Godview AI

Godview is one of the most promising new AI-powered maps to have emerged in the past 18 months. It is an interactive map that allows users to perform geographical searches using natural language queries.

This week, Godview introduced an exciting new feature called "Discover." This addition enhances the user experience by enabling individuals to click on any location on the map and instantly access AI-generated insights about the area.

Discover

The Discover feature is particularly valuable for a wide range of users, from travelers and researchers to everyday individuals curious about their surroundings. For example, as I plan a trip to southern Italy, Discover should allow me to simply click on towns and cities in Puglia to identify which ones might be most interesting to visit. It could also help me pinpoint must-see attractions in the towns where I eventually decide to stay.

Or Don’t Discover

Unfortunately, Godview appears to be somewhat myopic, and its Discover answers can be geographically imprecise. For example, when I clicked on West Ham in London, I was informed that it was Wanstead—"a generally affluent area." While the two locations are only two miles apart, in a city like London, those two miles can make a world of difference (West Ham actually has one of the highest poverty rates in the UK). 

I encountered similar inaccuracies when clicking on towns and cities in southern Italy. For the large city of Bari, Discover provided an accurate and useful general description. However, for smaller towns in Puglia, the results were more inconsistent. Godview frequently defaulted to a broader description of the Puglia region rather than providing details specific to the town in question. In some cases, it even confused locations with nearby towns.

Final Thoughts

While Godview’s Discover feature has significant potential, it still has room for improvement in accuracy and precision. The issue may stem from imprecise reverse geocoding rather than the AI itself - perhaps the geocoding service Godview relies on is returning overly broad results. Alternatively, the language model (LLM) powering Discover may currently be limited to generating insights in some locations at a broad regional rather than a local level.

This also raises the question of user needs - something Discover does not yet fully address. Different users clicking on Greenwich Village in New York, for example, may be looking for entirely different levels of detail. One might expect a general overview of New York City, another may want specifics on the neighborhood itself, while a third could be interested in insights about a particular block or street.