Latest / Sidecar Sync / How to NOT Commit ‘Leadership Malpractice’: 10 Predictions for AI in 2025 | 61

Transcript

How to NOT Commit ‘Leadership Malpractice’: 10 Predictions for AI in 2025 | 61 · Sidecar Sync

• 0:00
  Speaker 1: If you are not preparing your people for the world of AI, it is my firm belief that you are committing an act of malpractice, because you are entrusted with the futures, in many respects, of your team.
• 0:12
  Speaker 1: Welcome to Sidecar Sync, your weekly dose of innovation.
• 0:18
  Speaker 1: If you're looking for the latest news, insights and developments in the association world, especially those driven by artificial intelligence, you developments in the association world, especially those driven by artificial intelligence, you're in the right place.
• 0:27
  Speaker 1: 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.
• 0:35
  Speaker 1: No fluff, just facts and informed discussions.
• 0:39
  Speaker 1: I'm Amit Nagarajan, chairman of Blue Cypress, and I'm your host.
• 0:42
  Speaker 1: Greetings and welcome to the Sidecar Sync, your home for content all about associations and artificial intelligence.
• 0:50
  Speaker 1: My name is Amitnagarajan and my name is Mallory Mejiaz, and today we are recording our last live episode of 2024,.
• 0:59
  Speaker 1: If you can believe that, mallory and we are going to be running episodes throughout the holiday season, because we know that you guys all want to hear from us and hear all sorts of cool things about associations and AI while you're in your holiday break, but Mallory and I won't be recording again live until early January, so happy holidays to everyone.
• 1:18
  Speaker 1: Before we get into our episode, let's take a moment to hear a quick word from our sponsor episode.
• 1:26
  Speaker 2: Let's take a moment to hear a quick word from our sponsor Introducing the newly revamped AI Learning Hub, your comprehensive library of self-paced courses designed specifically for association professionals.
• 1:33
  Speaker 2: We've just updated all our content with fresh material covering everything from AI prompting and marketing to events, education, data strategy, ai agents and more.
• 1:45
  Speaker 2: Through the Learning Hub, you can earn your Association AI Professional Certification, recognizing your expertise in applying AI specifically to association challenges and operations.
• 1:57
  Speaker 2: Connect with AI experts during weekly office hours and join a growing community of association professionals who are transforming their organizations through AI.
• 2:06
  Speaker 2: Sign up as an individual or get unlimited access for your entire team at one flat rate.
• 2:12
  Speaker 2: Start your AI journey today at learnsidecarglobalcom.
• 2:21
  Speaker 3: Meath 2024 is almost at its end.
• 2:24
  Speaker 3: I'm wondering if you have any kind of general reflections that you want to share before we kick off this special prediction episode of the Sidecar Sink.
• 2:34
  Speaker 1: I don't even have any way to know.
• 2:41
  Speaker 1: I know this episode is about our predictions for 2025, but I really have absolutely no clue what's going to happen next year.
• 2:47
  Speaker 1: It's just this year has been batshit crazy and next year is going to be an order of magnitude crazier than this one.
• 2:53
  Speaker 1: So you know, we'll see what happens.
• 2:55
  Speaker 1: I am reflecting on the year as just an amazing time.
• 3:00
  Speaker 1: All sorts of great stuff has happened.
• 3:03
  Speaker 1: I think that you know, for our family of companies, it's been just amazing on so many levels.
• 3:08
  Speaker 1: It's hard to think of another word.
• 3:09
  Speaker 1: Honestly, I'm running out of vocabulary, but it's been really really good.
• 3:12
  Speaker 1: It's also been just really intense.
• 3:13
  Speaker 1: So I think it's good that we have a little bit of a slowdown although things are not really slowing down that much, but a little bit of a slowdown before we hit the end of the year and go into 2025.
• 3:24
  Speaker 1: But how about you?
• 3:26
  Speaker 3: I'm just feeling excited for the new year but also incredibly grateful for the growth we've experienced at Sidecar, for the network that we've built, the community, the connections that I've made within the association community.
• 3:39
  Speaker 3: I was thinking this morning I'm approaching my three-year mark having worked in this family of companies, but it wasn't until I was at Sidecar that I was able to truly forge personal connections and relationships I felt with the association space and it has just been so rewarding and fulfilling to do this podcast every week with you and kind of hear feedback from people in the community.
• 4:01
  Speaker 3: So I cannot imagine what 2025 will hold, but it seems like not to, you know, give too many spoilers.
• 4:08
  Speaker 3: It seems like we did pretty good with a lot of these predictions for 2024.
• 4:11
  Speaker 3: I don't know if we played it safe, maybe.
• 4:13
  Speaker 3: So we'll see what 2025 brings.
• 4:16
  Speaker 1: Yeah, maybe.
• 4:16
  Speaker 1: So I mean, I think predictions are always an interesting thing to venture into.
• 4:20
  Speaker 1: I think it's entertaining, if nothing else, to think ahead and say what will the future hold.
• 4:25
  Speaker 1: A 12-month window is theoretically a reasonable time span to predict historically, but of course, with AI doubling in capability roughly every six months, it is actually somewhat difficult to predict what's going to happen, partly because of how people adapt to the AI.
• 4:40
  Speaker 1: So the AI's capability is one thing, but how people use it is what really matters.
• 4:44
  Speaker 1: And so what will people do in 2025?
• 4:46
  Speaker 1: Will a number of people who've been somewhat on the sidelines, will they take it more seriously and will they really step on the gas and go for it?
• 4:52
  Speaker 1: I certainly hope the answer to that question is a strong yes, but we will see on.
• 5:00
  Speaker 1: And what you're excited about this year and the relationships you've built in the market that's why I've been in this space for as long as I've been is that I feel that the people who are in this market, doing the work day-to-day on the ground in the associations, are really making a difference.
• 5:16
  Speaker 1: They're making a big impact in their missions and, of course, our goal is to support all those missions through what we do.
• 5:22
  Speaker 1: So I'm really happy to hear that experience has been rewarding for you this year.
• 5:27
  Speaker 3: Absolutely Well.
• 5:29
  Speaker 3: In the first part of today's episode we're going to recap those 2024 predictions quickly.
• 5:34
  Speaker 3: I will let you know what the prediction was and then Amit will share a quick reflection on whether that prediction held true or not, and then we'll move into our 10 2025 AI predictions.
• 5:45
  Speaker 3: So first let's recap last year, or this past year, I should say.
• 5:50
  Speaker 3: Our first prediction was natively multimodal models go mainstream.
