Latest / Sidecar Sync / 25: Llama 3, Y Combinator’s Winter 2024 Class, & AI and the Middle Class

Transcript

25: Llama 3, Y Combinator’s Winter 2024 Class, & AI and the Middle Class · Sidecar Sync

• 0:00
  Speaker 1: Spend time on this, learn about it, go deeper.
• 0:02
  Speaker 1: Spend a little bit of time every week and learn this stuff, because the disruptive effect is not a question mark.
• 0:08
  Speaker 1: It's really a when is it going to happen to your field?
• 0:11
  Speaker 1: Welcome to Sidecar Sync, your weekly dose of innovation.
• 0:16
  Speaker 1: 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.
• 0:25
  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.
• 0:33
  Speaker 1: No fluff, just facts and informed discussions.
• 0:36
  Speaker 1: I'm Amit Nagarajan, chairman of Blue Cypress, and I'm your host.
• 0:41
  Speaker 1: Greetings and welcome back to the Sidecar Sync, excited to have you here with us today.
• 0:48
  Speaker 1: We have an exciting episode with a lot of interesting news at the intersection of AI and associations, and in fact, this is episode number 25 of the Sidecar Sync.
• 0:58
  Speaker 1: Can you believe it, mallory?
• 0:59
  Speaker 1: We're already at 25.
• 1:00
  Speaker 2: I really can't believe that.
• 1:01
  Speaker 2: I feel like this was just a seed of an idea a few months ago and now we're a quarter of the way to 100.
• 1:07
  Speaker 2: So we've got some big goals ahead of us.
• 1:10
  Speaker 1: All sorts of great conversations, interesting news.
• 1:13
  Speaker 1: Lots has changed since the first episode of the Sidecar Sync and had a lot of fun with it, so looking forward to this episode.
• 1:19
  Speaker 1: Before we get going, let's take a moment to hear a quick word from our sponsor.
• 1:24
  Speaker 2: Today's sponsor is Sidecar's AI Learning Hub.
• 1:26
  Speaker 2: 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.
• 1:34
  Speaker 2: 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 in the AI space.
• 1:44
  Speaker 2: 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.
• 1:57
  Speaker 2: You can purchase 12-month access to the AI Learning Hub for $399.
• 2:02
  Speaker 2: And if you want to get more information on that, you can go to sidecarglobalcom slash hub.
• 2:09
  Speaker 2: And in honor of the 25th episode of Meath, I guess we have a special edition storm episode today.
• 2:14
  Speaker 2: If you all have listened before, you know, meath and I are located in New Orleans and right before we hopped on this call to record the podcast, we got a severe thunderstorm alert with 80 mile an hour winds, so stay tuned for this episode.
• 2:28
  Speaker 2: Hopefully neither of us lose power, but you might hear some soothing thunder and rain in the background.
• 2:33
  Speaker 1: It'll be fun either way.
• 2:35
  Speaker 2: Of course, all right.
• 2:36
  Speaker 2: So today's topics we are talking about first, llama 3, very exciting news announcement.
• 2:42
  Speaker 2: We'll be talking about Y Combinatorators winner 2024 class and the AI startups that are within that class, and then we'll have an interesting conversation around AI and the middle class to wrap up this episode.
• 2:54
  Speaker 2: So, first and foremost, lama 3.
• 2:57
  Speaker 2: Meta plans to release smaller versions of its upcoming AI language model, lama 3, next week.
• 3:02
  Speaker 2: These compact models will make Lama 3's capabilities more accessible to developers and researchers without the need for expensive hardware.
• 3:10
  Speaker 2: While the full Lama 3 model is expected to be very large, potentially surpassing 140 billion parameters, the smaller variants will be optimized for running on consumer hardware like GPUs and cloud instances.
• 3:23
  Speaker 2: This follows Meta's strategy with previous Lama releases for providing scaled-down versions alongside the flagship model.
• 3:30
  Speaker 2: By making the model more widely available, meta hopes to accelerate research and development in areas like conversational AI, code generation and multimodal applications.
• 3:40
  Speaker 2: Meta's approach contrasts with the closed ecosystems of some AI companies.
• 3:45
  Speaker 2: The company believes that open sourcing key infrastructure, while retaining control over future model iterations, will maintain its competitive advantages in the long run.
• 3:55
  Speaker 2: So, amit, my first question for you here when talking about the size of models, we talk about parameters.
• 4:02
  Speaker 2: I don't know if we've ever really talked about that on this podcast, but what are parameters and how do these relate or how are they different from context windows?
• 4:11
  Speaker 1: Mal, that's a great question.
• 4:12
  Speaker 1: Parameters basically refer to the size of the model, whereas the context window is like it's short-term memory, it's the ability for it to remember information that you prompt it with and remember longer conversations.
• 4:22
  Speaker 1: So when we talk about a model having 140 billion parameters or 7 billion parameters or whatever the number is GPT-4 is believed to be over a trillion parameters what you're talking about essentially is the rough.
• 4:35
  Speaker 1: It's a rough way of describing the size of the model, which has generally correlated with the power of the model In the neural network.
• 4:43
  Speaker 1: When you're training, using all the different training techniques we've only touched on a little bit in the past, what ends up happening in this training process is the algorithm doesn't change.
• 4:53
  Speaker 1: The algorithm for the neural network essentially is the same, but the weights are being constantly recalculated through the training process, and that can also be adjusted through what's called a fine-tuning process, through the training process, and that can also be adjusted through what's called a fine-tuning process.
• 5:06
  Speaker 1: And ultimately the weights are the model and the number of weights essentially refers to this parameters, and so it's effectively like how big is the model?
• 5:13
  Speaker 1: That's the way to think about it.
• 5:14
  Speaker 1: So we talk about something like Mistral 7B, which is a 7 billion parameter model, and Mistral.
• 5:21
  Speaker 1: On their website they call that Mistral Tiny or the Mixtral model that we've talked about in this pod, which has eight times 7 billion parameters, because it's a mixture of experts model.
• 5:31
  Speaker 1: You have these models of various sizes which do different things.
• 5:36
  Speaker 1: What is really exciting about models to me is certainly bigger models in theory are interesting, but we found very clearly through research and also through practical applications that the quality of the training data is actually just as important, if not more important, than the rough size of the model.
• 5:54
  Speaker 1: So and Mistral really is the company at the forefront of this being super efficient with training models that actually have a lot smaller training sets and therefore have fewer parameters, but ultimately are as capable as some models considerably larger.
• 6:08
  Speaker 1: So the model size thing is certainly worth paying attention to, but it's not necessarily the most effective way to determine the power of the model.
• 6:22
  Speaker 1: So it'll be very interesting to see when they release Lama 3, even with the smaller versions that are supposed to be released next week what their capabilities are.
• 6:28
  Speaker 1: I'm sure they'll release benchmark scores and all the typical benchmarks and then, of course, the larger model that is expected to be available later this year, maybe this summer, sometime perhaps, will have GPT-4 or surpass GPT-4 capabilities.
• 6:42
  Speaker 1: That's certainly what I'm expecting.
• 6:43
  Speaker 1: I'm speculating by saying that last comment, but Lama 3 is Meta's flagship AI effort and they are not trying to be a second place contender, they want to win.
• 6:57
  Speaker 1: They want to have capable AI as strong as OpenAI's frontier models and everybody's been trying to chase GPT-4.
• 7:04
  Speaker 1: Until recently it was the only category only model in that category.
• 7:09
  Speaker 1: But now we have Cloud 3, opus and also Gemini Ultra and Gemini 1.5 Pro that are in various ways as good, in some cases better.
