.