• 5:55
  Speaker 3: So AI models designed to natively handle multiple data types text, images, audio and video would become widely adopted.
• 6:03
  Speaker 1: Amit, we get it right I think we nailed that one.
• 6:07
  Speaker 1: I don't even think people think about models as being one modality or another.
• 6:11
  Speaker 1: You just you know.
• 6:11
  Speaker 1: You go to cloud, chat, gpt now gemini with their newest model and you know you just provide images, you interact with text, you generate images, audio and video.
• 6:20
  Speaker 1: Not quite there yet for some of the models, but I think it's becoming really a commodity to be able to interact across all these modalities.
• 6:28
  Speaker 1: You're expecting that from any consumer type product.
• 6:32
  Speaker 1: You need to be able to speak to the AI as well as type and see text responses, and very soon I think you'll see audio and video be much more common across all sorts of third party apps.
• 6:45
  Speaker 3: All right, that's one counted yes for now.
• 6:47
  Speaker 3: Our next prediction for 2024 was that open-source AI models from companies like Mistral, Meta and or Microsoft would approach or surpass GPT-4's capabilities.
• 6:57
  Speaker 3: If you're an avid listener of the pod, you already know the answer to this one.
• 7:00
  Speaker 3: But, Amit, what's your take?
• 7:02
  Speaker 1: Yeah, we nailed that one.
• 7:03
  Speaker 1: That was kind of easy, though, because we already saw that trend line right around this time last year with what was happening with Lama and Mistral, and Microsoft has a model called PHI P-H-I, and they released PHI 4, which is fun to say in the last week or two, and that model as well is better than the GPT-4 that we had late last year, and that model is even smaller than the Lama 3.3 model we talked about recently.
• 7:26
  Speaker 1: So, yes, indeed, we nailed this one, and kudos to the community of open source practitioners who are driving this forward.
• 7:34
  Speaker 3: Our next prediction was that Microsoft 365 Copilot and Google's Duet AI would see widespread deployment transforming productivity tools across office environments.
• 7:44
  Speaker 3: Amit, what do you think on this one?
• 7:47
  Speaker 1: I would say we missed on that one.
• 7:48
  Speaker 1: I think Copilot has been somewhat of a disappointment this year in terms of adoption.
• 7:53
  Speaker 1: The use cases within Copilot throughout the year, in my opinion, have been underwhelming.
• 7:59
  Speaker 1: Compared to what you can get out of Claude, out of Gemini, out of OpenAI's ChatGPT Even though Copilot is powered by GPT-4, it has been underwhelming and I think it's by limitation designed into the first iteration of Copilot.
• 8:13
  Speaker 1: I'm really excited about Copilot Wave 2 that was announced and we spoke about recently, but I think that it's been fairly limited in use.
• 8:20
  Speaker 1: I know personally, I've tried many times and fairly recently to use Copilot and the only use case that I have personally gotten significant value from is being able to create a PowerPoint from a Word document.
• 8:32
  Speaker 1: But of course there's many tools that can do things like that.
• 8:34
  Speaker 1: So I'm optimistic about the future, but I think we predicted too rapid of an ascent of those inbuilt types of tools.
• 8:43
  Speaker 3: Our next prediction was the rise of the CDPs.
• 8:46
  Speaker 3: We predicted that common data platforms, or CDPs, would become essential for associations to consolidate data from disparate systems and fully leverage AI.
• 8:56
  Speaker 1: I'd say we got a B on that.
• 8:58
  Speaker 1: My thought process is we've definitely seen adoption of CDPs this year.
• 9:03
  Speaker 1: I still think, though, it's not at the level or scale where it's making a big impact broadly across the community.
• 9:10
  Speaker 1: We've seen a lot of organizations that are at the frontier of adoption of AI in the association market fully embrace the idea of a common data platform or an AI data platform, but for the most part, associations still are not educated in what these things are, why they're critical.
• 9:25
  Speaker 1: So I think the future is still very bright for this.
• 9:28
  Speaker 1: I think it's a critical technology to solve problems, but we were perhaps a bit optimistic on the uptake for 2024 itself.
• 9:37
  Speaker 3: We're an optimistic podcast, so that makes sense.
• 9:39
  Speaker 3: The next prediction we had was consumer expectation for AI-enabled brands.
• 9:44
  Speaker 3: We said consumers would expect AI-driven features like 24-7 chat support, and brands not adopting AI might be seen as outdated.
• 9:53
  Speaker 3: What do you think on this one?
• 9:54
  Speaker 1: I might be an outlier, mal, or you tell me what you think, but I already expect this out of every brand I deal with.
• 9:59
  Speaker 1: I get fed up really quickly now if I have more of, know, more of a classical robocall type experience Not that I actually think most brands are going to have it, but the ones that I am interacting with more regularly.
• 10:10
  Speaker 1: When they have it, I'm pleased, whereas I'm not pleased when I don't have that experience.
• 10:14
  Speaker 1: How about you?
• 10:16
  Speaker 3: I expect it, I guess, but I'm not surprised when I don't see it.
• 10:20
  Speaker 3: I do feel like most brands that I'm interacting with on a regular basis are not leveraging AI fully and completely.
• 10:25
  Speaker 3: I did have a relatively positive experience with State Farm filing an insurance claim, like only through their chat assistant, which I was really impressed by recently, but other than that, I feel like we're not seeing this as much as I would like.
• 10:39
  Speaker 3: Our next prediction was AI misuse in information integrity and cybersecurity.
• 10:46
  Speaker 3: 2024 being an election year, we predicted we would see increased risks from AI-driven misinformation, like deepfakes and sophisticated phishing attacks.
• 10:55
  Speaker 3: What do you think?
• 10:57
  Speaker 1: Sadly, we got this one completely right, and we've seen a lot of evidence of this, both in the public space, with content being used to misinform, but also the phishing and cyber attacks have increased dramatically and their sophistication has leveled up considerably, so that has absolutely been as big of an issue, perhaps bigger than what we had projected this time last year.
• 11:21
  Speaker 3: Next prediction dramatic reduction in AI costs, fueling broader adoptions.
• 11:25
  Speaker 3: Falling AI costs would democratize AI access, allowing more organizations to adopt AI solutions.
• 11:31
  Speaker 3: I think I know your answer here.
• 11:33
  Speaker 1: What do you think it is?
• 11:34
  Speaker 3: I think you're going to say, yes, we got that one right.
• 11:36
  Speaker 3: You know better than me like the math side of it, but we've seen the price drop a lot.
• 11:42
  Speaker 3: I will say Indeed.