• 7:16
  Speaker 1: So it's going to be really interesting to see how this evolves as we get more information next week with the official release.
• 7:24
  Speaker 2: Given all the models you just mentioned in your opinion, which do you see, which do you foresee being kind of the forerunner within the next few months or by the end of the year?
• 7:33
  Speaker 1: The short answer is it depends.
• 7:35
  Speaker 1: It depends on what you're doing.
• 7:36
  Speaker 1: So we know that.
• 7:37
  Speaker 1: You know GPT-4 continues to impress in many broad categories.
• 7:42
  Speaker 1: There are models that are much smaller that are really really good at specific things.
• 7:48
  Speaker 1: There's a company called Stability AI that is best known for stable diffusion, but they actually have models in a variety of other contexts and they have some language models too that are very small and they have a completely different strategy.
• 8:01
  Speaker 1: And an interesting side note in the open source AI world, their CEO actually just stepped down, which is a story unto itself to pursue a different approach to distributed AI.
• 8:13
  Speaker 1: His viewpoint is that, even with open source, we haven't gone far enough in distributing the possibilities of AI to the masses, and so he's working on something that seems to be at the fusion of Web3 and distributed computing and AI open source.
• 8:26
  Speaker 1: But I digress.
• 8:28
  Speaker 1: The point is is that these models you have to look at them like specialists in a way, and say what is it that I'm trying to solve for and which model might fit best, and ultimately, in fact, that's actually what these mixture of experts models are doing, like Mixtral 8x7b, which is a wonderful technical name, probably not the best marketing name, but I do think Mixtral is kind of cool.
• 8:52
  Speaker 1: What it's doing is it has eight different 7 billion parameter models, and it has this dynamic capability to pick two or more of those models for each request and then to use those models based on what would be best in the sequence of token generation, essentially, and so what's powerful about that is Mixtrel has performance in excess of GPT 3.5, even though it's considerably smaller.
• 9:18
  Speaker 1: So I think there's a lot of interesting things to talk about there.
• 9:23
  Speaker 1: Ultimately, there is no one model to rule them all, and we're increasingly going to see that.
• 9:28
  Speaker 1: Now we don't know what OpenAI is up to, because they're famously secretive about what's happening with GPT-5 or GPT-4.5, whatever they want to call it model.
• 9:35
  Speaker 1: This year it'll probably reset the high watermark for AI capability and language models, and it'll be interesting to see where it goes from there.
• 9:47
  Speaker 1: But it's going to go faster and more aggressive from this point for sure.
• 9:52
  Speaker 2: Yep, I like what you said there.
• 9:53
  Speaker 2: I think the key is it really does depend Thomas Altman, who I've mentioned on this podcast before.
• 9:58
  Speaker 2: He's co-founder of Tasio and we co-host the Intro to AI webinar every month through Sidecar and our last version of that webinar in March he actually said he thinks Claude is the best tool to use when writing marketing copy, which is interesting because I myself typically use ChatGPT.
• 10:13
  Speaker 2: But I think that is kind of the future figuring out what tools or what applications, what models perform best in certain scenarios.
• 10:20
  Speaker 2: Also, on that note, we do have another version of that Intro to AI webinar next week, actually on April 16th, so we'll include that registration link in the show notes if you're interested.
• 10:31
  Speaker 1: That will be a great webinar.
• 10:33
  Speaker 1: And one thing I wanted to quickly comment on is, even if you've attended prior Intro to AI webinars, it's always worth checking out the new edition because these webinars are constantly updated as you follow along in this journey.
• 10:44
  Speaker 1: If you listen to Mallory and I speak about these topics, you know they're changing all the time.
• 10:49
  Speaker 1: So the Intro to AI webinar is a way to reinforce what you already know, even if you're already gone down that path, and it's certainly a way to get the fresh new content.
• 10:57
  Speaker 1: And if you can't make it but you're interested, register anyway, because then you'll get access to the recording.
• 11:03
  Speaker 2: Absolutely, and Thomas shows Claude Gemini Chats UBT.
• 11:07
  Speaker 2: We talk about Midjourney, we talk about Munch, which we've discussed on this podcast before, lots of cool demos of tools and tons of association-specific use cases.
• 11:16
  Speaker 2: So I would highly recommend that you check that out, amit, it seems like more and more within the podcast maybe it's confirmation bias, but I do feel like in the last few episodes we've spent a lot of time talking about smaller models, whereas we weren't talking so much about smaller models back in November and December.
• 11:32
  Speaker 2: What do you see as the most exciting business use cases, potentially for smaller models?
• 11:39
  Speaker 1: Well, if we talk about smaller models that are at least as capable as GPT 3.5, which is a very capable model, that's when ChatGPT burst onto the scene in fall of 22.
• 11:48
  Speaker 1: That was powered by GPT 3.5.
• 11:50
  Speaker 1: And in fact I believe the free public version of ChatGPT is still powered by GPT 3.5.
• 11:55
  Speaker 1: So it's a very capable model, not as strong as GPT 4, but very good.
• 11:59
  Speaker 1: So, as long as we're talking about models of that caliber and better, when you talk about small models, you're talking about efficient, super low cost, very fast models.
• 12:09
  Speaker 1: So the things you can do really open up possibility, because when you're dealing with a GPT-4 or a Gemini, ultra, opus type model that category, they're big, they're heavy and they're also slow.
• 12:21
  Speaker 1: There's a lot of work being done to speed things up in the world of AI inference, which is the runtime speed, but right now the size of the model is directly correlated to the inference speed.
• 12:32
  Speaker 1: So smaller models are faster, they're cheaper, they're more energy efficient and they're also things that, because of open source, you can use in lots of creative ways that you can't necessarily do with models that are closed source.
• 12:46
  Speaker 1: So I think that small models are a more exciting area to play with, because so many problems can be solved by GPT-3-5 class models that people haven't even touched on.
• 12:56
  Speaker 1: You know we've talked about use cases for associations a bunch in this podcast.
• 13:00
  Speaker 1: I spend time with a lot of executive leaders in this market talking about the same types of things.
• 13:04
  Speaker 1: I spend time with a lot of executive leaders in this market talking about the same types of things, and any time there's language involved, you know a GPT-3-5 class model can help you.
• 13:10
  Speaker 1: As an example, we often see unstructured content meaning like just text or images coming into an organization.
• 13:19
  Speaker 1: Someone might submit a response to a call for speakers or a call for papers, and in that process it's either all or none.
• 13:27
  Speaker 1: People think about automation as being binary.
• 13:29
  Speaker 1: They can either automate it all or automate none of it, and in reality there are shades of gray there.
• 13:34
  Speaker 1: You can automate portions of it, and the interesting thing in a unstructured content process, like when you get content from a person submitting, let's say, a talk for your annual conference, is that there's a lot that can be done with even a GPT-3-5 class models.
• 13:50
  Speaker 1: First of all, just to classify the talk, let's say that you're in a particular branch of science, like physics, for example.
• 13:56
  Speaker 1: Within your conferences, you probably have many different tracks, many different sub-areas, and the people that are going to evaluate each of these papers that are coming in the door have to be specialists in those fields.
• 14:09
  Speaker 1: They can't just be random people.
• 14:10
  Speaker 1: And so in that context you can use a very small model as a basic classifier to just route these applications to the right people, eliminating the need for a person to look over the proposal and actually route it.
• 14:23
  Speaker 1: So that's a very simple example.
• 14:25
  Speaker 1: We're not automating the entire process of selecting or rejecting a submission for a call for speakers, but we're taking a little chunk of it off and you can do that very inexpensively.
