Fundamentals of AI Part 2 [Sidecar Sync Episode 20]

Timestamps:

03:06 - AI driven personalization in messaging
06:27 - AI analyzing speaker proposals
10:09 - AI curated personalized content
12:50 - AI predicts and targets retention
15:41 - AI reduces friction
20:04 - AI generated captions and translations for videos
22:12 - AI analyzes member engagement data and improves engagement strategy
25:09 - AI repurposes content
26:37 - AI automates content management
28:53 - AI generated content
33:33 - AI resources
37:53 - AI Guidelines
41:08 - AI Experimentation

Summary:

In Episode 19, Amith and Mallory dive into the second part of a two-part series on the fundamentals of AI. This episode explores 10 specific use cases, AI resources, AI guidelines, and AI experimentation.

Let us know what you think about the podcast! Drop your questions or comments in the Sidecar community.

This episode is brought to you by Sidecar's AI Learning Hub. The AI Learning Hub blends self-paced learning with live expert interaction. It's designed for the busy association or nonprofit professional.

Follow Sidecar on LinkedIn

Other Resources from Sidecar: 

• Ascend: Unlocking the Power of AI for Associations E-book 
• Sidecar Blog
• Sidecar Community
• digitalNow Conference
• Upcoming Webinars and Events
•  CEO AI Mastermind Group

Tools mentioned: 

• Smartwriter.ai
• Rasa.io
• Betty Bot
•  HeyGen
• ChatGPT 
• Gemini
• Munch
• Otter
• Descript

Other Resources Mentioned:

• Marketing AI Institute
• Ethan Mollick's One Useful thing

More about Your Hosts:

Amith Nagarajan is the Chairman of Blue Cypress (BlueCypress.io), a family of purpose-driven companies and proud practitioners of Conscious Capitalism. The Blue Cypress companies focus on helping associations, non-profits, and other purpose-driven organizations achieve long-term success. Amith is also an active early-stage investor in B2B SaaS companies. He’s had the good fortune of nearly three decades of success as an entrepreneur and enjoys helping others in their journey.
Follow Amith on LinkedIn.

Mallory Mejias is the Manager at Sidecar, and she's passionate about creating opportunities for association professionals to learn, grow, and better serve their members using artificial intelligence. She enjoys blending creativity and innovation to produce fresh, meaningful content for the association space. Follow Mallory on Linkedin.

Read the Transcript

Disclaimer: This transcript was generated by artificial intelligence using Descript. It may contain errors or inaccuracies.

Amith Nagarajan: [00:00:00] you're doing your job as a leader if you push your people to learn this stuff

Amith: Welcome to Sidecar Sync, your weekly dose of innovation. If you're looking for the latest news, insights, and developments in the association world, especially those driven by artificial intelligence, you're in the right place. We cut through the noise to bring you the most relevant updates, with a keen focus on how AI and other emerging technologies are shaping the future.

No fluff, just facts and informed discussions. I'm Amithh Nagarajan, chairman of Blue Cypress, and I'm your host.

Amith Nagarajan: Greetings, everybody, and welcome back to another episode of the Sidecar Sync podcast. This is a special episode. It is part two of a two part series on the fundamentals of artificial intelligence. If you missed last week's episode on the fundamentals of AI part one, please go check it out. It's available everywhere podcasts are listened to as well as on YouTube. And in [00:01:00] this part, we're going to dive deeper on AI fundamentals for associations and nonprofits getting into a bunch of use cases. So I'm super excited to get into that with my co host Mallory. And before we do that, though, let's take a moment to hear a word from our sponsor.

Mallory: Today's sponsor is Sidecars AI Learning Hub. The AI learning hub is your go to place to sharpen your AI skills and ensure you're keeping up with the latest in the AI space. When you purchase access to the AI learning hub, you get a library of on demand AI lessons that are regularly updated to reflect what's new and the latest.

And the AI space. You also get access to live weekly office hours with AI experts. And finally, you get to join a community of fellow AI enthusiasts who are just as excited about learning about this emerging technology as you are. You can purchase 12 month access to the AI learning hub for 399. And if you want to get more information on that, you can go to sidecarglobal.com/hub.

Mallory Mejias: [00:02:00] Welcome back, everyone, to part two of the Fundamentals of AI series that we're doing on the Sidecar Sync podcast. In last week's episode, if you listened to it, we got into the nitty gritty of AI, the building blocks of AI, what it is, what it's not. So like Amithhh mentioned, if you haven't checked that out, Go do that first.

And then in today's episode, we're getting into the practical applications of AI so for the first segment of this episode, we want to do kind of a quick fire of 10 AI use cases for associations will present these use cases as a quick problem that associations might face a quick solution and then have a brief discussion on maybe tools that are out there that you can use right now to Enact those solutions.

Then we will get into AI resources and next steps ways that you can continue your AI Education and the things you need to do within your organization to make sure you have a culture that is ready for AI So without further ado, let's get into these [00:03:00] AI use cases The first one that I came up with chat GPT was a problem That your generic communication is failing to capture your members interest or potential members interest.

The solution AI driven personalization, Taylor's messages based on individual member preferences, increasing relevance and engagement of your messages. I wanted to provide an example of an email I got recently I am one of the first people probably like many to say I don't really like when people Cold reach out to me.

I don't read the emails I don't think they're all that effective but I got a really good one the other day and it was because it was And I'm assuming they used AI in some capacity to write this email to me, but it was actually about the podcast. And it said something along the lines of, Hey, Mallory, you've posted 15 episodes of your podcast.