• 13:33
  Speaker 3: Moving on.
• 13:35
  Speaker 3: Ooh, I'm curious to get your take on this one.
• 13:37
  Speaker 3: An increase in AI interpretability would lead to safer use and broader adoption.
• 13:42
  Speaker 3: What do you think on that one?
• 13:45
  Speaker 1: I think, at the frontier of science, where the leading AI labs that are deeply involved in this notably Anthropic I think is probably the leading lab in this area have made notable improvements in their ability to understand what's actually happening inside these models.
• 14:00
  Speaker 1: Right, they're still somewhat of a black box.
• 14:02
  Speaker 1: They produce amazing outputs, but how they actually work we're not exactly sure, which is both fascinating and a little bit scary.
• 14:09
  Speaker 1: So for those who are a little bit AI hesitant, this area, I think, will lead to a lot more comfort and a lot more breadth of adoption.
• 14:20
  Speaker 1: I would say this year we probably haven't made as much progress in the field as I would have liked to have seen, partly because I think the people who are at the frontier of AI that are pushing forward are so focused essentially by necessity, because it's such a competitive landscape that this is taking a backseat.
• 14:37
  Speaker 1: People can give it as much lip service as they want, the reality is the resources invested in interpretability all the top labs in academia are a tiny number of folks compared to those that are working on actually pushing the boundaries of what these systems can do.
• 14:51
  Speaker 1: I still have optimism that this will be a critical and achieved area of AI capability over time, but this year we did not see as much growth there as I would have hoped.
• 15:04
  Speaker 3: Next up.
• 15:04
  Speaker 3: We predicted that there would be many specialized AI models that would drive breakthroughs in fields like material science and renewable energy.
• 15:11
  Speaker 1: The short answer is absolutely.
• 15:13
  Speaker 1: We've seen an abundance of evidence that there has been incredible advance in the world of scientific discovery aided by AI models.
• 15:21
  Speaker 1: Perhaps the most publicly visible evidence of this is the Nobel Prize being awarded to Demis Hassabis, one of the founders of DeepMind, for the work on AlphaFold3, which has been truly groundbreaking in the world of biology and the world of drug discovery.
• 15:36
  Speaker 1: So very excited about the future across all of these categories.
• 15:40
  Speaker 1: Right now, these models, to be clear, are assisting people who are making discoveries, so the AI models are not coming up with hypotheses to then test and iterate.
• 15:49
  Speaker 1: The AI models are helping with predictions, they're helping with understanding, and so the idea that the models will perhaps one day be able to also formulate their own hypotheses and then go on to run experiments, validate or invalid the hypotheses, etc.
• 16:02
  Speaker 1: That holds tremendous promise, and a lot of the leading labs are working on exactly this problem, so I'm super pumped about it.
• 16:09
  Speaker 1: I think that's definitely a checkmark for us in 2024 as well.
• 16:15
  Speaker 3: The last prediction we had was that generalized AI models would exhibit enhanced reasoning and math capabilities, aiding complex problem solving.
• 16:24
  Speaker 1: Yes, I think this has been a big area of progress.
• 16:29
  Speaker 3: Well, awesome, with that we're moving on to next year, to 2025.
• 16:34
  Speaker 3: Our first prediction might not surprise you.
• 16:36
  Speaker 3: We're predicting that AI audio will match human latency and conversational response times.
• 16:42
  Speaker 3: So, essentially, ai-driven voice assistants and chatbots will achieve response times indistinguishable from human conversation, reducing delays to under 200 milliseconds, and we predict this improvement will make interactions with AI and customer service, virtual assistants and real-time translation feel seamless and natural.
• 17:01
  Speaker 3: So, amit, what do you see as kind of those immediate, near-term outcomes of AI audio matching human latency?
• 17:10
  Speaker 1: My first hope for this would be that Siri doesn't suck.
• 17:13
  Speaker 1: That would be helpful.
• 17:15
  Speaker 3: That would be nice I just used Siri this morning.
• 17:17
  Speaker 1: It was like are we still using this thing?
• 17:19
  Speaker 1: I mean, I use it because it's built into my iPhone and I'm trying to like text people while I'm driving.
• 17:23
  Speaker 1: Of course, I'm not actually using my phone, but I'm talking to Siri, so think about that though.
• 17:27
  Speaker 1: So Siri is an extremely broadly adopted AI type tool.
• 17:32
  Speaker 1: We're using ancient technology, so if we had basically low-latency conversational ability with Siri, we can have natural conversations with that kind of AI, perform a bunch of tasks, because, as limited as Siri is from a language comprehension and language generation capability, siri's actually more agentic than many contemporary models, if you think about it, because Siri can send text messages.
• 17:54
  Speaker 1: Siri can do a lot of things on device with your phone interact with apps.
• 17:58
  Speaker 1: Most of these other AIs cannot do that.
• 18:00
  Speaker 1: So if you imagine that environment, that's pretty cool, but that's a little bit of a separate path from audio latency more generally, as you put it, and I'm excited about this prediction because, first of all, as cost comes down and as speed goes up partly driven by innovation all up and down the stack in terms of algorithmic improvement, also improved by hardware optimizations, also improved by scale you see improvements there, but you also are going to see a lot of this, driven by free, open source models, which already are abundant out there.
• 18:31
  Speaker 1: 200 milliseconds, or two-tenths of a second, is what we expect in conversation with a human.
• 18:37
  Speaker 1: Now, you can have a little bit longer pauses when someone's really thinking about something, but if you have to wait more than half a second, three-quarters of a second, it starts to feel unnatural.
• 18:46
  Speaker 1: So really fast response time is key.
• 18:49
  Speaker 1: If you think about what's happening in that 200 milliseconds.
• 18:52
  Speaker 1: You have to have an AI interpret what you said.
• 18:55
  Speaker 1: It has to be then processed by a model, which essentially means understanding the request, whatever it is generating the response, and then audio has to be synthesized.
• 19:04
  Speaker 1: Whether that's done as part of a native multimodal model, as we've talked about, or if it's done with audio to text, text to text in terms of the response, and then text backs to audio, you have a lot going on.
• 19:17
  Speaker 1: So being able to do that in a very short amount of time, and then obviously, all the network traffic to get that stuff going on, is an interesting thing, which, by the way, is one of the reasons I'm excited about on-device AI inference.
• 19:29
  Speaker 1: So when we talk about models getting smaller, it not only means they're cheaper and more available.
• 19:33
  Speaker 1: It also means that some of these models can run on device, so they can run on your phone, they can run on your laptop, and so therefore, when we have like, let's think of it as a coprocessor for AI locally that can do some of the lifting, that not only distributes the load, which is good in a number of ways, cost being one, but it also is much faster because there's no network latency.
