@lawrencegis.bsky.social
In that case, what we can do is pass real-time, accurate information to LLMs via tool use invocation. I'll test that approach another day.
I'm Lawrence, and I talk about #geospatial and #AI. Follow me to learn more.
(7/7)
However, there are billions of Points of Interest on our maps. Labeling them all in training datasets would be incredibly laborious, not to mention that these POIs change over time and require continuous updates.
(6/7)
If you break down how LLMs work, you'll realize they function so well with text generation because they're word-by-word generation engines pre-trained on enormous amounts of well-labeled datasets that are either machine-generated or manually verified before training.
(5/7)
AI-powered geospatial mapping interface testing LLM location accuracy on Golden Gate Bridge with blue location pin marker, satellite view, and AI assistant chat showing coordinates longitude -122.4783 latitude 37.8199
Satellite aerial view of Singapore urban area showing Takashimaya Shopping Centre location test with AI mapping tool, featuring blue location pin marker and red highlighted building outline demonstrating LLM location intelligence inaccuracy
Geospatial AI testing interface displaying Singapore LaunchPad location at Ayer Rajah Crescent with location pin marker and red polygon overlay, demonstrating LLM spatial accuracy limitations in mapping software
1. When I asked for prominent landmarks like the Brooklyn Bridge, it accurately pinpointed the latlong on the map.
2. However, when I queried less well-known locations, it gave locations that were off by hundreds of meters from their actual positions.
(4/7)
But LLMs' ability to understand such questions relies on a fundamental assumption: Do LLMs actually have all locations accurately memorized?
To test this, I built a simple demo with OpenAI's GPT-4.1.
Here's what I found:
(3/7)
After ChatGPT went viral, we learned that LLMs are text-in, text-out models using transformer architecture.
Naturally, we ask questions like "How many Starbucks are there near Union Street?" or "Which housing estate in Singapore has the highest growth potential for the next 10 years?"
(2/7)
A combination of 3 screenshots of a geospatial model I made using ChatGPT, demonstrating inaccuracies with lesser-known places.
How far are LLMs from becoming location intelligence that actually works?
Geospatial and LLM integration is a highly hyped combination among devs right now, including myself. I'm grateful to be part of this generation witnessing these changes unfold before our eyes.
(1/7)
@owenboswarva.bsky.social hey thanks! Love your work. Im gonna make a thread about LLMs in geospatial later. Any first level thoughts?
10.06.2025 09:38 β π 0 π 0 π¬ 0 π 0@tech-timc.bsky.social hey thanks! Love your work. Im gonna make a thread about LLMs in geospatial later. Any first level thoughts?
10.06.2025 09:38 β π 0 π 0 π¬ 0 π 0what are your hot takes for AI in geospatial? Gonna post a thread soon
10.06.2025 09:36 β π 0 π 0 π¬ 0 π 0
See the first episode of my "Build Geospatial with Me "series!! Showing how I built a complete GIS vector processing pipeline capable of rendering multi-million vertex datasets directly in the browser. Out now on YouTube: youtu.be/LPxVgj6I9Nw
#Geospatial #NikaPlanet
Need to be more active here. Still not in the habit of checking bsky every day
27.05.2025 05:37 β π 0 π 0 π¬ 0 π 0Posting a video soon about the thinking process behind building NikaPlanet + the daily life of a startup CTO (spoilers: it's coding on a beach... the view is nice!)
Want to post it on X too but its not allowing me to change username, citing "your rate is limited" forever ??? anyone got a clue?
My first post in Bluesky! Hope to contribute for all #geospatial #remotesensing #gis contents
16.05.2025 08:16 β π 1 π 0 π¬ 1 π 0