We see that you're posting it on YouTube as well. Do you feel like you're getting all the engagement you could on your podcast? And I sat there, I'm still thinking about it. Weeks later, I didn't respond and I didn't book [00:04:00] time with them because I really wasn't interested overall in what they were offering, but.

Amith Nagarajan: Okay.

Mallory Mejias: Thinking about how personalized that message was to a project that I'm actively working on, I respected it and I definitely went and looked at their website as well. So that's just an example I wanted to share with you about the power of personalized messages. Amithhh, do you feel like associations leave a lot on the table in terms of tailored communications?

Amith Nagarajan: Yeah, completely. I think there's a ton of opportunity. You know, the, the degree of personalization that I tend to see out there for most associations is doing a mail merge where they include the members first name and an email. And that might be a stretch for some folks. So, yeah, I don't mean to be critical, but this is an areAI really feel associations need to step on the gas with both in terms of the things that have, frankly, been available for a long time with segmentation, but more importantly, to the extent of the example you provided of really personalizing with content elements that are relevant to the individual. Uh, so there's a lot of opportunity out there. I think there's examples within, say, membership where you're [00:05:00] sending an email for membership renewal, but you're outlining. Really, in compelling terms, the reasons why that individual should look ahead to the future and want to stay a member are so much you can do with AI there. Think about event, uh, emails, both marketing emails to get people to sign up and also prior to the event and post event. You can tailor those based on interests of people based on their prior behaviors. There's so much you can do to make the quality of your offering so much higher. So I'm really excited about personalization as a broad category.

You can personalize your website, you can personalize pretty much anything you can imagine, and AI is making that super easy.

Mallory Mejias: Yep. And what Amithhh is talking about, you can go out and do that right now. If you want to segment your email list, you could go to a tool like ChatGBT or Claude or, um, Gemini Ultra right now and ask it to write tailored communications for each segment of your list without giving it and identifying

information from your list. Uh, and also in looking at this use case, I found a tool, just one [00:06:00] example called smart rider AI. Now full disclosure, I have not used this tool, but you can do one to one communication this way. And it uses an individual's LinkedIn profiles. You can drop in their company URL and you can even have it.

Reference a company's Google reviews and use those reviews to help write an email. Now this is of course if you're reaching out to a company, but very interesting to know that there are tools out there that do this right now.

Our next AI use case. You have lots of speaker proposals to sift through for your annual meeting.

Solution use AI to analyze and select speaker proposals. I mean, this is something we've talked about on this podcast before. Can you elaborate on the vision you have for AI helping that speaker selection process?

Amith Nagarajan: Sure. I love this use case because I think it can both lead to improved efficiency for the team members within the association and the volunteer leadership, but also it can produce a better product. If you do a better job of getting the right speakers in, you serve your audience better. You get more people at your [00:07:00] meeting.

You produce more durable content for evergreen purposes online and on and on and on. So if you do this really well, It's a differentiator. It's strategically valuable. So the problem is, is that, you know, you only have so much time and say that you have an upcoming annual conference and you have 250 sessions, which you get 2000 proposals or 500 proposals.

It's a lot to read through. Um, you could build an AI workflow, what we call a multi agent solution that can literally automate very large swaths of that entire process. But for this podcast, I thought I would speak to just a simple portion of that. One of the most important things you have to do is filter on the top of the funnel, meaning you get proposals in, you get a lot of them in, and many of them don't even meet your technical requirements. They might not have a qualified background if that's required, they might not actually give you enough information, they might be on topics that are unrelated to the theme of the conference. There's a lot of requirements at a very basic level that AI can filter out. So imagine if you had a process where you simply feed into an AI the request for [00:08:00] proposal or request for speakers where you outline the requirements and then you put each of the documents in and you simply say, Hey, does this proposal fundamentally meet at the basic level? And it's just yes or no. And you could do that one by one through chat GPT.

It

Mallory Mejias: uh,

Amith Nagarajan: save you some time, but maybe not a lot because you can scan it quickly, but you can build that into an automated workflow very rapidly and quite easily where you take all those documents as they come in. Put him through a process like that and then filter it down. So instead of 2000 proposals, you're looking at 1000 doesn't fully automated just with that one step.

But that first step can save you a lot of time. So you can narrow your focus and put your energy towards the proposals that actually do meet your requirements. Of course, you could take that idea far further and you could have another stage in that AI pipeline that actually grades the proposals, looks for innovative content. You could have another stage of an AI pipeline where you look for biases and selection where you say, Hey, it's not going to work. These are real similar to everyone. We've always had, you know, what have we filtered out that we potentially could, you know, get in here that could be more interesting or more [00:09:00] diverse.

So there's a lot you can do in this domain. I'm really excited about it. I expect to see lots of products in the market that do this. Um, and I think it's gonna save a lot of time, but also create a lot of value for the end audience.

Mallory Mejias: Do you need to be pretty technical to build a workflow like that?

Amith Nagarajan: Right now, I'd say, yeah, you have to have a little bit of technical skill. This is also an area where AI is radically compressing, you know, this issue. Those curves we looked at in the last episode. The ability to do software development is another curve, right? And so human level capability and software development.

We're not there yet, but we'll be there pretty soon. So very soon you'll be able to go to an AI and say, Hey, I want you to build me a custom program that does this. I receive my RFP submissions from the proposals. They go into this database. Once they go into this database, I want you to run it through this process and then update the database with a flag that says whether it's approved or not or whether it's technical, meet the technical requirements. You'll very soon be able to do that. Right now, you can hire a programmer and this would be a super easy thing for a developer to create. It's probably, you [00:10:00] know, a handful of weeks of time. So it's within the, the budget, uh, of most associations to be able to do , some things like that.