• 19:55
  Speaker 1: So I think this is a key area of almost guaranteed improvement, because we're already seeing a lot of these applications.
• 20:02
  Speaker 1: Some of the stuff we're working on, actually across Blue Cypress in our labs group is exactly in this area, because we think audio is the primary user interface for the future for most associations and their members.
• 20:14
  Speaker 1: So we have a lot of projects in the work in this area and they all presuppose what I just said that the cost is extremely low and the speed is really fast.
• 20:21
  Speaker 1: So I'm very confident this is going to happen.
• 20:25
  Speaker 3: Our next 2025 prediction a bit more on the negative side.
• 20:28
  Speaker 3: This one but a Fortune 500 company will fall victim to an advanced AI scam.
• 20:33
  Speaker 3: So we're predicting a major corporation will be targeted by a sophisticated AI-driven scam using deepfakes, voice cloning or AI-generated phishing.
• 20:41
  Speaker 3: The scam will lead to financial loss or data breaches data breaches and we predict an incident like this will serve as a wake-up call for businesses to invest in advanced cybersecurity measures, ai detection systems and employee training to recognize AI-based threats.
• 20:57
  Speaker 3: So, amit, this is pretty timely for us.
• 20:59
  Speaker 3: We actually just had a conversation in our AI mastermind group that Sidecar hosts about AI and cybersecurity Right now.
• 21:06
  Speaker 3: What must associations do immediately to start mitigating these threats?
• 21:11
  Speaker 1: Pay attention.
• 21:12
  Speaker 1: That's the first thing is you have to pay attention, you have to get trained on it, you have to increase awareness.
• 21:17
  Speaker 1: There's a number of cybersecurity tools and trainings that you can invest in for your team.
• 21:21
  Speaker 1: This is something you have to invest in every single team member, as well as close-in volunteers.
• 21:27
  Speaker 1: What I mean by the latter is people have access to your systems from the volunteer side who potentially can have social influence.
• 21:33
  Speaker 1: So if you have certain board members that potentially could call someone up or could be faked to call someone up and say, hey, I need you to transfer this money or I need to do this other thing, that is a major attack vector opportunity for the bad actors.
• 21:48
  Speaker 1: So, first thing is awareness and pay attention.
• 21:51
  Speaker 1: Invest in some training for your team.
• 21:53
  Speaker 1: This isn't even AI training.
• 21:54
  Speaker 1: This is like AI aware cybersecurity training.
• 21:58
  Speaker 1: At Blue Cypress, we use a company called Ninjio, which we endorse as a great resource.
• 22:03
  Speaker 1: We've got no affiliation with them, but we just like their training.
• 22:06
  Speaker 1: All of our employees get a weekly cybersecurity training nugget from them, which is just like their training.
• 22:09
  Speaker 1: All of our employees get a weekly cybersecurity training nugget from them which is just easy to consume I think three-minute video or so every single week, and that's the important thing is just this constant reminder and a little bit of education to be slightly more aware.
• 22:21
  Speaker 1: Unfortunately, the world of cybersecurity is like the world of physical security in that the weaker boundaries tend to get tested.
• 22:28
  Speaker 1: Everybody's boundaries are getting tested all the time, but the weaker organizations tend to get exploited more easily, and so it's kind of like if you live on a block and you have an alarm system, but your neighbor doesn't, you're safer than the next guy.
• 22:41
  Speaker 1: Not that you want someone else to fall victim, obviously, but if you are the weaker opportunity, then you're probably going to get attacked.
• 22:50
  Speaker 1: Unfortunately, this year, in 2024, we know directly of several associations who have fallen victim to advanced AI-driven scams.
• 22:59
  Speaker 1: It's going to happen even at larger scale in 2025.
• 23:02
  Speaker 1: I'm hopeful that our little part of increasing awareness here through the pod will have an impact, but I almost think this is not even a prediction.
• 23:10
  Speaker 1: It's just like a wake-up call that people just have to pay attention to this.
• 23:14
  Speaker 3: Our next prediction for 2025 is that AI video models will be capable of generating high-quality videos over 10 minutes long from a single prompt, complete with AI-generated audio narration and dynamic visuals.
• 23:27
  Speaker 3: This evolution will combine creative storytelling abilities of tools like Sora with the practical avatars of Haijin, making it easier to create educational content, marketing videos and entertainment.
• 23:39
  Speaker 3: We predict this will democratize video production, making professional-quality videos accessible to all associations and content creators.
• 23:47
  Speaker 3: So, amit, I think you and I had a slight differing view on this.
• 23:50
  Speaker 3: I think I had a few minutes long and then you had over 10 minutes long.
• 23:54
  Speaker 3: For how long these videos might be, it seems like a leap from right now, where we're seeing the ability to make clips between five seconds and 20 seconds long.
• 24:02
  Speaker 3: How confident are you on this one?
• 24:06
  Speaker 1: I would say I have a pretty high level of confidence in it, partly because Google's VO2 model has two minute videos already, because the competition they have with Sora and some other tools that are out there that are in that 20 second length, google's like, well, we're going to crush you guys with the VO model to version two.
• 24:23
  Speaker 1: So that's a two minute video.
• 24:24
  Speaker 1: I think between competition and improvements and performance in the compute stack as well as the models themselves, you're going to see this happen.
• 24:32
  Speaker 1: I'm pretty confident of it.
• 24:34
  Speaker 1: And whether or not we're right about 10 minutes or if it's eight minutes or something like materially longer, that enables use cases that are non-trivial right.
• 24:42
  Speaker 1: So, right now they're actually non-trivial use cases today.
• 24:45
  Speaker 1: They just have more to do with like really short nuggets of content, more perhaps for advertising, creative stuff like that.
• 24:52
  Speaker 1: But you're not going to stitch together a bunch of Sora clips to make an e-learning course, partly because, also, the modality is purely on one side of the wall or the other.
• 25:00
  Speaker 1: It's either kind of the more cinematic type thing or if it's more like the hey Jen avatar world, where it's more like a person talking.
• 25:07
  Speaker 1: I actually think the latter case is where there's more practical use cases today for AI-generated video, at least for associations right straight down the center of the use case line.
• 25:17
  Speaker 1: But I think you're going to see a synthesis of these modalities come together.
• 25:20
  Speaker 1: So whether it's five minutes, 10 minutes, 15 minutes, is part of the question.
