 Hello everyone and welcome to today's webinar by Product School and how PM's can leverage AI. My name is Marc Rita Fakina and I've been a Senior Product Manager for over a decade. I've also been an entrepreneur myself having launched two successful startups both in the food industry. I've worked at press industries, especially e-commerce, food tech and food tech. I've also worked as a Senior Product Manager at companies like Amazon, huge, Tiva and Mercado Libre. Today's webinar we're going to be covering four specific topics. We'll start by delving into some of AI capabilities. We'll explore tools by that allow us to work with AI hand in hand. We'll then look at limitations of AI and finally we'll look at some of the ethical and privacy considerations. We'll start looking first at some of the AI capabilities which leave at the intersection of both ML machine learning and NLP natural language processing. AI itself is an enigmatic force. It has the ability to drive innovation through analysis, predictions and creation. It is a broad field that actually creates machines capable of intelligent behavior in this way mimicking human behavior. A subset of AI is actually machine learning, which is the one that focuses on algorithms that learn from data and improve their tasks over time. It is essential to spotlight the synergistic relationship between ML and NLP because they're both pillars that actually elevate AI itself from a computational tool into a predictive partner, especially for us in product management. The synergy of both machine learning and natural language processing actually opens up a plethora of capabilities and applications within product management. When we think about this, there are technologies that not only pave the way for more personalized user experience, but also equip product managers with actionable insights. AI capabilities include computer vision. This is technology that enables machines to actually interpret and understand the world from a visual perspective, visual information. It is employed today in various applications like facial recognition, autonomous vehicles and image analysis. Another important aspect is predictive analytics. By analyzing historical trends and data, AI actually can predict future trends, behaviors and events with great accuracy. This capability specifically for product managers allows us to anticipate market changes, user needs and potential issues enabling proactive rather than reactive strategies. Another big capability is decision making, which it provides data driven insights and recommendations that can be acted upon. This capability supports especially us as product managers in making informed decisions when it comes to product development, experimentation, marketing strategies and user experience improvements. The fourth one is creativity. AI has the potential to generate creative content from writing, marketing puppy to also helping us within ideation sessions. It helps us transform brainstorming sessions into ideas that can actually inspire human creativity. Another big portion is robotics and automation. This has been kind of upgrading itself into complex and sophisticated systems that are autonomous or semi-autonomous. And the last big portion is personalization and recommendation engines. AI has the ability to move us from a one size fit all into a tailored experience that actually improves each customer journeys to curate content and products that actually adapt to their journey. This increases satisfaction and engagement and has the capability to impact user retention and conversion rate. Other capabilities include sentiment analysis, speech recognition and chatbots and virtual assistants, which have become part of our daily activities. As we can see the convergence of capabilities within AI enables product managers to not only enhance existing products, but also innovate new solutions that meet our evolving customer needs. Now, as we think about AI capabilities, there's a big structure that actually feeds into it. As we mentioned a little bit earlier, it all starts with data collection, which is the ability to gather data from diverse datasets for in-depth analysis. This can include user behavior logs, social media interactions, purchase history. The third, the second big portion is data preparation. Once we've ingested all this information, we need to refine this to provide quality and accurate data that can help us train our models. And this includes actual data cloning, handling missing values and data normalization. The third portion is machine learning analysis, which is actually using all this information to analyze data patterns used with advanced algorithms. And this actually includes training predictive models, cross-validation and feature engineering. The outcome is prediction outputs. This are generating actionable foresight and model insights, which as product managers could include estimating sales volume, customer turn rate, or even next month's trend forecast. As we look into the before and after of product management, it has been broadly and abruptly changed by AI capabilities. Product management a few years ago was marked by limited data analysis. We heavily relied on decisions that were based on limited data that forces to make rapid decisions sometimes based on gut feeling. Another big problem was this little feedback loop, which was inadvertently followed by infrequent ingestion and collection of feedback. Which because it was so infrequent, it was hard to act upon. The other port is that a lot of the results that were driving product development were reactive. We were making a lot of changes based on market trends or competitors moves. Also generalized features and finally static reports, which allowed us to look at data from a single optic. With AI, we've revolutionized each one of this. We now have the ability for gathering dip data insights. We have real-time feedback, allowing us to really learn about customers on a daily basis and be able to get instant feedback incorporation into our product. We've moved from reactive into proactive. We're able to actually anticipate market trends, user