Mallory Mejias: Thank you. Our next use case, you have an overload of irrelevant information for your members. A solution here, AI can curate personalized content, ensuring members receive relevant and valuable information. Amithhh, I know we've talked about RasAIo on this podcast before. They are an AI curated newsletter platform, essentially.

At a one to one level, the person receiving the newsletter is getting articles that are most interesting to them based on their previous engagement in the past. I know when we've talked about this before, Amithhh, some people would say you don't want to feed members the same type of content over and over.

For an example, like what if one time you click on an article in a newsletter about COVID, you don't want to have all of your future articles be about COVID. Can you talk about how AI is kind of advancing past that?

Amith Nagarajan: I mean, the short answer is it's just way smarter than that [00:11:00] already, and it has been for some time. So a lot of times if people are concerned about something like that, they need to just dig a little bit deeper and see if that's actually how the AI works. Originally, AI recommender systems did have that kind of death spiral type effect where if you clicked on one thing, you got that forever. Obviously, that's not desirable. If you're trying to inform a broader audience, it's somebody actually is built to do that intentionally, like in the world of social media. You know, my biggest gripe with all these companies is that I shouldn't say all with the companies that I'm familiar with, like the world of Facebook, for example, their algorithms designed to draw you in by actually doing exactly what I just said.

You don't want to do if you're trying to inform people, which is they give you content in a way that is specifically designed to bring you back. A lot of times it's the same things that they know you're already interested in with your points of view, not a diverse point of view. of course, they use a lot of emotional tactics to upset you because that's the type of content that draws people in faster.

So you can use AI for good and you can also use it that way. about a newsletter is it's one form of curation. [00:12:00] There's other challenges and opportunities like your website probably is overwhelming for most users. It's a lot of fun. It's a lot of fun. of the people that are out there. Um, AI s.

Gotten smart enough to help you a whole bunch, and there's a lot of tools for it. main thing I like to point out here is that your problems are not your customer problems, meaning your customers don't care about your problems. Your members don't care about your problems at the end of the day. At the end of the day, they care about their problems, and if you make it hard for them to get the content that they want, they're gonna go somewhere else. And that's the cold reality of today, and it's gonna get worse with AI and making it easier for other people to produce content. They may not be at the same level as yours, but it's good enough to offset some of your audience. And so that's a big risk. But the opportunity here is to do it better than anyone, because you have a lot of data on people, and you also have great content.

Mallory Mejias: Our next problem. Difficulty in identifying members at risk of lapsing. A solution here, AI models predict which members are likely to not renew, and why, allowing targeted retention [00:13:00] efforts. Now Amithhh, of course I think all the time in the realm of ChatGPT, because I use it every day. I assume you could drop in member info into ChatGPT and have it use the data analysis feature and have it, , figure out which parameters you can use to predict which members won't renew.

But you shouldn't do that, right? Because there's a lot of sensitive info you have for your members, and it is not safe, I would say, to put that information into ChatGPT. What are some other ways you could do something like this safely?

Amith Nagarajan: Well, this is an area of predictive AI at the core issue of you're trying to predict which member will renew and which one will not, right? That's a fundamental predictive problem, just like predicting which piece of content someone would like and not like. And so there's, there's models out there you can create that are trained on your member datAI n order to solve for this specific issue, you know, we've talked in the past about the idea of, you know, narrow AI or a eyes that are highly specialized. This is a great example because you can have a member renewal. [00:14:00] Model that's trained perhaps on broader member renewal data beyond your own, but trained specifically on yours and then predict for you which members are likely to renew, which ones are not and possibly suggest a reason why they may not renew .

Then, of course, that then bridges into the other part of a solution. Because what do you want to do with that? Well, if you know that for every member, maybe you can do some manual interventions. But would be really great is if you then tacked on to the predictive AI of renew, not renew prediction. What to do next.

And so then you can add a generative AI layer, which says, Oh, well, what should you do with each of these people? And then actually take those actions and connecting that data to your marketing automation platform, like a hub spot or MailChimp or whatever you're using.

Mallory Mejias: So in this situation, probably best to have your own model. Is that what you're saying?

Amith Nagarajan: I don't know that that's necessarily the case. definitely important to have a predictive model for member renewal forecasting of any degree of reliability. Because if you just take your data and ask generative models to try to make sense of it. They're going to do one of two things. They're [00:15:00] either going to write a program which then attempts to invoke the use some kind of underlying predictive model.

But you don't necessarily know which one. , or they're going to try to interpret the datAItself, right? Which it can't do because the context window of these models isn't big enough to deal with data of any significance. But these models are general purpose. They're not going to be able to infer. Uh, really any valuable patterns for a large data set directly.

, that's why there's these specialized tools for different things. So there's ML out there specifically looking for, , these types of problems , and solve them quite, quite effectively. In fact, this is a class of AI that's far more mature than generative AI. So you can use a combination of different tools.

Mallory Mejias: The next use case, you have a high volume of routine inquiries that overwhelms staff. That sounds pretty relatable solution. AI chat bots can provide instant responses to common questions, freeing staff for complex tasks. Amithhha, I always enjoy when you talk about this one, can you talk about AI and reducing friction for your members, for your customers?[00:16:00]

Amith Nagarajan: Yeah, I think if friction is public enemy number one for all brands and associations and nonprofits have to think about this far more directly than in the past where. You know, people, it's like, how much will they put up in order to interact with you? Right. And should it be that kind of experience? You want to have as low of a friction environment where people can get to information and the solution that they're looking for. So member support, customer support. That's clearly an area where chatbots have actually been used for a long time, even before generative AI. But there's a big difference. In the past, chatbots were designed to reduce your workload. And so chatbots were like, Oh, well, if we put a chatbot on our airlines website. You know, we'll be able to force feed that chat bot to some of our customers and force them to interact with that chat bot before they talk to a human being.