• 25:24
  Speaker 1: But also, can you have a tool that makes like a studio, that makes it possible for you to combine these different things together so that you can say, hey, I want to create a video that's an e-learning lesson about cybersecurity for associations in the age of AI?
• 25:39
  Speaker 1: I want to cover these topics.
• 25:40
  Speaker 1: I want you to give me some dramatizations of what happens to an organization when their cybersecurity is penetrated by a bad actor, and so then the AI comes up with multiple scenes, including the ability for Sora-style videos, kind of showing the bad actor perpetrating the attack, and then maybe there's an avatar talking about the problem, what went wrong, how to solve it, that kind of a thing right, a very highly professional produced lesson.
• 26:07
  Speaker 1: You can visualize that based on what I just said, and so can the AI.
• 26:11
  Speaker 1: So I think you're going to see solutions like that come to market.
• 26:14
  Speaker 1: It may not be actually a model advance.
• 26:16
  Speaker 1: It might be an application that stitches together multiple pieces and parts to create an output, because you can do everything I just said right now with multiple different tools and doing a lot of work to stitch them together.
• 26:27
  Speaker 1: I think that's actually in general.
• 26:33
  Speaker 1: Where you'll see a lot of advancement in 2025 is people building apps that pull this stuff together.
• 26:35
  Speaker 1: I think this is a category that you'll see it in because the value opportunity is extremely high.
• 26:39
  Speaker 3: I can imagine someone listening to this right now and saying well, I'll just wait.
• 26:43
  Speaker 3: I'll wait until we have this a few years down the line.
• 26:45
  Speaker 3: I don't really want to play around with these tools now and stitch them together.
• 26:48
  Speaker 3: What would you say to that?
• 26:51
  Speaker 1: I would say that that is, I mean, that's a pragmatic point of view in some categories.
• 26:55
  Speaker 1: I wouldn't agree that that's a good idea.
• 26:57
  Speaker 1: In general, with AI, I think you should definitely take advantage of what's a practical, immediate use case today.
• 27:02
  Speaker 1: But I think for some of these perhaps more complex outputs where it does take a lot of extra work to produce them Like the way we did the Thanksgiving edition of this podcast you know that took a little bit of work.
• 27:14
  Speaker 1: So waiting to do stuff like that for another year, two years, sure, but just don't let that be an excuse to wait on everything.
• 27:25
  Speaker 3: Our next prediction is that AI models will continue to improve their ability to reason and solve complex problems by following logical, step-by-step processes similar to human thinking.
• 27:35
  Speaker 3: This advancement will make AI more useful in fields requiring analytical skills, like strategic planning, legal analysis and scientific research.
• 27:43
  Speaker 3: So we've seen this a little bit in 2024 with OpenAI's O1 model, and I want to be clear here, though, amit, that when we're talking about human-like reasoning, we're not talking about artificial general intelligence next year, right.
• 27:59
  Speaker 1: I mean, it depends on how you define AGI.
• 28:01
  Speaker 1: Agi is a moving target because when people have defined it initially, they said, hey, it's PhD-level intelligence in all fields.
• 28:05
  Speaker 1: Initially they said, hey, it's PhD-level intelligence in all fields.
• 28:07
  Speaker 1: So if you have PhD-level intelligence in all fields, o1 can arguably do that today, so is that AGI?
• 28:16
  Speaker 1: The problem is is that AGI, I think, in the more generally accepted definition, or whether it's AGI or ASI for artificial superintelligence I don't really care about the terminology but models that are able to broadly reason, solve complex problems and then also do novel thinking?
• 28:33
  Speaker 1: So currently, models themselves, including O1, cannot come up with new ideas.
• 28:39
  Speaker 1: They're essentially solving for things that are within their scope of knowledge, and so that, to me, is a key thing.
• 28:47
  Speaker 1: It's not just exhibiting knowledge and applying it in the context of what they already know.
• 28:51
  Speaker 1: To me, true AGI means being able to create something fundamentally new that doesn't exist, like, for example, discovering new physics.
• 28:57
  Speaker 1: That would be something that would, in my mind, be characterized as AGI ASI.
• 29:02
  Speaker 1: That being said, I don't think it matters.
• 29:05
  Speaker 1: I think the ability to leverage human-like reasoning capabilities will improve the applications we can build, so we'll be able to do a lot more in the way of decision-making, for example, being able to have models take on decision-making in more complex, higher-stakes scenarios.
• 29:21
  Speaker 1: So that to me, is almost a guaranteed thing, because we already see that in the frontier.
• 29:28
  Speaker 1: Now what's happening is there's know, there's this big innovation around Strawberry, now O1, and everyone was excited about it, and I think there's reason to be excited about it.
• 29:35
  Speaker 1: At the same time, what really they did is they just threw more what they call test time, compute, more ability for the model within it to basically iterate and to ask questions of itself and to kind of deliberate, using slow thinking, right, thinking fast and slow.
• 29:50
  Speaker 1: We've talked about that before when we talked about O1, strawberry, and that's exactly what the model's doing.
• 29:54
  Speaker 1: Well, guess what?
• 29:55
  Speaker 1: Agents have been doing that for a lot longer.
• 29:58
  Speaker 1: So agents that have a lot more compute available can say hey, I'm going to take an answer from a model, I'm going to determine whether or not that answer is good.
• 30:06
  Speaker 1: I'm going to come up with a plan.
• 30:08
  Speaker 1: I'm going to iterate through that plan.
• 30:09
  Speaker 1: I might even use multiple models to come up with different possible answers, compare them against each other, pick the best one.
• 30:15
  Speaker 1: That's essentially what we're talking about, and there's different algorithmic approaches to this, whether it's tree of thought or other types of techniques, but these can be encoded either into the models themselves or into systems that leverage models.
• 30:27
  Speaker 1: In fact, people are working on that right now across a number of different companies.
• 30:36
  Speaker 1: So I think that reasoning capability whether it's from the model or from the agent or from the application layer, it doesn't really matter.
• 30:39
  Speaker 1: Ultimately, particularly in this sector, it's about what capabilities does this unlock?
• 30:43
  Speaker 3: Our next prediction for 2025 is that AI will fuel, or continue to fuel, breakthroughs in scientific research.
• 30:50
  Speaker 3: Specialized AI models will accelerate discoveries in fields like material science, renewable energy and pharmaceuticals.
• 30:56
  Speaker 3: Tools like DeepMind's GNOME, which identified millions of new crystal structures, hint at the potential for AI to unlock knowledge that would take decades through to traditional research.
• 31:06
  Speaker 3: Amit, is there an area of science that you are particularly excited about in terms of AI innovations in 2025?