• 14:36
  Speaker 1: In fact it's pretty close to free already with a model of the caliber of GPT-3.5.
• 14:41
  Speaker 1: So I'm really interested to see what the smaller LAMA-3 models do.
• 14:45
  Speaker 1: We've talked in this pod briefly in the past about a Microsoft model called PHY-2, which is spelled P-H-I-2,.
• 14:50
  Speaker 1: Certainly the Mistral models.
• 14:53
  Speaker 1: You can do so much with these things.
• 14:55
  Speaker 1: We often also say that this is the worst AI you're ever going to get the AI you have today.
• 14:59
  Speaker 1: And even if AI never advanced from where we are today, if you simply took the time to integrate into your business and into your profession the things you can now do with these models at low cost, very close to no cost.
• 15:13
  Speaker 1: It would take you years to take full advantage of this stuff.
• 15:15
  Speaker 1: So to me, the small models just compound that.
• 15:18
  Speaker 1: The other thing is is that small models are getting smarter.
• 15:21
  Speaker 1: They're not staying at GPT-3.5 levels.
• 15:23
  Speaker 1: Models are getting smarter.
• 15:26
  Speaker 1: They're not staying at GPT-3.5 levels.
• 15:27
  Speaker 1: Very soon you'll see an open source tiny model like a 7 billion parameter model that's as smart as current GPT-4.
• 15:30
  Speaker 1: It might take about another year or so, but you're seeing the smaller models getting smarter and smarter.
• 15:35
  Speaker 1: That's partly algorithmic improvement, but it's also heavily driven by the fact that the large models are actually training the small models, which makes the small models smarter.
• 15:44
  Speaker 1: So it's kind of this crazy loop that's happening, and I'm actually honestly more excited about what you can do with the small models, because you can weave them in everywhere and then, for those organizations that are deeply concerned about privacy and security, you can run them yourself.
• 15:57
  Speaker 1: You can inference on your own hardware, in your own virtual private cloud or literally on your own hardware if you wanted to, and you can't do that with the frontier models.
• 16:06
  Speaker 2: That makes sense.
• 16:06
  Speaker 2: I'll speak from an average user, perhaps a power user of ChatGPT, for example.
• 16:12
  Speaker 2: I understand that you're saying the smaller models are faster and are more specific in terms of what actions they can take, like classifying speaker submissions, for example.
• 16:23
  Speaker 2: However, from a user perspective, I want to use the model that's most powerful, right?
• 16:28
  Speaker 2: I want to use GPT-4.
• 16:29
  Speaker 2: I want to use Cloud 3 Opus.
• 16:30
  Speaker 2: I'm still not fully understanding from an individual point of view.
• 16:34
  Speaker 2: When would I opt to try out a smaller model versus the most powerful one?
• 16:40
  Speaker 1: As an end user, you probably have no reason to care about that, because you're going to use an interface like a Perplexity or you're going to use a ChatGPT or Cloud's UI, and those models actually are.
• 16:51
  Speaker 1: Those are actually not single models like behind the scenes there's a lot going on in terms of looking at the request and handling the request in different ways, because when a request comes in, you can look at it and say what is this Like?
• 17:02
  Speaker 1: What's Mallory asking for?
• 17:04
  Speaker 1: She asking what the weather is today.
• 17:06
  Speaker 1: Well, I don't need to employ the most state-of-the-art, biggest powerful model to handle her request.
• 17:11
  Speaker 1: I can just give it to a GPT-3.5 class model.
• 17:14
  Speaker 1: It can, in turn, talk to a weather service, get the information, pass it back to you.
• 17:18
  Speaker 1: Similarly, a lot of classification and summarization tasks can be done very well with smaller models, if you want something that is more of a kind of approaching reasoning logic, things like that.
• 17:32
  Speaker 1: That's when we get into really needing the heavier duty models, because the GPT-35 class models are terrible at that.
• 17:37
  Speaker 1: Gpt-4, frankly, is not the best at it either, and that's really what is most exciting about on the frontier.
• 17:42
  Speaker 1: When we talk about frontier, we're saying the biggest newest models being able to do things that the current models cannot do at all.
• 17:49
  Speaker 1: So what you're going to get in your consumer experience and you very much see this in Copilot inside the Microsoft suite is they're bringing you down to like the model level you need and then occasionally they'll burst up to the bigger ones.
• 18:02
  Speaker 1: In fact Copilot in my experience for the last couple months is not nearly as good as just using GPT-4 inside ChatGPT, because I think they have tamped it down quite a bit to make it scale.
• 18:12
  Speaker 1: So I think from an end user perspective a lot of this conversation is just going to go away, because if you think about GPT-4 and certainly with GPT-5 class models, not those specific models you're going to get so much out of it that most users will get everything they need from it and won't necessarily be too concerned.
• 18:29
  Speaker 1: It's kind of like you don't really know what processor you have in your laptop and it just works really well.
• 18:33
  Speaker 1: And if you get a new computer in three years you're probably not going to notice that much of a difference.
• 18:38
  Speaker 1: You're probably replacing your computer because you just kind of need a new computer.
• 18:41
  Speaker 1: The physical hardware has gotten to the point where it's broken down a little bit.
• 18:45
  Speaker 2: Okay, that helps.
• 18:46
  Speaker 2: So from a user perspective, we're just going to keep using the models that we use On the back end.
• 18:50
  Speaker 2: Perhaps things will change.
• 18:51
  Speaker 2: We'll have these mixture of experts, architectures or small models doing certain tasks, but the way we interact with it will stay roughly the same.
• 18:59
  Speaker 1: I think that's true for a period of time.
• 19:01
  Speaker 1: I mean, there's always an opportunity for some disruptive user experience.
• 19:04
  Speaker 1: I think that's where one of the strategic advantages that you can build if you're building these tools is to think of a novel user experience, something that's even lower friction, more natural, more intuitive.
• 19:14
  Speaker 1: Chat is great.
• 19:15
  Speaker 1: It's so different than what we've had in the past, and we've had chat for years.
• 19:19
  Speaker 1: There's been different kinds of chat apps.
• 19:21
  Speaker 1: They've just all been universally terrible.
• 19:23
  Speaker 1: Now, all of a sudden, they're good, so it's a great experience.
• 19:26
  Speaker 1: But people are rapidly kind of building hybrid experiences where it's part chat and part structure.
• 19:32
  Speaker 1: There's things like chat being infused throughout apps.
• 19:35
  Speaker 1: So I think there's a lot of innovation that's going to happen in user experience.
• 19:39
  Speaker 1: That's where I'm really excited for consumer apps and don't get me wrong, improvements in underlying models are going to power new capabilities that are very exciting for the average consumer.
• 19:48
  Speaker 1: I just think it's one of these things where most people, like you, definitely are a power user.
• 19:54
  Speaker 1: Most people aren't doing what you're doing with ChatGPT already and they have a ramp up in terms of just what they know to ask for and what they will want out of these systems.
• 20:03
  Speaker 2: Mm-hmm.
• 20:03
  Speaker 2: So we've seen a wave, certainly, of open source models from Google, for example, Elon Musk's XAI, like we discussed I think that was last week Mistral, like you mentioned, and then we've also seen some powerful closed source models like Cloud3, Opus.
• 20:18
  Speaker 2: Do you see large language model advancements leaning one way or the other in terms of open versus closed?
• 20:25
  Speaker 1: I think it's going to be fun to watch.
• 20:26
  Speaker 1: That's really my only conviction here.
• 20:28
  Speaker 1: I believe in open source personally, because I think that the only way to fight bad things with AI is to develop really powerful, good AI.
• 20:36
  Speaker 1: And closed source is really hard to know if it's good or bad.