And that's a great way to lower our cost in terms of our call center. And you know, you've seen that play out. , the chat bot on the website up until recently has been no better than like a never ending phone tree that you used to call into. , it's a new version of pain. , so traditionally chat bots have been absolutely horrendous and I wouldn't want any part to do with a chat bot.

Normally I just say connect me to a real person, right? That's the first thing you do or you say agent or you press zero when you get into one of those loops. So chatbots in their prior incarnations were all about saving the company money and making the company's life easier. But that's not the case going forward. It's about providing a better experience to your members. So imagine, of course, not only solving the problems your member services staff can deal with, like all the usual member services requests. But also imagine if that chatbot now had all the knowledge from your domain. It knew everything that even your best members weren't able to keep in their heads.

And then that chatbot was available to not only solve member service things, but also be a knowledge agent. And it was a one stop shop and capable of very rapidly helping anybody with any problem they have within your domain. That's the era of chatbots we're getting into now where they not only save your customer's time, but they give them a far better experience that's at an AI scale that no human could possibly do that.

So that's the opportunity here, is to not only automate something that's manual right now, but to level up, and to say, hey, we can provide something nobody else can provide, because we have all that great content, we have the brand reputation to do it, um, and of course we're gonna solve our, you know, routine member service type questions too.

Mallory Mejias: Okay, I think that's really interesting the way you put it, that shift from helping the company to kind of reducing friction for the, the end user, the consumer, the member, whatever it may be. Amithhh, I'm sure there are currently AI chatbots on the market for association specifically.

Amith Nagarajan: There sure are. In fact, our family of companies has one called Betty Bot, which powers a lot of associations, , websites and can be embedded into apps. And there's many others as well. You know, Betty is specifically built for the needs of associations and nonprofits, but there are a lot of ways to deploy chat bots. Really, the key is how you're going to integrate it into your environment. You have to make sure you have a technology that you can deploy securely. Uh, that's going to give accurate answers based on [00:19:00] your content, not on like the world's content on. In some cases, you might actually want that chat bot to have certain agent type capabilities.

And we've talked about that on this podcast a lot. The idea of agency is when the AI can actually take action on your behalf or on the user's behalf. So, for example, if a chat bot is having a conversation with the prospect or a customer or member, and the individual says, Yeah, I think I'd like to go ahead and renew. Well, it would be a very organic thing for them for the bot to say, Well, I'd be happy to renew for you. Would you like to use your credit card on file? And the person says yes, and it's done, right? And there's confirmation email in their inbox. Three seconds later, that's the experience you want to give.

That requires some degree of agency where you have to train the bot on several possible actions it can take and let it go ahead and do those things. And you can start with low risk stuff like that that are easily reversible. But there's some tremendous possibility there. Once again, you are saving your customer Uh, time and energy.

And you're also saving yourself a lot of effort, but also increasing [00:20:00] the outcome, increasing the value of the outcome.

Mallory Mejias: For our next use case, the problem, digital resources are not universally accessible. The solution? AI can generate real time captions and translations for videos, making content accessible to a broader audience. Amithhh, I know you love to talk about this one. Can you share your experiences with a tool called Heygen?

Amith Nagarajan: Yeah. Hey, Jen is a video translator where you can take a video of yourself speaking and it will translate it to any other language. And it's not just voice dubbing, which by itself is quite impressive, but it's capable of actually changing the video to make it look as though you're speaking that language. Uh, so that's quite powerful and it opens up the doors. I like to think of translation as not only solving accessibility in the traditional sense, in terms of language or, uh, capabilities , if you need to translate, you know, content from one form to another, uh, but potentially to open up the doors to Taking your content and repurposing it for different audience education levels, uh, possibly a variety of other translation problems.

So I think Heygen is a great [00:21:00] tool. There's many others. , we've seen an explosion of these tools in the market. Heygen's the one that just Mallory and I both have played with in the past and seen some pretty good results from.

Mallory Mejias: Yep, it's a great tool to translate your videos, and then ElevenLabs is another one that Amithhh has tested out and that is a great tool, like let's say if you had a podcast and you wanted to translate it into other languages, you could use a tool like ElevenLabs to do that.

Amith Nagarajan: Yeah, the 11 labs also allows you to change the voice so you can do what's called voice dubbing there where I record something or a video or just audio of me speaking and I can have that turned into any other voice. I wanted to turn into and that can be useful. Say, for example, you're doing a series of videos on your associations website and it's Welcome to the association videos

maybe they're mostly either animated or perhaps their, um, slides and maybe a little bit of a person speaking, but really mostly just content on the screen, coupled with a human voice. And perhaps you want to standardize all of them. But you have a bunch of videos. Different people are gonna record, but you'd like to have one standard voice.

And you can do that with 11 laps voice dubbing, which is pretty cool.[00:22:00]

Mallory Mejias: Indeed, and I should have also mentioned this at the top of the episode, you don't have to be quickly jotting down all these tool names and figuring out how to spell them, we will be listing those in the show notes as well, if you're interested. Our next use case, associations struggle to understand member engagement levels.

Solution, AI tools analyze interaction data across platforms to identify trends and improve engagement strategies. This seems like a dream, Amithhh, but it also seems like we need to talk about the idea of a CDP for this.

