There is no shortage of people who are ready to tell you how AI just changed everything.

They'll sell you something, too, if you're not careful, but I imagine you are seeing and hearing from the same people I am seeing and hearing from, and they are LOUD.

There's a denominator under these claims that's been making my earlobes itch. So let's say it changed everything in how we capture and use data. I'll concede that point without hesitation.

But isn't the data all still more or less the same?

The inputs - outside of some updated collection and querying measures - it's still all about the inputs, right?

Garbage in, garbage out vs. good data, sorted well gets good outcomes.

Eric Pachman came back on Excess Returns and we got deep into the BLS data and the new visualizations they're coming up with over at Data4thePeople. It had my head swirling around this idea all the more.

The BLS (Bureau of Labor Statistics, ahem, sorry for the alphabet soup) has been publishing labor force participation numbers, occupational wage surveys, quarterly census employment data for years. They share it for free. It's a public utility. If you have a Bloomberg you are way familiar with it. If you have internet access, you can find your ways to play with it.

The point is it wasn't hidden. It was right there the whole time. The raw inputs. Just on the internet. It just was clunky to deal with.

Eric's got a background as a chemical engineer with loads of logistics experience. There's an Intentional Investor and a Just Press Record (or two) if you want to get further into his background. The common theme for his entire career though, is that if you notice something in the data, he’s got ways to help you make something with it. That's the ethos, and it's contagious.

The point is that Eric is default curious in making sure the data he's looking at is useful for the way it's being discussed.

That matters for investors, journalists, community advocates, whoever - basically, anybody who wants to actually see what the numbers are saying.

So you start talking to Eric and he starts explaining what's in the almost impossible to download and read spreadsheet the BLS puts out. His solution is AI. The problem is the presentation of the data. It’s wild to hear it said so bluntly.

He starts telling you about how the labor force participation rate for men 45 and over is at its lowest point on record. How 32 states would have had negative job growth in 2025 were it not for the Medicaid care economy. How home health aides - and there are 4.3 million of them, and growing, mostly made up of women and immigrants, many of which are below the poverty line - have been the source of American job growth in the country for the past year and a half.

So when people say "Jobs are fine, look at employment" Eric can give you the data to answer, "Oh yeah? Look!"

The data was always the data.

But AI is changing how we look at it.

This is a pivotal moment in storytelling. It's not about editorializing or manipulating narratives (although, it could be, you must be good - please), but is about how the tools anyone can use are so much more powerful now.

As Eric puts it - "the crank is cheaper now."

And that's true. What used to take two weeks takes two hours. The visualization that would've required a team now requires one curious person and a high schooler who noticed something worth building.*

But cheaper crank-turning just means the story needs you more, not less. This is the issue we’ve been pounding the drum on for the past year or so at Panoptica (and even more so at Perscient!). Eric is clear about this too: AI will overclaim, underclaim, and occasionally drop one sentence that torpedoes the credibility of everything around it. Catching that - knowing this is the wrong sentence in exactly the wrong way - that's not in the model. That's in the person who has enough at stake to care whether the piece holds up.

The data was always there. The crank is cheaper now. What that frees up is the part that was always the point: the story, the stakes, and the taste to know when something's off.

*He's also running a data storytelling fellows program - and one of his fellows is a rising high school senior named Jonathan Pickens, who built an interactive wage visualization of 830 occupations out of the same BLS spreadsheet that crashes Excel if you look at it wrong. So Eric’s a teacher, too, and these kids are doing crazy, AI-assisted work with him!