• 20:41
  Speaker 1: I'm not accusing anyone at Anthropic, which makes cloud, or open AI, which makes chat GPT, of malicious intentions whatsoever, but you just don't know because these are very small companies with an extreme amount of power and open source.
• 20:54
  Speaker 1: The theory behind it is it potentially democratizes some of this technology, and I think that's important.
• 20:59
  Speaker 1: That being said, closed source systems right now are still far more powerful because of the many billions of dollars it takes to stand them up.
• 21:06
  Speaker 1: So, to the extent that model training continues to be that expensive, I think closed source is going to continue to have some advantages.
• 21:14
  Speaker 1: But it's hard to say and algorithmic improvements are happening very rapidly.
• 21:19
  Speaker 1: There's a lot that's also going on in terms of training efficiency.
• 21:22
  Speaker 1: What used to take $10 million to train a model of a GPT-3.5 class capability, you can now train with $5,000, the same model capability.
• 21:31
  Speaker 1: So the same things are happening over and over again that we talk about here in terms of exponential growth.
• 21:36
  Speaker 1: So the short answer is I really don't know, but I think it's good to have both.
• 21:39
  Speaker 1: I think that it's a healthy competitive environment.
• 21:43
  Speaker 2: I'm sure it's something we'll continue to discuss on this podcast well into the future.
• 21:47
  Speaker 2: I'm really excited for topic two because we will be chatting about some interesting AI use cases, some interesting AI startups that are in Y Combinator's Winter 2024 class.
• 21:58
  Speaker 2: If you don't know, y Combinator is an American technology startup, accelerator and venture capital firm that was launched in 2005.
• 22:04
  Speaker 2: Notable companies that have gone through Y Combinator include Airbnb, dropbox, instacart and Coinbase, among others.
• 22:11
  Speaker 2: Y Combinator's Winter 2020 4 class features a mega batch of 157 AI startups, a trend which highlights the growing adoption of AI across various industries.
• 22:23
  Speaker 2: So I want to give you a little bit of an overview of how some of these startups are grouped.
• 22:27
  Speaker 2: So first we have software development 44 startups in the class are dedicated to innovating software development with AI, automating routine coding tasks and enhancing the development process.
• 22:38
  Speaker 2: There's a company called Codient AI that focuses on automatically fixing bugs, and then a company called Momentic AI that specializes in AI-powered automated testing.
• 22:48
  Speaker 2: The next kind of grouping we have is customer service, with 12 startups aiming to optimize customer interactions using AI-powered solutions like AI-driven receptionist services and AI-powered platforms for customer engagement.
• 23:02
  Speaker 2: Retail AI is one of those companies that builds infrastructure for voice artificial intelligence, another one of those companies is Kiosk, which is an AI-driven platform focused on WhatsApp marketing, which sounds pretty interesting to me.
• 23:15
  Speaker 2: The next group we have is Biotech and Healthcare, with 10 startups in that group.
• 23:21
  Speaker 2: They're focusing on areas where AI can reduce the time and cost of developing new treatments.
• 23:26
  Speaker 2: Radmate AI aims to serve as an AI co-pilot for radiologists.
• 23:31
  Speaker 2: Metafyco offers no-code data analysis for life sciences.
• 23:36
  Speaker 2: This next category, I think, is probably most well, I won't say most interesting for me, but Amit and I had an interesting talk about these companies which is AI as a creative partner.
• 23:45
  Speaker 2: So 14 startups are enabling creative individuals to leverage AI tools as collaborative partners in their creative process.
• 23:53
  Speaker 2: Pocketpod creates fully personalized news podcasts made just for you every day, potentially a big competitor to the sidecar sync.
• 24:03
  Speaker 2: And then we've got Magic Hour and Infinity AI.
• 24:06
  Speaker 2: That are both video creation platforms, which is really no surprise given how popular and how much conversation has existed around Sora, openai's text-to-video tool.
• 24:17
  Speaker 2: One of the last groups we have is finance and investing.
• 24:20
  Speaker 2: Ai is being applied there, with 11 startups developing AI-powered solutions for various use cases, like Clarum AI, focusing on AI-accel, ai accelerated due diligence, and Powder, which is an AI co-pilot for wealth managers.
• 24:33
  Speaker 2: And finally, our last group.
• 24:35
  Speaker 2: Seven of the startups within this mega batch seek to revolutionize the physical world with InspectMine providing faster inspection reports.
• 24:44
  Speaker 2: Hey Purple is a company that provides AI property managers.
• 24:49
  Speaker 2: And then there's another company called DraftAid that converts 3D models to CAD drawings using AI.
• 24:56
  Speaker 2: So I personally had a ton of fun clicking on all these links, exploring all these companies.
• 25:00
  Speaker 2: It was kind of a rabbit hole.
• 25:02
  Speaker 2: Amit Y Combinator has a knack, you could say, for identifying potentially very successful companies.
• 25:08
  Speaker 2: Are you surprised that 44 of the startups in this group fall into the AI and software development category?
• 25:15
  Speaker 1: Not at all.
• 25:16
  Speaker 1: I think that's a natural concentration because there's, first of all, you build what you know, and so there's so many people YC is very much a group that has had a great track record of picking big ideas, swinging for the fences, going forward.
• 25:32
  Speaker 1: That's really what they're about and so there is so much inefficiency and opportunity in software development.
• 25:38
  Speaker 1: The two you mentioned focus on bug fixing and testing are where probably 70, 80% of the software development lifecycle is spent in that maintenance, fix, test kind of mode.
• 25:49
  Speaker 1: Building completely new software is also an area I can help a ton with.
• 25:53
  Speaker 1: But once you have a code base and you want to maintain it, fix it, upgrade it, test it, those are naturally things that are going to be automated For a particular startup.
• 26:03
  Speaker 1: I never really look at it and say, oh well, this particular one or that one is going to be a great one.
• 26:07
  Speaker 1: I think that these are categories that are so broad that there's going to be a lot of acquisition activity.
• 26:12
  Speaker 1: If someone figures something out, that's good.
• 26:14
  Speaker 1: I think you're going to see major platforms from providers like Atlassian and Microsoft and others incorporate these capabilities as native features, either that they've built themselves or that they have acquired, and you're going to see a lot of the bug fix test cycle go away.
• 26:29
  Speaker 1: So to me that's the kind of the obvious place and it's obviously a category I know well.
• 26:36
  Speaker 1: Software development is kind of this intersection of creativity and science.
• 26:39
  Speaker 1: There's this mix that goes on when you build software.
• 26:41
  Speaker 1: So I think it's a great area for generative AI to have an impact.
• 26:44
  Speaker 1: We certainly have seen it across all of our companies.
• 26:47
  Speaker 1: You know it's interesting because just yesterday the team at Member Junction released version 1.0 of their software, which is the open source common data platform for associations and nonprofits and that's hosted on GitHub.
• 26:59
  Speaker 1: And you know GitHub is a leader in AI with Copilot.
• 27:03
  Speaker 1: They were actually the original Copilot out there for software development and they're going to be introducing a whole bunch of other capabilities in that platform.
• 27:09
  Speaker 1: I'm certain that will do exactly what you described.
• 27:13
  Speaker 2: You are a software developer at Meath and you, at least from what I've heard, you like to develop software for fun as well as for business purposes and professional purposes.
• 27:21
  Speaker 2: What do you think about AI impacting software development, or what have you seen thus far?
• 27:28
  Speaker 1: Well, I think software is fun to develop because it's a combination of engineering and building things and science with creativity.
• 27:35
  Speaker 1: That's what makes software fun.
