Is Your Association Ready For AI?

Ever since ChatGPT exploded onto the scene a couple years ago, setting a new standard for how people are able to interact with computers, I've talked to a lot of people about potential projects and iddeas for AI. I've heard everything - from organizations needing Betty to unlock their content all the way up to all-knowing analysts that can consume everything and do anything with minimal effort. Some of these are more doable than others, currently, but the enthusiasm to take advantage of this rapidly advancing technology has been absolutely amazing to see.

But even for the ones that AI can handle right now, what does it actually need to be effective? What does your association need to be thinking about and preparing to make sure you're able to take advantage of what exists today and what will be out there tomorrow?

Just a heads up!

Before we get too far here, this post is going to be focused on technology preparedness rather than policy and culture - we would need a book to cover the rest!

Intelligence + Information

With the way generative AI has evolved and continues to develop so quickly, the biggest gap for most tasks is the question of whether the AI is going to have enough information and the right information when it needs it. Everything from preventing hallucinations to tackling extremely complex tasks seems to be solved by getting the right information at the right time.

That could be like with Betty where we need to train on journals, news articles, educational content, and whatever else your knowledge assistant needs access to for her to help members with all their questions. Or it might be all your AMS data and purchase history for data analytics tasks. Or it could be the wider internet if you want AI to be able to go out and research ideas and interact with other systems.

The models are smart and cheap and always getting smarter and cheaper, but the value for your organization is based on what your organization knows that *it doesn't*. So, access to the right information to provide the right answers and take the right actions is key.

So, for your organization...

That means the question about preparedness is primarily this: if you want to use AI for tasks, do you have a way to give it the information it needs? Can you do so in a secure and repeatable way?

When working on training, we receive content in a variety of ways. Some methods are easy to automate and very secure - most often, this would be an API connection directly to the source of content. Some methods are easy but require manual intervention - like uploading specific files and having to rename them or provide useful citation links.

There are a ton of potential systems in play when talking about feeding information to an AI and knowing how those systems and the vendors involved can share content is key in knowing whether that information is easily usable for AI or if it will require additional work to figure out.

Well, are you prepared?

Ultimately, most systems have a way to get access to content and, with how fast AI is moving, getting involved whether you have the ideal setup is the only way to have a shot at keeping up. Whether your setup is ideal or not, the best way to do that is to start experimenting. The tech will only continue to evolve faster and faster, so the most important thing is making sure your learning curve does the same.