I live in Oaxaca, Mexico, which I think may be like the Nashville of Mexico. Big town, tourism center, good on culture and history – not a commercial powerhouse. If you want to talk about hotels, tourism, restaurants, Mezcal, then sure, it’s got some great stories. But there aren’t big libraries or universities in town. There are no networking events (at least now where people consider “analyst” a job). If you want information, you have to use the internet – or make it up. I think 99% of the world CHOOSES to use the internet as their one and only source of information, but for many of us, there is no option.
Right now I’m researching real estate price levels for different cities in Mexico for a client. There’s plenty of information out there – but it’s all marketing copy. No stats. No databases. I can get plenty of commentary, but no numbers. The only hard data I could find was from a private real estate services company, and it was from 2020. Not recent. I traced THEIR source, and it pointed me to a 404 Page Not Found message on another real estate company’s site. Not verifiable.
The good thing is that my other interest, besides global supply chain, is AI. I turned to a major search engine that used AI technology for searches. Sure enough, it gave me a beautifully succinct and authoritative answer: “the cost of industrial space in Mexico can vary widely. As low as $2.50 per sq. ft. for land and as low as $4.20 for existing facilities.” Great start. I still needed a lot more information, but I was on my way.
Then I checked the source of the AI’s information – and it was the same 3-year-old real estate company quote. The AI made it look like all the other data I accept without a moment’s thought. Not only was the data outdated and unverifiable but it was passed off as current, relevant, authoritative information. I’m concerned that there is no reliable AI data about Mexico or other emerging economies.
I had three takeaways from this exercise:
1. Mexico and developing markets will be misrepresented with arbitrary, irrelevant, and inaccurate data, disqualifying them for many investors and decision-makers. Mexico and other low-tech companies will go from being data-poor economies to being poor-data economies as the quality of common information about them gets distorted and fabricated by AI. Potential investors attracted by pre-COVID quotes are going to feel burned when they learn the real pricing and are likely to abandon Mexico (and the consultant who used the misleading information).
2. ChatGPT and other AI platforms are likely to supplant free-range searches due to convenience – and because more and more data is firewalled. Corporate websites, news sites, and NGOs used to shovel databases at you on their homepage, but now it’s all commentary and conclusions for free – the database is “members only” (if that). The problem is that AI is notoriously opaque about where they are sourcing their raw data, so it will be difficult to assess quality. That’s not a problem in busy nodes like New York and Shanghai, where there is a lot of analysis and data and anecdotal reports to build a complete picture. In less active places, one or two bad data points can distort the picture.
3. There’s an opportunity for all of you Luddite rebels. Every time Silicon Valley “solves” a problem, they create oligarchies and bubbles. Tech businesses tend to concentrate in a few interlocked hands until every platform is indistinguishable from the other, and everyone focuses on the same big markets. The news sites started it, then the financial sites did it, travel sites did it, dating sites did it, and AI-search will be the same. If you look off the beaten path, you may be able to discover some overlooked opportunities.
Final Word – This started out being about real estate pricing in Mexico, but it’s now more of a cautionary tale against blindly drinking the AI Koolaid. Managers and investors focusing on big cities will probably get access to good data, but if you are relying on AI to keep you in the loop in developing markets, you may be heading for trouble. Data is the lifeblood of AI platforms, but less developed economies and cultures that don’t have a habit of sharing online are not only going to be underrepresented in AI results, but they will also be misrepresented. Bad data will be passed off as vetted fact once the AI blender gets done with its mixing.