• 27:36
  Speaker 1: It is a form of expression in a lot of respects, and so, for me, for 40 years now, basically, I've been doing this stuff and I love it, and I still do.
• 27:45
  Speaker 1: My point of view is that I view these things as catalysts, as accelerators, not taking anything away from me, but actually empowering me to go faster and do better.
• 27:55
  Speaker 1: Because I don't like doing the redundant, boring things.
• 27:58
  Speaker 1: I like to have the AI take care of everything I don't want to do and then focus on thinking about the architecture of software, working with other team members and saying, hey, how do we want to build this and what do we want it to do, and building some of the interesting codes.
• 28:10
  Speaker 1: Still, I think there's definitely a place for that.
• 28:13
  Speaker 1: I think that that might extend to other fields as well.
• 28:16
  Speaker 1: I don't know, but I think there's room for AI to take on a lot of the redundant things.
• 28:20
  Speaker 1: No-transcript, fixing them, finding them, retesting it's just really laborious and it's a perfect fit for AI.
• 28:42
  Speaker 1: So I don't see that taking anything away from the creative side of software development.
• 28:46
  Speaker 2: Personally, Looking through this list of AI startups that I have in front of me, I think pretty much every one of these could impact an association in some way, either the way an association runs at its core or even, like the industry, specific ones.
• 28:59
  Speaker 2: There are likely associations for property managers and things like that.
• 29:02
  Speaker 2: Are there any of these categories that you find are most exciting for associations?
• 29:09
  Speaker 1: I think the things that you mentioned about.
• 29:10
  Speaker 1: If you look with the external lens of what the association does, you have to pay attention to this stuff.
• 29:15
  Speaker 1: So if you're in anything around radiology or similar fields, understanding what's happening there with these types of co-pilots is key, of course, in that field.
• 29:23
  Speaker 1: This is nothing new.
• 29:24
  Speaker 1: Ai has been leveraged as an assistive tool in these fields for many years, but it's accelerating, so there's opportunities there.
• 29:33
  Speaker 1: I think that if you're in a field like real estate, these tools around property management or home inspection or real estate inspection are super interesting.
• 29:42
  Speaker 1: So you have to look at it with the external lens.
• 29:44
  Speaker 1: How are these disruptive forces going to affect your field, your profession, your industry, and think about the way you need to adapt to provide services, training, education, et cetera to what the future needs will be based on that disruptive likelihood.
• 30:00
  Speaker 1: And then, of course, on the internal side, how can you do that work better?
• 30:03
  Speaker 1: How can you essentially employ an army of AI capabilities to supercharge what you're able to do and how you can deliver it, which might create opportunity to develop entirely new products and services as an association that you previously could not do.
• 30:21
  Speaker 1: My favorite example of that is this idea of a knowledge assistant.
• 30:23
  Speaker 1: People are deploying chatbots, certainly to do customer service stuff.
• 30:27
  Speaker 1: We've talked about Klarna, for example, in a prior podcast, which is one of these buy now pay layer companies, and they've automated a large portion of their kind of rote customer service with actually great results.
• 30:37
  Speaker 1: Customers are as happy with the AI as they have been with human respondents, but also they get an answer much faster.
• 30:43
  Speaker 1: That's a great use case.
• 30:44
  Speaker 1: But what if you could take all of the knowledge of your association, package it up into an AI with high accuracy and then unleash that to serve your industry, serve your profession, to advance what everyone in that space does?
• 30:57
  Speaker 1: That gets exciting, and that was not a capability you could even dream of pre-AI.
• 31:01
  Speaker 1: So to me it's thinking about new services you can provide that raise the bar in terms of your impact in your sector.
• 31:08
  Speaker 2: Okay.
• 31:08
  Speaker 2: So, for example, the tool I mentioned the company I mentioned called DraftAid, which I'll go over one more time converts 3D models to CAD drawings using AI.
• 31:16
  Speaker 2: Should the Hypothetical Architects Association there is one, and I don't know the name of it off the top of my head Should they be strategically thinking, creating content around companies like this, like interacting with companies like this?
• 31:30
  Speaker 1: Should they sit back and wait and they need to be deep in this stuff.
• 31:36
  Speaker 1: They need to get people who are experts in the field, because a lot of associations within the association don't have deep expertise in the field, or only a little bit of it, and so someone like the American Institute of Architects should absolutely be all over this.
• 31:48
  Speaker 1: They should be really studying it I'm sure that they have people that are working on this and to go deep and to understand what's happening in the space, because it's going to affect architects, it's going to affect people who do drafting, it's going to affect all of it, and I think that's true for every branch of engineering and science and basically everything Right.
• 32:08
  Speaker 1: So my point of view is that the association has to develop a competency in AI in the context of their audience.
• 32:17
  Speaker 1: You have to do that because otherwise, you're going to very quickly be out of touch with what people in your field are doing and what they need to do, and it's your job as an association, in my mind, to help lead the way, to help say hey, this is where we need to go as a profession.
• 32:29
  Speaker 1: We need to embrace this tool in this way in order to be more effective, more productive, create more opportunity, have better quality, et cetera.
• 32:38
  Speaker 2: And what if someone listening is thinking well, we don't have the subject matter experts, this stuff is so new.
• 32:45
  Speaker 2: We don't have anyone who can speak on DraftAid, because this is an AI startup.
• 32:47
  Speaker 2: What would you say to that?
• 32:48
  Speaker 1: You go find them.
• 32:49
  Speaker 1: That's your job to go figure it out.
• 32:51
  Speaker 1: And there's lots of resources out there.
• 32:52
  Speaker 1: There's lots of people talking about this, so you go to the YC list and look at what people are doing.
• 32:56
  Speaker 1: You call up those startups and say, hey, you know, we're such and such association, we represent this profession.
• 33:01
  Speaker 1: We want to talk to you, we want to hear what you're up to, Be curious, Go and explore and investigate and learn.
• 33:05
  Speaker 1: Maybe get some of your more closely aligned, highly engaged members who are, you know, have a knack for technology or whatever to get involved as well.
• 33:15
  Speaker 1: I think there's a lot of ways to approach it, but sitting still is not the right approach.
• 33:18
  Speaker 1: I think that that's the key is just, you got to get there and explore and learn and in your field, you're going to see disruptive forces.
• 33:31
  Speaker 1: Some associations that I talk to are looking at it as a protectionist kind of thing, like how do we stop AI from negatively affecting our people, our industry, our profession?
• 33:37
  Speaker 1: That's certainly a concern for a lot of labor unions, which broadly fall into the category that we focus on helping, and why I totally empathize with that.
• 33:46
  Speaker 1: I also think that it's the job of those organizations to figure this out and say how will AI disrupt this job type and then to learn that content and then help bring that to their people so that their people don't become obsolescent.
• 34:00
  Speaker 1: You know we repeat ourselves a lot in life and in this podcast and we say that you know you're not likely to replace by AI by itself, at least at the moment, but you're very likely to be replaced by someone who knows how to use AI to do the job that you do.
• 34:13
  Speaker 1: So if you're a podcaster and you don't know how to use AI tools, I don't know how you can keep up with a podcaster that is good at AI.
• 34:19
  Speaker 1: Same thing for writing, same thing for engineering, same thing for software development, and I think it's the association's job to figure that out.
• 34:26
  Speaker 1: Like for whatever that profession is, how can AI help that profession and how do we train people to understand the technology?
• 34:33
  Speaker 1: And you know whether or not you like it.
• 34:35
  Speaker 1: It's there.
• 34:36
  Speaker 1: No-transcript.
• 34:59
  Speaker 2: Yep, I love that sentiment.
