Problem Identification and AI
So far, AI has been a solution looking for a problem to solve. While there are some amazing breakthroughs because of the models we now have, mostly it’s a monetization scheme: Do this cool new thing and give us lots of money. In the meantime, we’ll destroy the environment, accelerate the pace at which humans get dumber, and monopolize the conversation and job market.
I don’t write this because I’m anti-AI. I’m not. I am an AI skeptic, and I’d love to be proven wrong. I write this in hopes of provoking a better conversation around a potentially existential threat to our society.
In his May 31, 2026, New York Times essay, “There’s Something Else We Should Be Worrying About,” Ezra Klein posited that there should be a public aspect to AI. Some of the computing power and tokens should bought by the government and spread across universities, governments, libraries, etc.—entities that serve the public good but don’t have endless budgets.
Two other ideas from the article bring me here:
Identifying a problem before using AI to solve it
Needing data to work with
Problem identification & data sources
Instead of handing out AI tokens and randomly having people use AI for whatever, let’s start with a problem. (#2 on my list of responsible use of AI.) From the essay:
“AI’s benefits will not emerge automatically or inevitably. It will take work to identify the problems AI can help the public solve and then create the data, financing, and ‘compute’ … needed to actually solve them.”
This gets to what I’ve been pondering the last few months: What is AI good for? What are its benefits?
I’ve come up with a pretty short list for my professional interests (content strategy, information architecture, and user experience). Most of the benefits are slim margins that could be gained by doing the foundational work I’ve been espousing for more than a decade.
Which leads directly into the second point: For AI to do its work, it relies on knowledge represented as information and data in a place it can be accessed. Klein notes:
“It takes work to structure a problem in a way that allows AI to be useful…The AlphaFold advance was possible only because of the Protein Data Bank, a laboriously created database of protein structures that the NSF began funding in the 1970s. No Protein Data Bank, no AlphaFold.”
The breakthroughs AI has enabled didn’t come because someone entered a prompt into a chatbot in 2026. They happened because someone or a team or people put their heads together and said, “What if we could use all this data to do this specific thing?”
And they could do that because the work of collecting the data was already done! Someone had the foresight long ago to collect data knowing it would be useful because of their deep knowledge of a field – even if they did not necessarily know how it would be useful in the future.
We need human-created information and data to power the AI tools that will help us solve real problems.
The business part
The example Klein cites relates to the public good, which is important and admirable. But it’s not the only way it’s applicable.
Many people in my network talk about being excited about AI. Most of them don’t say why. Most of them are selling products or services that will make them money if they can hype the technology enough to get executives to buy the product or service they are selling.
When I think about AI in my field—particularly content strategy and management—I see one use case in particular that would move our field and organizations forward. And it’s not sexy. Governance.
Governance is how things get done in an organization. When it comes to content, it is the most neglected aspect of operations and keeps AI from being optimized. Getting the governance right will allow organizations to keep their content shiny and bright while allowing AI to automate the governance and optimize the use of the content itself. Not creating content faster or increasing the output, but to clean up the content, make sure you’re only creating content people need and ensure it stays timely and relevant.
This requires someone/a bunch of people to make decisions about policies and standards that have never been set—or followed or updated if they have been previously set. Then they have to be put in place and tested to get the bugs out. With clear governance rules and a process to follow, automation can happen. This was possible with the right technology before AI, but now AI can accelerate the implementation and do more evaluation than previously. Yes, we still need a human in the loop to discern and make decisions about what the robot tells them should be done.
Short of this, I’m not sure there are real novel use cases for AI that aren’t just throwing money at a solution in search of a problem. What am I missing?
I don’t want more AI slop built on SEO slop. I want content for humans by humans that serves a purpose. Computers—sometimes generative AI—can help us to find the information we need faster than ever. This is the real advance since the dawn of the World Wide Web, but what good is it if we only find content created to be found, not to solve a problem? Let’s get back to basics and see what we can build!