[Week 34 of 2025] Prompting, Packaging, and Patriotism

Welcome back to Price and Prejudice with a few musings from Week 34 of 2025.

What Should AI Do?

This FT article reports that Bernstein handed typical research tasks to ChatGPT, Grok, Gemini and others. The models did well on transcripts—pulling out concerns, testing management’s answers, spitting back summaries. They struggled on the grander stuff: historical context, judgment, tone. My prior, though, is that this isn’t really an intelligence gap so much as a prompting one. It’s not obvious why a system that can juggle reams of data should suddenly falter at cross-cycle analogies; more likely, it just wasn’t asked the right way.

Where AI shines is in the work that is either easily verifiable or where precision isn’t the point. Easily verifiable: reconciling financials, reformatting spreadsheets, catching inconsistencies, generating alternate drafts. Low-stakes: summarizing an earnings call, sketching a translation, extracting themes from transcripts, flagging sentiment in headlines. In both cases, “pretty good” is sufficient—either because mistakes are obvious, or because the task itself is about compression rather than accuracy.

Which leaves the human analyst in an odd position. If the machine does the heavy lifting, what’s left is judgment, narrative, and—above all—liability. The grimmer version is that humans become less knowledge workers and more insurance policies: bodies attached to mistakes so that someone can absorb the blame. It’s a little like The Matrix, except instead of drawing power from our nervous systems, the system draws cover from our signatures. Another variant is more Kafka than sci-fi—humans as formalities, present mainly so the process still feels official. Either way, survival looks less like being indispensable, and more like being expendable in a useful way.

Packaging as a Leading Indicator

This Bloomberg piece makes the case that cardboard boxes are one of the better leading indicators of the US economy. If you think about it, nearly everything gets shipped in a box—so if fewer boxes are moving, that usually means fewer fridges, TVs, and groceries are moving too. Box demand isn’t ordered years in advance, so when companies hesitate to buy packaging, it’s a real-time signal that demand is softening.

The slump now seems tied to tariff uncertainty: retailers aren’t stockpiling packaging until they know what imports will cost. But the bigger idea is that leading indicators work precisely because they’re mundane. Nobody buys boxes for fun; they buy them because they have stuff to sell. Watching something as boring as packaging can sometimes tell you more than watching retail sales themselves.

The same logic shows up at a bigger scale. Apollo notes that FedEx’s stock price tends to lead global trade volumes by a few months. It makes sense: if FedEx is struggling today, world trade usually looks weaker tomorrow. Planes full of packages are basically boxes with wings, and FedEx’s earnings embody whether people are moving goods around the world. In this sense, packaging is the purest economic signal: fewer boxes today, fewer sales tomorrow. That’s also why investors keep circling back to these unglamorous measures. Housing starts, trucking volumes, shipping containers, FedEx’s stock—they’re all cousins of the same idea: watch how stuff moves. Packaging just happens to sit at the junction between factories, retailers, and households, which makes it surprisingly powerful.

Return of the (SPAC) King

Chamath Palihapitiya (who had already taken ten SPACs out for a spin, with outcomes ranging from Virgin Galactic to Clover Health) is back with a new blank-check company, this time branded American Exceptionalism Acquisition Corp. A. The focus is energy, AI, DeFi, and defense. He’s looking to raise $250 million, and the sponsor shares only vest if the stock rises 50%, which is at least an improvement on the pandemic-era giveaways. This is essentially a bet on the political moment: Washington is more receptive to Silicon Valley defense start-ups, and VCs are pouring into the sector. A SPAC wrapped in patriotic branding is a way of saying the next one will be different.

It probably won’t be, of course. But the persistence of the format tells you something: SPACs survive not because they worked brilliantly, but because they remain useful to certain kinds of companies and certain kinds of promoters. The rest of us just get to watch the same experiment rerun, only with slightly new marketing.