• 35:01
  Speaker 2: I think everyone has fears around this, everyone has questions, and how powerful could it be if your organization, your association, is the place that people can go to learn about it.
• 35:07
  Speaker 2: I think that's really powerful, amit, on a lighter or maybe darker note, depending on how we look at it, what do you think about PocketPod?
• 35:13
  Speaker 1: The company that's creating on-demand podcasts for everyone every day.
• 35:15
  Speaker 1: What do you think about that?
• 35:15
  Speaker 1: I mean, it's just an example of personalization at scale.
• 35:18
  Speaker 1: Taking the idea of personalization of content and saying, hey, we're going to create a personalized podcast just for you, mallory, and it's going to know what you're interested in, whether you tell it that or somehow it figures it out through other signals and then generates the script for that podcast based upon what it knows about you and then synthesizes audio, maybe even video, just for you.
• 35:35
  Speaker 1: It's very much something that's within the realm of current AI.
• 35:40
  Speaker 1: If you break down some of the things we've talked about in this podcast and said, hey, if you have an idea for a podcast episode, you can definitely get the AI to create a script for you If you have the script, we know that there are tools like 11 Labs and there's tools from OpenAI and other people that can synthesize text to audio, and so the idea of being able to do this is really a matter of connecting some building blocks together.
• 36:02
  Speaker 1: Doing it well, of course, is another level, like doing it in a way where you find it engaging and interesting and there's some art along with the science.
• 36:10
  Speaker 1: So I'm excited to see stuff like that.
• 36:12
  Speaker 1: I actually think that when you have a podcast like the Sidecar Sync or much bigger shows, obviously that are out there, believe it or not.
• 36:19
  Speaker 1: There are some shows bigger than the Sidecar Sink, yeah, and you know, it amazes me too, mallory.
• 36:29
  Speaker 1: But those other shows that have large followings, like Tim Ferriss or someone like that or Lex Friedman, you know, those are shows where the personality of the producers of the content really are a big part of what draws people in and then certainly keeps them.
• 36:39
  Speaker 1: If people like the host, if people like the content, their way of communicating, I think there's a degree of protection for podcasters around that, at least to some degree.
• 36:48
  Speaker 1: You could say, hey, well, sidecar Sync, will we embrace this idea?
• 36:52
  Speaker 1: So if PocketPod is a great tool, will we have a Sidecar Sync personal edition right, where you can basically train the PocketPod tool on all of our podcasts we've ever done?
• 37:03
  Speaker 1: By the time that tool is available, maybe we'll be on episode 50 or 100 or something.
• 37:07
  Speaker 1: We feed every episode we've ever done into PocketPod and then we make it available to our listeners.
• 37:13
  Speaker 1: To create a personal version Could be interesting, right.
• 37:15
  Speaker 1: It could be an interesting thing to do where it overweights on most recent podcasts but then also has availability in older ones.
• 37:23
  Speaker 1: And then, to take that idea even a step further, beyond what PocketPod's talking about, what if it was a two-way conversation?
• 37:28
  Speaker 1: So you're listening to the podcast I know some people probably do this but you start talking back to the podcast and saying, well, what do you think about this?
• 37:35
  Speaker 1: But if you had that kind of interactive experience, that could be very, very fascinating as well.
• 37:41
  Speaker 1: So, um, I think when you kind of draw, pull these things and draw them out to logical conclusions, it gets fascinating.
• 37:47
  Speaker 1: Um, who knows if this particular company will be successful?
• 37:50
  Speaker 1: There's a lot more than just the tech that has to make that.
• 37:53
  Speaker 1: You know has to be there to make them successful as a business.
• 37:55
  Speaker 1: But I think the idea is great and there's some lessons to be had for the association market.
• 37:59
  Speaker 1: Where you're talking about new content modalities, you're talking about something at scale that wasn't possible pre-AI.
• 38:05
  Speaker 1: It's an AI scale problem.
• 38:07
  Speaker 1: We like to say that a lot.
• 38:09
  Speaker 1: When we're talking about personalization is a great example and it's certain kinds of personalization are becoming really easy to do.
• 38:15
  Speaker 1: Most associations I know are not doing any personalization or any meaningful personalization.
• 38:20
  Speaker 1: I talk to a lot of association groups and one of the questions I ask them is where are you on content personalization?
• 38:26
  Speaker 1: First of all, do you believe in the idea?
• 38:28
  Speaker 1: Are you bought into it?
• 38:29
  Speaker 1: Most people say yes to that and I say, well, have you done anything?
• 40:22
  Speaker 1: And the most common answer is basically no.
• 40:25
  Speaker 1: We maybe put in Mallory, your first name, maybe we segment a little bit and say we'll send you content in category A instead of category B, but that's about it.
• 40:35
  Speaker 1: It's not true one-to-one personalization.
• 40:37
  Speaker 1: And they all pretty much say that not all.
• 40:40
  Speaker 1: But like, many people say, well, we tried it.
• 40:41
  Speaker 1: It was really expensive, it was really hard, it didn't work very well, and that bias, from that failure or that challenge will oftentimes influence people's plans going forward.
• 40:50
  Speaker 1: They'll say, well, we tried personalization four years ago and it was hard and it didn't work very well.
• 40:55
  Speaker 1: And I would say to you well, the reason you tried it is because the promise is actually incredible.
• 40:59
  Speaker 1: Try it again, because the technology has doubled and doubled and doubled and it's better and better and cheaper and cheaper.
• 41:05
  Speaker 1: So, coming back to PocketPod, it's just a form of personalization at the extreme level and we're going to see stuff like that pop up all over the place.
• 41:14
  Speaker 1: I wouldn't be surprised if you log into Netflix a year or two from now and it just starts generating content for you based on what you.
• 41:21
  Speaker 1: You talk to Netflix and it knows everything you liked and didn't like and you start to say I've got 25 minutes.
• 41:27
  Speaker 1: I want to watch something that's an action comedy and it just spins up something and there's all sorts of interesting issues around intellectual property and so forth, but that's where we're headed Very interesting.
• 41:39
  Speaker 2: I said it on last week's episode, I think for now and this could change, my personal stance is, I hope, to seek out creations that are made by humans and consume those.
• 41:49
  Speaker 2: So like human made podcasts, human made movies, TV shows, whatever.
• 41:53
  Speaker 2: But I will say, on the podcast front, I think there's an exciting use case where there just aren't podcasts around certain things.
• 42:00
  Speaker 2: Like I know, there's certain parts of history that I don't remember learning when I was younger and I'm like man.
• 42:05
  Speaker 2: It would be really interesting to get like an interesting version of what happened in Latin American history in these years that I could see myself using something like PocketPod for.
• 42:14
  Speaker 2: So I guess it just it kind of.
• 42:16
  Speaker 2: It depends on the situation.
• 42:18
  Speaker 1: Yeah, I mean the way I would describe that is you're separating, kind of the raw utility of the thing from the art.
• 42:23
  Speaker 1: And so when you want to consume art, it's because of the enjoyment, it's because of the feeling you get from it.
• 42:28
  Speaker 1: It's not just because you captured, you received that information, right, you got that information and that's it.
• 42:33
  Speaker 1: It's like saying, okay, well, let's say, I could get the same historical content from a 15-minute personalized pod about the movie Oppenheimer and what happened there.
• 42:45
  Speaker 1: I just want to consume the history, get the facts.
• 42:48
  Speaker 1: I want them given to me in a way that's relevant to my level of education, what I already know, which the personalized pod can do.
• 42:54
  Speaker 1: It doesn't give me things that I already am aware of, but it really gives me the pieces and bits and pieces, but I don't experience any of the art of that film, right.