• 31:14
  Speaker 1: I mean, I think I'm excited about all this stuff.
• 31:16
  Speaker 1: I spend a lot of my time listening to people who are deep in these fields talking about what they're doing.
• 31:23
  Speaker 1: I just find it fascinating.
• 31:24
  Speaker 1: I don't have any scientific training in biology or physics or chemistry personally, but I like to learn as much as I can and hear what these people are talking about, and it's fascinating.
• 31:34
  Speaker 1: I think you're going to see significant improvements in 2025 in these areas.
• 31:39
  Speaker 1: My specific prediction within this realm is that we will see a drug that was discovered by AI, probably through isomorphic labs such as the DeepMind spinoff, using the AlphaFold3 and other related models and proprietary processes.
• 31:56
  Speaker 1: They built on top that a drug from that process will enter phase three clinical trials and potentially have significant impact on the world in some disease area that right now seems unsolvable.
• 32:10
  Speaker 1: So I'm super excited about that.
• 32:12
  Speaker 1: I know they have stuff in their pipeline in earlier phase trials, so I think that it's reasonable to say that they might be going into a phase three trial, which is a broader, wider trial, and that is super, super exciting.
• 32:23
  Speaker 3: Next up.
• 32:24
  Speaker 3: We are predicting that AI education and upskilling will take center stage in both for-profit and nonprofit sectors next year.
• 32:32
  Speaker 3: Companies and associations alike will invest in AI training programs to address the growing skills gap and ensure their workforce can effectively use AI tools.
• 32:41
  Speaker 3: This trend will lead to more AI literacy programs and certification courses tailored to different industries and professional needs.
• 32:48
  Speaker 3: Amit, I've been really wanting to ask you about this phrase I think you coined on the podcast, so I'm really excited to ask you what is leadership malpractice and how does it relate to this prediction?
• 33:01
  Speaker 1: Well, it's my belief that the number one role and responsibility of a leader is to grow their people Full stop.
• 33:08
  Speaker 1: You, as a leader, should be judged not by the work and the output while you're at a place, but what happens after you leave, which is only predictable based upon the work that you do to help other people grow.
• 33:20
  Speaker 1: So, to me, that's the number one thing leaders have to do, and that's leadership at every level.
• 33:25
  Speaker 1: It doesn't matter if you have formal direct reports or if you're the CEO of an association with a thousand employees.
• 33:31
  Speaker 1: You have a responsibility to grow your people.
• 33:34
  Speaker 1: And where we are today in the world forget about associations where we are today in the world is that AI is affecting everything and you ain't seen nothing yet.
• 33:42
  Speaker 1: So if you are not preparing your people for the world of AI, it is my firm belief that you are committing an act of malpractice, because you are entrusted with the futures in many respects of your team.
• 33:56
  Speaker 1: They're looking to you as the leader to help guide them into the future, not only as an organization, but for them as people in their careers.
• 34:03
  Speaker 1: So I think it's your responsibility to help them find their way in the age of AI, as challenging and as scary and as difficult that is in terms of for all of us.
• 34:12
  Speaker 1: Right, we have no idea what's happening next at the end of the day, certainly beyond two years.
• 34:17
  Speaker 1: So what do you have to do?
• 34:18
  Speaker 1: You have to get people started on their journey.
• 34:25
  Speaker 1: You have to encourage them and, frankly, in some cases force them.
• 34:26
  Speaker 1: You have to do that because otherwise, you are guaranteed to be walking your people into a meat grinder.
• 34:31
  Speaker 1: They are going to be unemployable in a very short period of time and it will be your fault at least largely your fault if you are not helping them learn.
• 34:40
  Speaker 1: So I know those are very strong statements.
• 34:42
  Speaker 1: That's intentional.
• 34:43
  Speaker 1: I am urging every leader in this market to learn themselves, to learn and to push their teams to learn, and we see a lot of people seeing success with this right.
• 34:54
  Speaker 1: Sometimes people on a staff are concerned.
• 34:58
  Speaker 1: They're scared about what the future looks like with AI, and the best thing a leader can do is say listen, we share those concerns.
• 35:04
  Speaker 1: We don't have answers, but we do know that if we don't learn this stuff, we are definitely going to be obsolete.
• 35:10
  Speaker 1: So you know the future is going to be one where AI plays a major role, and if you're not helping your team prepare for that.
• 35:17
  Speaker 1: That is why I use such a strong term as saying you're committing an active leadership malpractice, so that is why that's my logic behind that.
• 35:24
  Speaker 1: It also tends to catch people's attention, so my belief is that in all sectors, there will be a heavier emphasis on AI education.
• 35:33
  Speaker 1: Associations, I think, have to do two things.
• 35:35
  Speaker 1: Number one they have to train their team.
• 35:37
  Speaker 1: We just talked about that, and part of what I mean by team is both staff and, particularly close in volunteer leadership, the board committees that are highly active.
• 35:45
  Speaker 1: Maybe you could extend that to all of volunteers.
• 35:47
  Speaker 1: Beyond that, though, I think the association both has an obligation and an opportunity with respect to AI learning in their field.
• 35:55
  Speaker 1: So if you're an association of accountants or lawyers or architects or engineers, most likely your members are wondering what to do in the future.
• 36:04
  Speaker 1: How does AI affect them?
• 36:06
  Speaker 1: And they look to you as their professional or industry association as perhaps a form of leadership in this area.
• 36:13
  Speaker 1: If they don't find it in you, they'll find it somewhere else, of course, so that's where the opportunity and the risk lies.
• 36:18
  Speaker 1: You should be out there training your members on the future of their sectors with respect to AI.
• 36:24
  Speaker 1: That's the opportunity that's there and I think you're going to see more and more associations do that.
• 36:28
  Speaker 1: Just as an aside, we at Sidecar do, in fact, help associations with both of those things.
• 36:32
  Speaker 1: We've talked, obviously, about our learning hub for association people on this podcast a bunch of times.
• 36:37
  Speaker 1: What many people here probably don't know is that we also partner with associations providing what we call AI learning for members, where we take the AI learning hub content and we tailor it for your sector and then you are able to provide that as your AI learning hub for your associations members with your branding, and we do that in a revenue partnership with associations.
• 36:57
  Speaker 1: So just wanted to mention that, because if you think this idea makes sense but you're not sure where to start now, you do.
• 37:03
  Speaker 3: You heard it here.
• 37:04
  Speaker 3: Don't commit leadership malpractice.
• 37:06
  Speaker 3: Invest in AI education for your organization, for your staff, for your members, even if it's not with Sidecar.