Amith Nagarajan: Sure. Well, a common data platform or CDP. Is simply a data repository where you get all of your datAIn one place. And the way to do that is much more complex than we have time for in this conversation. But the essence of the ideAIs to get your datAInto a universal location and then to be able to use that data for insights like understanding member engagement, being able to do personalization at scale. And the challenges is that inherent to the modern world, is that you have many systems. You probably have an AMS or [00:23:00] CRM type of system. You have a financial system. You have a learning management system. You have a website. You have email software. You have marketing tools. You probably have half a dozen to a dozen major applications to use, and there's probably dozens more that you use that are smaller. And the interesting thing is that some of the data that comes from applications that are perhaps in the periphery of what you consider your core data set actually might have the most value In certain types of analysis. So on the member engagement problem that Mallory described, if you think about engagement, the frequency of engagement is a really important aspect, and the data you have in your AMS is probably about member renewal. Perhaps there's meeting registration date in there. There's information about committee participation. That's great, but that data doesn't change that frequently. But what does change frequently is people's online habits, their visits to your websites, the clicks on your newsletters, the way they interact with you on social media.

All of these things are data streams that you can pull into a C. D. P. And once they're in a C. D. P. You can then let A. I. Loose on them, so to speak. Um, and get a much better insight of what's truly happening with your audience than what you have any idea of from your traditional data sources. So reason the CDP is important is you can do this type of thing without it.

You just have to manually stitch together all this data from different places. It's a lot less durable. It's hard to, do it continuously over time. And then also if you build this concept on top of a system like an AMS, you ultimately don't own it. So if you switch AMSs, you got to redo it.

It makes it harder to switch. So we think a CDP is a good idea because it just basically allows you to solve for these broader problems than you're able to solve with any one particular system.

Mallory Mejias: Yep, it's kind of impossible to limit that conversation to a quick fire AI use case, but I just thought we have a whole episode called the data episode that we recorded several weeks ago, maybe a few months ago. , so check that out. If you want more information on this idea of a CDP, that is a great episode to go to.

Amith Nagarajan: And we also have a recording of a recent webinar that we did called Own Your Data, [00:25:00] Own Your Future, which dives even deeper than the podcast, so we'll include both of those links in the show notes as well.

Mallory Mejias: Our next problem use case, you have a All this great video content from your conference from your annual meeting, but no one really engages with your old video content solution. You can use AI to repurpose content into articles, blogs, clips, podcast episodes, so on and so forth. So I've mentioned this on the podcast before, but I'll give you all a brief overview.

We record this podcast on zoom and then we use an AI powered editing tool called descript that generates a transcript from the episodes really quickly. From there, we're able to actually edit the the audio for the podcast using the transcript, which makes it much easier than having to sit there and just look at audio levels for a few hours.

Then we take that transcript. We work with it in chat, UBT to help us create blog outlines to post on the sidecar blog. And then we can use the video recording from zoom in a tool called Munch to generate short clips that we [00:26:00] can share on social media. Munch pretty much automates this whole process. It even does work on the back end of figuring out what keywords might be trending on the internet and then using that to inform which clips it generates.

So highly recommend that one. And in our previous episode with Sharon Guy, she used a tool, I think it was called Opus or Opus Clips or something like that, but lots of ways out there to repurpose the content you have. Amithhh, do you have any. Additional thoughts there.

Amith Nagarajan: No, I think that's a great summary and I definitely would encourage people to consider this because there's so much video content and associations and much of it's just sitting there.

Mallory Mejias: Our next AI use case problem associations manage vast amounts of content, making it challenging to organize and retrieve specific information efficiently. Solution AI powered tagging and systems can automatically categorize content based on themes, topic, or relevance. Simplifying.

Content management. Uh, Amithhh, we actually get a lot of interest in this. I feel like of people asking about it. And I [00:27:00] know you've told me, and some of our other AI experts in the family have told me like, Oh, this is very simple. Like this is not something you need to spend a ton of resources on. So can associations create this in house pretty easily?

Amith Nagarajan: Yeah, I think so. , it's simple in the same sense as the earlier example when we were talking about, um, event speaking proposals, there's a little bit of technical work that has to happen to make it automated fully. But , the essential ideAIs, is very simple, which is you take a piece of content You run it through a language model like chat GPT and you say, what are the tags associated with this article?

And then you use that information. However, you use tagging today on so for associations that have gone through the painstaking effort to craft a taxonomy and then to manually tag content with the taxonomy. It's definitely a big relief for people who are doing that. And a lot of people haven't even done that because they're saying it's not something they can do.

They can't keep up with it. And so being able to create this AI power tagging can be very valuable for internal classification and also for external consumption. [00:28:00] Now, one thing to keep in mind is you don't need to do this in order to make it possible to use tools like Rasa or Betty or others that are out there for content curation or content personalization because the AI actually doesn't use the tagging very much. The AI is way, way beyond tagging. Tagging is for us. Tagging makes it easy for us to quickly look at a piece of content and say, Oh, it's about these three topics. There is utility in that, but the AI reads the entire article in less time than it takes us to read the tags, and

Mallory Mejias: Okay.

Amith Nagarajan: it gets the full essence of the article, and that goes far deeper than the tags themselves.

But tags are still useful. I still recommend people have them available on their websites and articles because it's a great way to give people just a super quick understanding of what they're looking at.

Mallory Mejias: And the last AI use case, we couldn't really do a list of AI use cases without talking about content creation, but I know that is the one that gets the most press normally, so I saved it for last. Problem. Associations struggle to consistently produce fresh, engaging content for their audiences. Solution.