• 43:02
  Speaker 1: And so, from the perspective of a consumer, if you're looking for raw utility and just achieving a thing, I think AI-generated stuff can be great.
• 43:11
  Speaker 1: But I do agree with you that I very much look forward to consuming, for enjoyment, true artistic works from other humans.
• 43:21
  Speaker 1: And we talked about AI music recently and it's the same thing there.
• 43:26
  Speaker 1: I think there's a tremendous utility there and there's going to be an explosion of use cases where you take advantage of AI-generated music for business and for communicating with friends and so forth.
• 43:35
  Speaker 1: But you're also going to have situations where is that really art?
• 43:39
  Speaker 1: Is that really something you would?
• 43:40
  Speaker 1: Would you listen to AI-generated music for fun?
• 43:42
  Speaker 1: I don't know.
• 43:43
  Speaker 1: Maybe some people eventually would.
• 43:45
  Speaker 1: There probably will be a genre called AI art or AI music, but maybe I'm old enough where I'm not going to particularly seek that out.
• 43:51
  Speaker 1: I'm looking for music created by humans.
• 43:55
  Speaker 2: Yeah, I think there'll be a market for those people, absolutely, and for people who want the Netflix on demand film made just for you, and I get that.
• 44:01
  Speaker 2: I can respect it.
• 44:02
  Speaker 2: But I love what you said separating utility from art, so I'm going to kind of I'm going to hold on to that as part of my belief system.
• 44:08
  Speaker 2: Thank you, amy.
• 44:09
  Speaker 2: In topic three today we are talking about AI and the middle class, not something we've spent a ton of time talking about on this podcast before.
• 44:18
  Speaker 2: David Autor, an economist at MIT, has changed his stance on AI's potential impact on the workforce.
• 44:24
  Speaker 2: Previously known for research showing how technology and trade have hurt middle-class incomes, autor now argues that generative AI could reverse this trend and benefit the middle class.
• 44:35
  Speaker 2: Autor believes modern AI is fundamentally different from past automation waves.
• 44:40
  Speaker 2: He contends that AI can assist more people, including those without college degrees, in performing valuable work currently done by expensive experts like doctors, lawyers and professors.
• 44:51
  Speaker 2: By enabling a broader segment of the population to take on higher-skilled work, ai could increase their earnings and lift more workers into the middle class.
• 45:00
  Speaker 2: Autor's perspective is notable given his background studying the negative effects of technology on workers' wages.
• 45:07
  Speaker 2: However, he explains that the facts have changed with the advent of powerful generative AI capabilities, prompting him to reevaluate AI's potential economic impact.
• 45:16
  Speaker 2: Amit, what are your thoughts on this?
• 45:19
  Speaker 2: When you shared this article with me and it was an interesting read, so I'd love to hear your thoughts on it.
• 45:25
  Speaker 1: Well, first of all, I love the idea of an academic who's as well known as celebrated in the field as this guy flipping his stance on anything, because I think the idea of going against what you've previously said is it's just so taboo in so many professions that people don't consider it.
• 45:45
  Speaker 1: They don't consider their own views being subject to change, and I think that's a giant mistake.
• 45:50
  Speaker 1: You know, I think it's really important to maintain as much mental agility as you can to approach the world based upon what you've learned rather than what you've said.
• 45:58
  Speaker 1: So, and there's a big difference there, right?
• 46:02
  Speaker 1: So I think that's one thing that's notable, and AI is clearly a remarkable enough technology that someone in a field that's obviously adjacent to AI, but not in AI, would flip his stance on this particular topic.
• 46:15
  Speaker 1: Now, as for the substance of the conversation, I think there's an interesting argument to be made that people in the middle could benefit from AI, subject to two conditions First being that what he is referring to holds true that people who have a middle level of skill can elevate their game because of AI.
• 46:35
  Speaker 1: So can a paralegal perform many of a lawyer's tasks?
• 46:40
  Speaker 1: Can a medical assistant perform more of a doctor's tasks?
• 46:44
  Speaker 1: Right, and if you can do more of the tasks theoretically, you can go up the value chain and you can achieve greater level of utility and greater economic gain through your role.
• 46:53
  Speaker 1: Now, at the same time, what we have to assume is that there is increased demand, because if there isn't, the question is who's competing?
• 47:02
  Speaker 1: Because even if you are able to do more if demand is fixed which I don't believe it is, but if it was, you have a problem with that equation because ultimately you just actually drive down price.
• 47:13
  Speaker 1: If demand is fixed and supply increases, that equilibrium pushes down price ultimately will suffer, and that's great from a consumer perspective, but terrible if you're a provider in the field, and so the middle class would be hurt in that scenario.
• 47:26
  Speaker 1: But the growth of the middle class in America in particular and most developed nations that I'm aware of, has gone on through major technological disruptions now for centuries, and particularly in the last 70 years with the IT revolution, there's been downstream effects for people well outside of the IT sector and a lot of that has gone from because of the increase in demand, essentially.
• 47:49
  Speaker 1: So we talk a lot in our AI briefings that the 10x increases are a really interesting order of magnitudes to study, certainly in computing and certainly in AI, but it's also interesting to study in the economic realm, and so if you study economic growth over a long period of time, I'm talking about pretty much all of human existence.
• 48:10
  Speaker 1: We say how long did it take us to get to $1 trillion in global GDP on a current basis meaning, inflation adjusted so equivalent to current 2024 US dollars?
• 48:20
  Speaker 1: How long did it take humanity to achieve a $1 trillion global output?
• 48:24
  Speaker 1: And that took about 1,700 years, from modern era right.
• 48:30
  Speaker 1: So it's really hundreds of thousands of years, millions of years, but from recorded history, basically, we know that roughly 1700, 1710 in that era is what economists estimate we achieve $1 trillion in global GDP in current US dollars.
• 48:45
  Speaker 1: It's essentially all of human history to get to a trillion.
• 48:48
  Speaker 1: Then the next question is how long did it take to get to the next order of magnitude?
• 48:52
  Speaker 1: 10 trillion, and it took about 250 years.
• 48:55
  Speaker 1: Through the Industrial Revolution, the steam engine, electricity, all these things that happened, we had a 10x increase over 200 plus years.
• 49:01
  Speaker 1: And then the question is well, what's happened since then?
• 49:04
  Speaker 1: Because right now we're at about $130 trillion global GDP, and when we got to 100, it was around 2014, I believe.
• 49:13
  Speaker 1: So from 1950 to 2014,.
• 49:15
  Speaker 1: 64, 65 years, compared to 250 years before that, for the preceding 10x increase in global GDP compared to all of human history, for the preceding one before that, thousands of years.
• 49:26
  Speaker 1: So something is definitely accelerating and what it is is demand increases when you have the ability to produce more value at lower cost.
• 49:35
  Speaker 1: Demand is not fixed.
• 49:36
  Speaker 1: You can't argue that demand is fixed.
• 49:38
  Speaker 1: If you look at that period of time.
• 49:39
  Speaker 1: Now you could say with AI, is it possible to have a quadrillion level of global GDP?
• 49:45
  Speaker 1: And I think the answer is yes.
• 49:48
  Speaker 1: I think that the opportunity exists to create new products, new services, and the demand will likely grow to consume those things.
• 49:55
  Speaker 1: So, coming back to your question and my point of view on it, if demand continues to grow, even at the rate that it's been for the last 67 years, I think there'll be a ton of opportunity for the middle class, because using AI assistive technologies will definitely level up skills.
• 50:12
  Speaker 1: Last thing I'll say about this is that there was a study last year that was done in partnership between Wharton and the Boston Consulting Group that looked at the different levels of people.