• 37:11
  Speaker 3: Our next prediction for 2025 is that AI agents will continue to gain autonomy and specialization.
• 37:17
  Speaker 3: Is that AI agents will continue to gain autonomy and specialization.
• 37:18
  Speaker 3: Ai agents will become widely adopted, capable of autonomously handling complex, multi-step tasks with minimal oversight.
• 37:25
  Speaker 3: Agents will specialize in domains like project management, marketing automation, customer support and data analysis.
• 37:32
  Speaker 3: Again, this is no surprise if you've listened to any of our podcast episodes before, but we've discussed knowledge agents on the pod data agents, customer service agents for associations, amit, where do you expect to see the biggest focus of dollars and impact when it comes to AI agents within associations next year?
• 37:54
  Speaker 1: I think the applications where you'll see the most of that is where you have a benefit internal to the association in terms of efficiency and a value creation opportunity for the member in parallel, right?
• 38:03
  Speaker 1: So to me, the biggest opportunity is around member service.
• 38:06
  Speaker 1: There are various flavors of this, but when you think about what we talked about earlier and the expectation that brands interact with their customers, their members, in an AI forward way, people are going to increasingly be frustrated with brands that are not basically instantaneous in responding to them with high quality answers.
• 38:26
  Speaker 1: So you can't do that without AI, right?
• 38:28
  Speaker 1: If the expectation is that I can immediately solve my problem, whether it's a knowledge inquiry like I have a deep domain question I'm a doctor and I have some deep domain question or perhaps I just want to cancel a registration for an upcoming webinar because I can't go anymore, I want all that handled.
• 38:45
  Speaker 1: I want to handle it immediately, professionally.
• 38:47
  Speaker 1: The only way to handle that is through AI, obviously.
• 38:49
  Speaker 1: So I think you're going to see those kinds of applications.
• 38:52
  Speaker 1: So customer service, member service, to me is the number one category that I think we'll see adoption this year, because there's immediate value to the association, because there's more efficiency there.
• 39:03
  Speaker 1: Member services teams, event services teams they're overworked, people are having a hard time keeping up and typically you hear things like oh, we have a service level agreement, meaning we want to be able to adhere to this level of responsiveness and our goal is a 24-hour response time, which, pre-ai, sounds like a pretty good thing, right?
• 39:22
  Speaker 1: If I email a member services team like info at associationorg, that's not too bad, except 24 hours is a really long time in an age of AI.
• 39:31
  Speaker 1: So what if we could respond within 24 seconds?
• 39:33
  Speaker 1: And that's what we can do with AI agents, right, and we can probably produce objectively a better response every single time because there's more knowledge available than any individual member services rep could be reasonably expected to know.
• 39:45
  Speaker 1: And, of course, the speed of that response is also a value creation element by itself.
• 39:50
  Speaker 1: So to me, that's the area that you're going to see.
• 39:52
  Speaker 1: A lot of growth is stuff around member services stuff around member services.
• 40:01
  Speaker 3: Next prediction is the rise of domain-specific AI tools.
• 40:03
  Speaker 3: Industry specific AI models will rise in prominence, offering tailored solutions for fields like healthcare, finance, legal services and, of course, associations.
• 40:09
  Speaker 3: We predict that these tools will provide more accurate and context-aware outputs than generalized AI models, enabling specialized applications like legal document analysis, financial forecasting or even personalized health treatment plans.
• 40:26
  Speaker 3: You kind of touched on this a bit earlier, amit, but do you see it as a responsibility of an association a particular domain legal healthcare, manufacturing to kind of keep tabs on these industry-specific AIs and decide to endorse or maybe not endorse them?
• 40:45
  Speaker 1: I think the keeping tabs part 100% is a responsibility of the association, because your job is to know what's going on in that field and if you don't know about the AI capabilities of this level of importance in your field, you're not really serving your sector, because you're teaching them how to do things in a world where these tools don't yet exist.
• 40:57
  Speaker 1: And if the tools do exist and they're having an impact, it's a very important thing for the association to be aware of it and to inform their membership about it.
• 41:05
  Speaker 1: As far as whether to endorse them or not, that's going to be a decision that depends on the association whether that's part of their business model, whether that's appropriate, considered a conflict of interest, or not.
• 41:15
  Speaker 1: Some associations do a great job in doing this in a fairly objective way and generate a lot of revenue from that, and others stay away from it entirely.
• 41:22
  Speaker 1: I think it really depends.
• 41:23
  Speaker 1: If you go look at a lot of scientific associations that are academic in nature, they would very rarely, I think, venture into this space because they want to be seen as a totally neutral, objective ground for their journals and for their conferences.
• 41:36
  Speaker 1: But you see other associations that are more trade-oriented or profession-oriented that tend to be a little more open to this type of thing.
• 41:43
  Speaker 1: So I think the last point about whether it's a revenue stream, there's endorsements or not, is a different question for each association, but keeping up on top of this stuff, I think, is important for every association.
• 41:52
  Speaker 1: I'm particularly excited about democratization of services that up until now have only been available for really a very narrow slice of the population.
• 42:01
  Speaker 1: Think about, like the best healthcare in the world.
• 42:03
  Speaker 1: Think about the best legal services in the world.
• 42:06
  Speaker 1: You basically have to be rich to access that stuff right now.
• 42:09
  Speaker 1: And not only do you have to be rich, but you have to know people who can give you advice on how to get to the right services.
• 42:14
  Speaker 1: If you can democratize access to the best that we as a species have to offer in health care, in financial advice, in legal services that's amazing.
• 42:24
  Speaker 1: So I'm excited about that, just in terms of uplifting humankind, and I think the association plays a critical role there.
• 42:32
  Speaker 3: Our ninth prediction for today's episode is a shift from SEO to AEO or AI engine optimization.
• 42:36
  Speaker 3: Is a shift from SEO to AEO or AI engine optimization.
• 42:38
  Speaker 3: As AI driven search engines and chatbots like ChatGPT, gemini and Perplexity, just to name a few, reshape how people find information, we are predicting that traditional search engine optimization, or SEO, will give way to AI engine optimization.
• 42:53
  Speaker 3: Organizations will need to optimize content for AI generated answers and recommendations to stay visible and relevant.
• 43:01
  Speaker 3: Amit, if you were an association CEO right now, would you stop investing in deep SEO work?
• 43:09
  Speaker 3: Or do you not think that we're there yet?
• 43:12
  Speaker 1: I would not stop investing in SEO.