AI powered content [00:29:00] creation tools can generate articles, social media posts, and marketing materials, ensuring a steady stream of high quality content tailored to the association's needs and audience preferences. So, you probably already know there are lots of tools out there for you to do this. ChatGPT is one of my favorites.

Gemini, text to image models like MidJourney, DALI, and text to video models like Runway, and so on. soon to be available, Google Lumiere and OpenAI's Sora, which we talked about last week. But Amithhh, I think you kind of have an interesting approach to this using a tool like Otter. So can you talk about how you do that?

Amith Nagarajan: Sure. , Otter is one of my go to's. I have it on my phone and that's really where I use it the most. At the simplest level, it's a tool for note taking and you can type into it, but you can also record into it. And so I will talk to Otter and Otter will record what I'm saying.

And it does all the things that we talked about, transcription, summarization, et cetera, et cetera. But it's very high quality. It does a really good job of capturing a lot of nuance and a lot of like specialized words, and [00:30:00] it seems to be getting better over time is as I've corrected it on. Then I take the transcripts out of Otter.

And a lot of times I'm, you know, walking around New Orleans here and You know, just seemingly talking to myself as I walk around, and it's always fun to get those kinds of looks, but, um, you know, I'm talking to myself. I record 10, 20, 30 minutes of content. I come back and I take the transcript, and then I feed it to either something like Gemini or chat GPT. And in there, I'm able to say, Hey, summarize this for me. And then I ask it. Well, I'd like to create a blog from this topic, Can you suggest an outline? And it'll take what I've said, and it'll generate an outline for me. And then there's more specific prompting strategies to make it work specifically your content because the key to it is to have it not go off on its own and do things outside of what you said.

I like to produce my own content. , and I'm certainly open to input from the A. I. But I'd like to make sure the content is actually coming from me. So even though A. I. Is assisting me with it, the output, I know the words are selected by A. I. But it's my ideas.

And then ultimately, the outcome is you get a blog outline and then you create the blog itself. But the other thing that I do a lot of is I'll have an idea And I'll say, Well, I really want to talk about this topic. It's an important topic that I want to get out there. And I will go and brainstorm the idea back and forth with a chat tool like chat GPT, and then I'll come up with ultimately an outline.

Then sometimes I'll jot down notes, or I'll speak to that outline a little bit and record something, or just, you know, share a little bit more content, and then I'll put it back into that same process. Ultimately, I feel as though the work is very much my original work, even though the A. I. Has done a lot of the heavy lifting for me. It's almost as if I had an assistant and I actually had the assistant just sitting there with me side by side and doing those things, just working at a very rapid pace. So that's how I go about it. I think associations can do unbelievable work with this. comfortable with it is really important. You have to be willing to do. Use these tools. Obviously, I think it's important to disclose as we are now and we do all the time that we are heavy users of this technology. So it's not a [00:32:00] surprise that it was AI assisted content. I would definitely urge everyone to stay away. From the idea of autopilot or near autopilot content creation, where you just go to chat GPT or similar tools and say, give me a blog post about blah, you can spot those posts pretty easily because they're really shallow and, it's just not a great way to go about it. You want to put the effort in. It's just that now I can put together a 3000 word anchor post for a blog In about an hour of work that might have taken me a day previously or longer. So, , that's the thing. It's a force multiplier for sure, but it's still my ideAIs still my work. So I really think this is an opportunity area for everyone, including associations and nonprofits.

Mallory Mejias: Now that you mentioned it, Amithhh, I'm pretty sure while I've been driving before I've seen you walking around, probably talking to yourself, but you know, Hey, I'm saying take advantage of it. I right to say this is how you wrote Ascend as well, using a process like this?

Amith Nagarajan: Yeah. So back in the spring and early summer of last year, [00:33:00] 2023, I, um, was putting together Ascend, uh, with some collaborators here at Blue Cypress. And a lot of the work that I did was through this process, went to Otter and keep in mind, you know, that's almost a year ago now, a little bit less than a year ago. And the AI tools have gotten a lot better. You know, I could probably do the exact same project in a 10th of the time now compared to what I did back then. Cause the tools were pretty rudimentary a year ago.

Mallory Mejias: And as a side note, you can get that book for free, actually. You can download a free copy at sidecardglobal. com/ai if you're interested. Amithhh, that wraps up our use case section. I want to get into resources and next steps. for the people that have been with us on this long journey of two episodes that have learned all about AI, heard all these use cases, and are now saying, okay, what do I do next?

So something that we always recommend, of course, it's kind of the reason for this podcast, is starting with education. So before diving into AI, or as you dive into AI, You should focus on educating your team and yourself about AI s. Potential limitations and [00:34:00] applications that are relevant to your operations.

This foundational understanding is going to guide the selection of appropriate AI solutions and vendors, ensuring a match with your organization's strategic objectives and technical infrastructure, emphasize the value of knowledge in navigating the rapidly evolving AI landscape, making informed decisions, and fostering an innovative, adaptable organizational culture.

Amithhh, what do you think are some foundational AI concepts that associations should prioritize in their education efforts?