• 50:22
  Speaker 1: They took.
• 50:22
  Speaker 1: I think about a thousand employees at BCG and BCG is a well-known high-end management consultancy, but they, like anyone else, have a bell curve of talent, and so the people at the top of their performing curve and the people at the bottom, people in the middle, and what was most fascinating about their study wasn't just the productivity increases felt across the board from using, in their case, gpt-4.
• 50:45
  Speaker 1: Across the board there was an increase, but there was a larger increase in productivity in the middle of the bell curve, and so those people in the middle of the bell curve had a lift far greater than people at the high end of the bell curve, and so that leveling function is a lot of what David Autor is describing, I believe, and so I think there's an interesting case to be made there.
• 51:04
  Speaker 1: I think there's cause for optimism.
• 51:07
  Speaker 1: At the same time, there's other economists out there, like Anton Koronek from the University of Virginia's Darden School of Business, who we've spoken about previously and we've written about, who look at this from a different lens from what I understand, and he's looking at it more from the viewpoint of how quickly AI grows.
• 51:21
  Speaker 1: So if AI grows so rapidly that AGI is upon us, well then it's all bets are off, because if AGI can literally do all the work, then it doesn't matter how high you elevate your skill.
• 51:35
  Speaker 1: So in that scenario, you have a radical decrease in price, or aka, lots of unemployment.
• 51:41
  Speaker 1: Of course, the value being created by the AI is so great.
• 51:44
  Speaker 1: How do we tap into that to provide for society's needs?
• 51:47
  Speaker 1: That's a different question, right?
• 51:49
  Speaker 1: But ultimately, my point of view is that I think this piece is really interesting to read.
• 51:54
  Speaker 1: I'd encourage everyone to read this piece in the New York Times.
• 51:57
  Speaker 1: I believe We'll link to it in the show notes.
• 51:58
  Speaker 1: There are many layers to this conversation and ultimately, in your profession as an association, I think you need to read this article and then think about how this applies in your sector, whether that's nursing or engineering or life sciences.
• 52:13
  Speaker 1: You need to think about this critically as part of what we described in the earlier section.
• 52:17
  Speaker 1: How does this stuff apply to you and your profession?
• 52:21
  Speaker 2: I think one of the key points in this article is that less skilled quote unquote people will be able to do the same things that doctors or lawyers.
• 52:29
  Speaker 2: Not everything, but some of the same tasks that doctors and lawyers can do.
• 52:33
  Speaker 2: My question then to you, amit and really we're just speculating here because we don't know the answers but then what are lawyers and doctors doing?
• 52:40
  Speaker 2: Do you think there's just a whole new world of work that only they can do that maybe we can't even imagine now, just because we're not there yet.
• 52:48
  Speaker 1: Yeah, it's a great question.
• 52:53
  Speaker 1: I mean, and even independent of what we're talking about in terms of upskilling people in adjacent fields, like medical assistant to doctor, paralegal to lawyer, within the legal profession, just for lawyers or doctors, you have varying skill levels, right, and so you have varying levels of experience.
• 53:04
  Speaker 1: There's this whole process where people go into a profession like accounting or law or architecture, and there's this period where you're learning, where you're picking up the knowledge, where you know very little coming out of school in most of these fields and you learn a lot on the job to do in terms of thinking or arguing a case in front of a judge, but the junior lawyer doesn't have the opportunity to learn by virtue of doing the lower level tasks.
• 53:34
  Speaker 1: What do you do with those people?
• 53:35
  Speaker 1: And so the short answer is I have no idea, but I think that the rate of change in AI is the thing to really watch for, the fact that there's displacing technology we've been through that a bunch of times over the course of history but the speed at which AI is moving is the thing that's, I think, going to throw a lot of people off.
• 53:52
  Speaker 1: There are a lot of AI optimists who are saying, hey look, every time we've had this happen before we figured it out, we'll figure it out here, and that's the New Orleans tornado warning.
• 54:00
  Speaker 1: So hopefully.
• 54:01
  Speaker 1: Mallory and I will still be here to complete this podcast.
• 54:03
  Speaker 2: This might be the last episode of me.
• 54:09
  Speaker 1: It might the last episode of me, but it might be.
• 54:10
  Speaker 1: Yeah, hopefully not.
• 54:11
  Speaker 2: This is pretty fun, but we'll get AI to step in for us right, absolutely.
• 54:14
  Speaker 1: But over the course of time people have figured this out, society has figured this out.
• 54:17
  Speaker 1: Industries have evolved, and that's true, but that doesn't mean that there isn't a massive disruptive effect along the way.
• 54:24
  Speaker 1: So you look at these curves and say, hey, we've had all this growth, all this productivity, all this prosperity Along the way.
• 54:30
  Speaker 1: It's a jagged edge.
• 54:31
  Speaker 1: It doesn't mean it's a smooth curve.
• 54:33
  Speaker 1: And when you have a compression of timelines, with less time to react, less time to adapt which is what we need to do Technology doesn't need time to adapt.
• 54:40
  Speaker 1: Technology is just there.
• 54:41
  Speaker 1: But we need time to adapt.
• 54:47
  Speaker 1: Society needs time to adapt.
• 54:47
  Speaker 1: Your association needs time to adapt.
• 54:48
  Speaker 1: If you have a highly compressed time frame, it's much harder to predict what's going to happen and that disruptive effect potentially is more acute.
• 54:53
  Speaker 1: So to me, I think that's again coming back to what we talk about a lot.
• 54:57
  Speaker 1: Spend time on this, learn about it.
• 54:59
  Speaker 1: You're listening to us on this podcast or watching us on YouTube.
• 55:02
  Speaker 1: It means that you're taking time.
• 55:04
  Speaker 1: Go deeper, spend a little bit of time every week and learn this stuff, because the disruptive effect is not a question mark, it's really when is it going to happen to your field?
• 55:15
  Speaker 2: Do associations that impact specific professions have a responsibility to bridge the gap between their current member workforce capabilities and an AI-enhanced workforce?
• 55:26
  Speaker 1: I don't see how the answer cannot be a resounding yes.
• 55:29
  Speaker 1: I think that every association has the responsibility and the opportunity but the responsibility to serve their members, to serve their field and to serve society by helping their profession adapt to AI.
• 55:43
  Speaker 1: I don't think there's any question of that.
• 55:45
  Speaker 1: Whether you like it or you hate it, or you're somewhere in the middle, it's real, it's here, it's what's happening, and so you have to figure out how your workforce and your field is going to adapt, how you're going to serve the people that are in your field and the adjacent fields, and then the end consumers, whoever they are, that consume the services that your field produces.
• 56:03
  Speaker 1: You have to go do that.
• 56:06
  Speaker 2: Well, I think that's a great line to end today's episode on.
• 56:09
  Speaker 2: Thank you everyone for tuning in.
• 56:11
  Speaker 2: Amit and I made it through the storm and we will see you next week.
• 56:17
  Speaker 1: Thanks for tuning in to Sidecar Sync this week.
• 56:19
  Speaker 1: Looking to dive deeper?
• 56:20
  Speaker 1: Download your free copy of our new book Ascend Unlocking the Power of AI for Associations at ascendbookorg.
• 56:28
  Speaker 1: It's packed with insights to power your association's journey with AI.
• 56:32
  Speaker 1: And remember Sidecar is here with more resources, from webinars to boot camps, to help you stay ahead in the association world.
• 56:39
  Speaker 1: We'll catch you in the next episode.
• 56:41
  Speaker 1: Until then, keep learning, keep growing and keep disrupting.