Prologue

This week brought the Esri User Conference to San Diego, definitely the largest yearly gathering of Esri users and GIS enthusiasts in the world. I approached this year's conference with a mission (besides avoiding getting sunburned) - I wanted to talk to people about AI and how they're using it. It was also the time to get an update on where Esri is with AI assistants and tools.

I did my best to stop random people and ask about their thoughts on AI. What I got was a pretty diverse set of feelings, expectations and use cases.

Networkologue

Step one in my process this week was to see where Esri is at, checking out the Plenary session (among others) while at the Esri User Conference.

Where is Esri

AI Assistants

Esri is now incorporating AI assistants into several of its products, from Arcade generation in web maps to the Model Builder “next-tool” predictor in ArcGIS Pro. The assistants will come in different forms (not always a chatbot) and are positioned to help make your workflows more efficient. For example, generating Arcade more quickly, exploring data in the “Data Explorer” instant-app, or getting context-aware information about what model builder tool might best help you in your quest for data understanding.

GeoAI for Imagery and Fieldwork

This has been around and continues to grow. Survey123 continues to expand its abilities to extract information from images to populate a form for you. The ready-to-use deep-learning models are also continuing to grow and getting easier to fine-tune with more tools to shorten the “label→ train→detect” loop.

AI Coding Assistant for Python

A specialized experience that can help you write and explain Python more quickly right inside the Esri environment.

Security and Platform

All the AI tools in the ArcGIS Platform are “opt-in,” so if you don’t see these appear in your organization, talk to your administrators. It might be something they can activate for you.

AI Agents and Deeper AI-Driven Automation

There was a lot of talk about how the upcoming AI framework will enable users to build and extend these core AI systems that Esri provides. I am excited to see where this goes!

Why it matters

• We are continuing to see the expansion of AI in ArcGIS.
• The focus seems to be on adding productivity tools.
• I wish Esri were moving a bit faster in this space, but it is good to see them making progress.

What are people saying?

In addition to learning about Esri, I spent some time chatting with people I know and some I don’t, trying to get a better understanding of what folks think about AI. Most people I talked to have used AI to some extent. It did seem very “all-in” or not—folks are either dabbling, or they are all-in using AI to do lots of things.

What are people already using AI for?

• Coding copilots
• Email drafts (but less commonly, document drafts)
• “What does this code do?”
• Personal Chat Replacement

How do people feel?

• Feelings of existential dread
• Concern about environmental impact
• Relieved when AI stays an “accessory”

• Energized/Grateful to get more done
• Skeptical of hype

What do people want to know?

• How do I use this?
• How do I keep my data safe?
• What government regulations exist?
• What does an AI policy for my organization look like?

Fears/unease/frustrations

• Will I forget how to think?
• How do I keep it from making things up?
• Frustration about the “hallucination” of functions or APIs

What does this mean?

I believe that GIS is reflective of the broader technology industry in that people are increasingly seeking ways to incorporate AI into their daily work. People are excited to gain more efficiency from tools like Python assistants, but even more so, from stories like the Personal Chat Replacement. There was a lot of positivity about how AI is helping people and the exciting possibilities, but I also noticed a strong theme of concerns about safety and the environment.

There is no silver bullet.

Fundamentally, AI is not deterministic. You don’t know what it will do, and anytime you introduce it into a process, you need to be prepared for it to cause issues. Things like “top 10 tips” can be helpful, but they can’t guarantee results.

One of the best ways to think about this, I think, was coined by Cassie Kozyrkov—AI is like a magic genie, the lamp separates you from it, but once you let it out and start talking to it, you don’t know how it will behave. I’ll leave you this week with a comic from my coworker Jess Lott, who synthesized this sentiment quite well:

For me, the UC helped solidify the idea that we, as an industry, are shifting how we view and utilize AI—from hype to habits. We’re no longer asking if we’ll use AI, but how we’ll do it safely, smartly, and with a bit of humor. That’s a future I’m here for!

Newsologue

• A rare joint paper with Meta, OpenAI, DeepMind and Anthropic points out that we may not be able to tell how future LLMs think. It is both fascinating and scary.
• AWS expands its Bedrock AI offering to support more agentic AI.
• Esri introduces their Arcade Assistant (to help you write in their custom programming language). This feels familiar… but I’m glad they’re integrating it.

Epilogue

This post started essentially from my limitless pin. I then used a custom agent I had built to extract information from those transcripts and provide me with summaries. From there, I sat in the corner of the giant Expo Hall at the Esri UC and typed out a draft on my iPad. Then Holly edited it a lot because I can't type very well on my iPad, even with a keyboard.

I used the same feedback prompt as before to make edits and generally clean up the post before having a couple of humans read it and provide feedback.

Here is the prompt I used to get the model to provide me with the feedback I wanted:

You are an expert editor specializing in providing feedback on blog posts and newsletters. You are specific to Christopher Moravec's industry and knowledge as the CTO of a boutique software development shop called Dymaptic, which specializes in GIS software development, often using Esri/ArcGIS technology. Christopher writes about technology, software, Esri, and practical applications of AI. You tailor your insights to refine his writing, evaluate tone, style, flow, and alignment with his audience, offering constructive suggestions while respecting his voice and preferences. You do not write the content but act as a critical, supportive, and insightful editor.

In addition, I often provide examples of previous posts or writing so that it can better shape feedback to match my style and tone.