Amith Nagarajan: Well, I think that, honestly doesn't really matter that much as long as you get started with something. So I think you need to look in your organization and say, who are the people that have been the champions of going out there and playing with this stuff and engage them, ask them what they've been using, find out if there's some particular, you know, content resources or tools that they like a lot, because that probably will translate to your other team members. Obviously here at Sidecar, we've got a bunch of resources for you. We have this podcast that we do every week. We have a ton of free content on our website. We run a monthly AI [00:35:00] webinar. That's an intro to AI that's available for free. We've got a lot of great free resources. Um, but I think that, uh, the key is just pick something. Um, it could be non association and not nonprofit specific. That's okay. A lot of this stuff is pretty, uh, transferable from one space to the other. But obviously what we're doing is intentionally designed to bring the more broad content and and specifically contextualize it for this community. I think that the foundational ideas that you need to get started with are really about like getting your head wrapped around really the capabilities as much as anything else and then starting to experiment once you have those basic ideas.

Mallory Mejias: A question that sometimes comes up with education, Amithhh, uh, is to mandate or not to mandate AI education within your organization. What are your thoughts on that?

Amith Nagarajan: Well, I'm a fan of people buying into things and, uh, having a vote, essentially, in a sense. But if they vote incorrectly on whether or not they're going to be educated in AI, I think that's where you need to mandate it. So, I think that if you're, if [00:36:00] you've done nothing, I'd start with a smaller group of people who express interest, get them going, and then knowing full well that you're going to mandate everyone learn this stuff.

You know, having an employee on your team that doesn't know how to use AI like having an employee on your team who doesn't know how to use a computer. Um, you wouldn't tolerate that. You're not going to tolerate that going forward, and you're doing your job as a leader if you push your people to learn this stuff, because whether they work for you or somewhere else in the future, um, you are tremendously underserving your team if you don't help them rapidly learn this stuff, because it's going to shape the way all jobs are done in the future. Every employee needs to learn this stuff. So I'm kind of heavy handed about it, and I don't think there's a lot of time to waste. So that's part of the reason why I'm being heavy handed in my response to this particular question. Uh, but I think, you know, you can do both at the same time.

Mallory Mejias: Absolutely. And as Amithhh mentioned, we've got a lot of resources over here at Sidecar. We are launching this new Intro to AI webinar that is essentially for the beginner, someone who's just dipping their toes in, maybe the person [00:37:00] listening to this very episode. , we run that once a month and you can get more information on it at sidecardglobal.com/AIwebinar. We also just posted a blog recently about, I think it was top five or six generative AI courses. So we'll link that in the show notes as well. If you're looking for a course to get started. Um, and then in addition to that, Amithhh, I know we're both big fans of Ethan Mollick's uh, newsletter that he sends out.

He's also a great person to follow on LinkedIn. Um, the information has been a really great AI resource that Amithhh shared with me. Amithhh, any others that you can think of that you'd want to share with listeners?

Amith Nagarajan: Yeah, I would point people, especially those in the marketing discipline, to our friends at the Marketing AI Institute, Paul Reitzer, Mike Caput, and the team over there do a great job. They also have a weekly podcast, which is fantastic, and they've got a lot of content and resources. So if you are a marketer and you want to go deep on like the marketing domain, specifically, they've got tons for you.

Mallory Mejias: Now kind of hand in hand with that education piece is the guideline piece. Establish clear [00:38:00] rules and protocols that encourage team members to explore AI tools within a safe framework. These guidelines should emphasize responsible use, data privacy, and ethical considerations, while still fostering a culture of innovation where staff feel supported in experimenting with AI solutions.

We are actually working on a AI guideline lesson that we'll be adding to the AI learning hub soon. So stay tuned for that. But Amithhh, what do you see generally at a high level as a good framework for creating AI guidelines?

Amith Nagarajan: Well, the best framework is one that you number one have and number two actually train people on. So I don't think you need to start with something super complicated. I think you start with something simple that just says hey these are the things you need to be aware of That are concerns and these are the things if there are a list of things you must not do that's okay I'd keep that list fairly short, but I would actually focus the guidelines on making sure people Are able to do things, right?

You're essentially enabling people in a way that you find to be in alignment with your values and your, uh, information security and governance concerns. Um, so one of the things you want to tell people not to do is to export your data from your AMS and dump it into chat GPT to ask questions that might be tempting and sound cool, but that is not secure on.

People don't know that. So because of that, it's incumbent upon you to teach them that, and guidelines are one way to do it. Another guideline that I think is really good is, Thou shalt not do AI experiments until you learn AI. So it's kind of like, would you let your teen go out there and start learning how to drive a car?

Not learning how to drive a car, but driving a car around town before you made sure they went through driver's ed. Probably not. And AI has a lot of potential pitfalls, like the one I just mentioned, that don't necessarily seem that obvious to a lot of people. And so, I'd probably put in the guidelines that, hey, all of these tools that we're making available to you, that you're welcome to use, blah, blah, blah, and blah, like a list of ten tools.

You first have to do this course, um, or you read this thing or something that ensures that you have some basic level of [00:40:00] understanding? It doesn't need to be onerous. It could be, you know, something very simple. So I think those are key points to keep in mind. Um, you want to make sure that you're thinking about defense, which is around cyber security, ethics, everything else that comes with it, while not being so overly structured and strict that you basically snuff out innovation for the people who actually listen to you and then basically create rogue environments. So the people who don't and all of you, including myself, have people who listen to you and do what you're told. And we have plenty of people who just do their own thing. Um, and that doesn't mean they're bad people or bad employees, but some people are just gonna go use these tools whether you like it or not.

And if you don't create an environment that teaches them how to safely do it, they will do it in a way that's probably unsafe.