• 43:13
  Speaker 1: First of all, a massive amount of traffic still comes from traditional search, and that's important to most associations.
• 43:19
  Speaker 1: First of all, a massive amount of traffic still comes from traditional search, and that's important to most associations.
• 43:24
  Speaker 1: I would be aware of this whole idea of AI optimization or AEO.
• 43:26
  Speaker 1: I think it's an important concept.
• 43:27
  Speaker 1: Ultimately, I think this stuff is just basically traffic generation, right?
• 43:32
  Speaker 1: So whether it's through quote unquote AI tools or traditional search tools, it doesn't really matter.
• 43:37
  Speaker 1: It's how do you make sure that your content is found?
• 43:39
  Speaker 1: And so that question will always be relevant, because you have to be able to get people to your content to drive the rest of your business functionality.
• 43:47
  Speaker 1: So I think it's going to be very important.
• 43:49
  Speaker 1: I think you're going to see people start to pay much more attention to what's happening inside ChatGPT, inside other tools.
• 43:56
  Speaker 1: I mean.
• 43:56
  Speaker 1: Clearly, everyone's going after everything.
• 44:00
  Speaker 1: Google wants to be the AI company of choice because they want to retain the traffic.
• 44:02
  Speaker 1: So does Meta.
• 44:03
  Speaker 1: Meta has more visitors on their chatbot, which is the Meta AI site, than actually anybody else, because they have their entire, Because they include in that people who use it in app.
• 44:12
  Speaker 1: So they're definitely someone to pay attention to.
• 44:15
  Speaker 1: They didn't build the Lama 3 and Lama 4 coming models.
• 44:18
  Speaker 1: They didn't build that out of the goodness of their hearts.
• 44:20
  Speaker 1: They built that stuff because they want to dominate AI, to make sure they retain traffic in their social tools and generate a lot of revenue, a lot of profit.
• 44:28
  Speaker 1: So that is 100% one that you're going to be seeing more and more.
• 44:33
  Speaker 1: So bottom line is yeah, I mean AI and AI tools are going to have more traffic.
• 44:38
  Speaker 1: Therefore, there's going to be a need to find a way to be discovered in those kinds of tools.
• 44:43
  Speaker 3: Our last prediction for 2025, it's a fun one.
• 44:46
  Speaker 3: What did we call this, amit, at the top of the episode?
• 44:48
  Speaker 3: The rise of the robots or the robot takeover?
• 44:51
  Speaker 3: Not quite.
• 44:52
  Speaker 3: We're predicting humanoid robots will gain traction in industries like retail, hospitality and healthcare, performing tasks like customer service, logistics and even caregiving.
• 45:02
  Speaker 3: Advances in robotics and AI will allow these robots to work alongside humans more effectively.
• 45:07
  Speaker 3: While they won't be ubiquitous yet, their presence will increase, driven by labor shortages and efficiency demands.
• 45:14
  Speaker 3: I mean, some people might say robots are just another way for AI to take human jobs, but I know you often have kind of a positive spin on humanoid robots.
• 45:22
  Speaker 3: So what do you think of that?
• 45:25
  Speaker 1: So I mean, I think there's, like everything else, there's two sides to this.
• 45:27
  Speaker 1: First of all, like on the prediction itself, I think you're going to see these types of robots gaining traction for sure, definitely things behind the scenes, right in warehouses and environments, where tasks that require interaction with the real world are going to be aided by these kinds of tools.
• 45:45
  Speaker 1: As far as the perspective I have on it, I think it's both good and bad.
• 45:49
  Speaker 1: On the bad side, yes, potentially you have a radical and rapid displacement of a number of sectors of the labor economy, which is challenging on a number of levels, but it's been happening for a long time, obviously, as well with other technological shifts.
• 46:05
  Speaker 1: I just think it's happening faster now than ever before.
• 46:07
  Speaker 1: So that's a potential concern, obviously.
• 46:08
  Speaker 1: But then the other side of it is that there's a lot of categories where labor is an extreme shortage.
• 46:13
  Speaker 1: Think about healthcare.
• 46:14
  Speaker 1: You mentioned that.
• 46:15
  Speaker 1: So caring for the elderly both very difficult to find, for example, in-home care for an elderly person that needs support, needs someone to needs really the support to be able to even safely use the restroom, something like that and there's a lot of issues there.
• 46:32
  Speaker 1: It's both from a cost perspective and the availability of resources very, very difficult.
• 46:37
  Speaker 1: So could robots potentially make a big difference there?
• 46:40
  Speaker 1: I think so, and that's exciting.
• 46:41
  Speaker 1: That potentially opens the door to healthier, longer lifespans for people in that particular subcategory.
• 46:48
  Speaker 1: But, just more broadly, I think there's a lot of areas where we can get more done and create abundance.
• 46:54
  Speaker 1: So I'm optimistic about this.
• 46:56
  Speaker 1: I think you're going to see companies that are more, you know, kind of forward looking, starting to experiment with this stuff, even in a visible way for customers, and then you're going to see some companies that are a little bit more hesitant, that maybe do some small experiments behind the scenes but start to really think about, like, how they can level up and improve the quality of service to their customers as well.
• 47:17
  Speaker 1: So to me, I think that the technology behind this largely the solutions to this are certainly not complete, but if you think about the advancements just this year in a number of companies that have produced significant capabilities in these robots, I think that in 2025, seeing them deployed at scale at a number of industries is almost a given.
• 47:41
  Speaker 3: Well, that is it.
• 47:42
  Speaker 3: That's a wrap on our 2025 top 10 predictions and our last live episode of 2024.
• 47:49
  Speaker 3: And Meith and I want to thank you all for tuning in, whether that's audio only or you watch us on YouTube, we truly have a blast recording these episodes every week, talking about AI for associations, and we are so thrilled to be doing it all again next year, and our New Year's resolution is to make our content even better.
• 48:07
  Speaker 3: So, with that, everybody, happy holidays, and we hope you have a happy new year.
• 48:15
  Speaker 1: Thanks for tuning into Sidecar Sync this week.
• 48:17
  Speaker 1: Looking to dive deeper?
• 48:19
  Speaker 1: Download your free copy of our new book Ascend Unlocking the Power of AI for Associations at ascendbookorg.
• 48:27
  Speaker 1: It's packed with insights to power your association's journey with AI.
• 48:31
  Speaker 1: And remember, sidecar is here with more resources, from webinars to boot camps, to help you stay ahead in the association world.
• 48:38
  Speaker 1: We'll catch you in the next episode.
• 48:40
  Speaker 1: Until then, keep learning, keep growing and keep disrupting.