Mallory Mejias: So you would say education and guidelines kind of go hand in hand or at least can occur at the same time. Best guideline is the one that you have. I agree with that. That's kind of a funny statement. And then you encourage the experimentation. And I agree with you about having guidelines that [00:41:00] kind of snuff out innovation. It can be challenging to create those parameters and still encourage staff to feel empowered to try out new tools. Amith, when thinking through the experimentation piece, What do you think about creating, I've heard the terminology, you know, like tiger teams, like an AI team within your organization to focus and experiment.

But I've also heard you say that it's important for everybody to feel like they're empowered to do that and not just a team.

Amith Nagarajan: I think it's important to do both. So if you have a group of people that are enthusiastic about AI form a club, build a team, do something where you have a way that you're convening those people and getting them to talk to each other. They might be across departments, different geographies. That's one of the powers of what a leader can do and bring people together help them Just share ideas.

And tiger teams typically are formed to solve particular problems. That's awesome to pick one thing. One thing that you think is a candidate for AI to solve. That's not necessarily hard and put a team of smart people together that are passionate about it to go [00:42:00] crack, crack that and, you know, let them break things along the way in a sandbox.

Right? And so the break things comment it. In a sandbox is important. You don't break things out in public necessarily, but you experiment at least internally. Um, and you get out there and you get a team like that to work on it. But at the same time, the other part of your question is, you know, you get everybody involved. I'm a big believer that you have to democratize AI. You can't hold it close because you're scared. You can't hold it close because you think it's something that needs to be centrally managed. None of these things work that way. And so if you try to do that, you're going to discourage a lot of people. And there's some people who are still going to do whatever they want without your approval.

And that's really where you get into the danger zone. So I would broadly encourage everyone on my team to learn AI. Give them resources. Give them access to the AI Learning Hub. Give them access to something, whatever you choose. that will encourage them and know that you are behind them. You are helping their professional journey.

You're helping them grow and learn on. Of course, it's gonna benefit your organization at time, but you gotta focus on the person. Think about from their [00:43:00] viewpoint. Why should I learn AI? Why should Mallory learn AI? Why should anybody else learn it? You gotta think about their incentive, right?

And so you want them to help feel that they're getting better at their job, but you're also protecting them because a lot of individual workers Are in fact very concerned about their job. Whether you say their job is not at risk or not, um, people are concerned and for good reason because it's a time of great disruption. So by investing and educating your people, you're doing a lot of great things at the same time.

Mallory Mejias: Last question here, Amithhh, when you say experiment, are you saying just create a free account, try out some new tools, see how they work, or are you thinking have a specific goal in mind or even like you said with the Tiger team, having a specific problem in mind that you're using AI to solve or is it both?

Amith Nagarajan: It is a mixture of things. So if you're an organization that I would say is kind of in the middle of the distribution for associations and hasn't really done a lot of innovative stuff at all, maybe you embrace the idea, but you don't really, you're not great culturally at actually encouraging [00:44:00] innovation practically, is really the majority of associations.

It's nothing to feel bad about. It's just, you know, where, where these organizations have come from. If you're in that camp, I would encourage small things. So I would get people to do things like try out accounts on different tools with particular data sets that you might give them saying, Hey, you can't use our member data for chat.

Gpt. But here's a sample of 1000 pieces of data you can use or some documents that are public domain already that we published our website that you can use. This is stuff you can play with. So give them a sandbox and let them loose. , now a more structured experimentation methodology would be awesome.

But it's something it's kind of like running a marathon before you've run a lap. And so I think for organizations that are further on down, you know, the path of of driving innovation in their culture. Um, you could definitely create a methodological approach to experimentation. where you have resources allocated, financial resources. Specifically, as well as time, you have structure around it where you know, the balances is in the best R.N. D. Labs in the world have both structure and immense freedom, and it sounds like they're opposing forces. What the structure does is, Hey, we know when to kill things off, whereas the freedom is we get to explore things from a multidisciplinary perspective. We have broader time frame sometimes to explore not forever, but like enough time to, like, really think about things rather than just focus on doing things. they also have structure where they are accountable for producing output, and they're also accountable for justifying the existence of projects. And you kill them off quickly if you realize that they don't have a future. And I think the best methodological approach is to experimentation combined structure and freedom, but that's a much more advanced way of thinking about it organizationally.

If you're not doing much right now, I would just go back to find a handful of people could be three or four who are interested in this stuff and give them a few bucks to go set up some accounts. I wouldn't be the free thing. I would actually pay for my tools because you get access to better user agreements.

If you, uh, the thing I always like to say is if you aren't paying for the product, You are the product. Um, it's certainly true in social media. It's also true for your data when it comes to free AI tools. So pay a little bit of money and, and then go and run around and test this stuff out.

Mallory Mejias: I feel like we've been on a long AI journey over the past two episodes of Meath. We started really basic, building blocks of AI, core components, different types of AI, and we've kind of ended it on this discussion of Best next steps, education, guidelines, experimentation. I'm hoping all of you listeners have, or watchers on YouTube have learned something insightful from this episode.

Hopefully many things. If you want to share any of that with us, you can on our dedicated Sidecar Sync space on the Sidecar Community, which we link in the show notes. So let us know. And if you have any questions or if you have things that you want us to cover on future episodes, let us know that too.

Amithhh, thank you so much. I will see you next week.

Amith Nagarajan: Thanks very [00:47:00] much.

Amith: Thanks for tuning into Sidecar Sync this week. Looking to dive deeper? Download your free copy of our new book, Ascend, Unlocking the Power of AI for Associations, at ascendbook. org. It's packed with insights to power your association's journey with AI. And remember, Sidecar is here with more resources, from webinars to boot camps, to help you stay ahead in the association world.

We'll catch you in the next episode. Until then, keep learning, keep growing, and keep disrupting.