needs, and also it allows us to stay ahead of the curve. A big component of the AI capabilities that we mentioned earlier has been personalization. It has allowed us to tailor and customize each one of our user experiences to provide a unique journey for each one of our customers. And finally, our data is dynamic. We're able to use real-time data with visualizations that are actually actionable insights. Now, when we think about product management, there's many stages in product management. And at this point, we're actually going to be shifting to focusing on co-working with AI. AI itself as a collaborative approach that integrates AI into each one of the stages of product management. And when we think about the stages, it includes market research, ideation, design and prototyping, development, testing, lunch, user feedback, and iteration. In the instance of market research, AI actually helps us sift through vast amounts of data and detect actually merging trends. During the ideation phase, AI can actually assist us in generating novel ideas and validating concepts, and we'll explore a little bit more on this one in some future slides. In design and prototyping, AI can actually simulate user interactions and gather preliminary feedback. Tools have actually able to provide with a single input an entire presentation and entire wireframe, making it easier to transform ideas into actual concepts. With development itself, it's all augmented by AI's capabilities, especially in optimizing code and predicting potential issues with some of the code that we've written. When we think about testing, AI actually allows us to process real-time input and feedback from our users. It ensures that we continue to be agile in our response. Upon launch, AI actually helps us monitor a lot of these products that are in production. It gives us the ability to continue to capture feedback from our users using natural language and actually distil insights from customers' conversations. This is complemented by user feedback. And finally, iteration. It allows us to ensure that we're constantly optimizing our products based on decisions and making sure that our products are continuously improving and evolving over time. Each one of the stages actually poses massive opportunity for us to use AI to augment our capabilities as product managers ourselves. By using AI to collaborate in each one of these stages, we're actually able to augment our capabilities, streamline processes, and bring us closer to creating products that truly resonate with our customers. It is worth thinking about in this section, how can AI actually elevate my role and even where can I take my next project? Now, as we think about co-working with AI, one of the tools that I really enjoy using is chat GBT. And this is going to be now one of the most practical portions of this webinar. When we think about chat GBT, we're generally thinking about myriad of solutions that are actually enabled by this platform. But I like to use this specific one in two main focuses. The first one is responding to customer inquiries. And the second one is helping me brainstorm new features. Let's explore the first one. When we think about responding to customer inquiries, let's think about an scenario. A customer actually messages. They are very confused about two specific product models on our e-commerce platform. We can actually use chat GBT to upload all the input from both of our products and craft a detail prompt that enables us to compare between both models and provide input for our customers. When we think about doing comparisons and those are some tips for using comparisons within chat GBT, let's make sure that we include detail attributes. We explain the purpose of the comparison. As we know in product management, we work across multiple streams. We have buying guides. We have frequently asked questions and we interact directly with customers. So making sure we specify what we're utilizing this response for will leverage even further our AI, our prompts, sorry. We need to also make sure to provide information about our target audience. As we know, a lot of our responses should be catered to our audience. And we should also be able to indicate the style and the tone that we want to convey that information. When we think about potential customers, we can have professional audiences, which might require a formal tone with data driven insights. Now let's explore a second scenario that we can actually tap with the help of chat GBT. And this is brainstorming ideas for new features. A scenario for this is say that you're looking for new features that actually are enhancing the user experience on an e-commerce website and we want to develop new features to do so. For the prompt, we want to make specifically clear what kind of user is going to be producing the output. In this case, we use as a lead product manager. We also want to make sure that when we're brainstorming new ideas, we set clear objectives. Sometimes it is about novelty, but sometimes the main goal could be increasing sales, increasing conversion. And each one of those goals will actually define the outcome or the potential solutions that chat GBT will produce. We should also specify technologies if those are important. For example, if we're creating something that has to do with the new Vision Pro using augmented reality, virtual reality, like all of this decision should be informed so that we produce more accurate results. We should also provide inputs about competitors. If there is anything that, you know, a new competitor has lunch or that is considered benchmark in the industry, it is essential to include it here. As before, defining our target audience or demographics is of great importance and very important as well in this aspect of brainstorming. Oftentimes, as product managers, we have a lot of limitations, whether that come from a budget perspective, a technological limitation or itself a team or capacity aspect. Mentioning those will also allow to brainstorm ideas that adjust to those outcomes. And finally, as an important output, it is essential to request practical examples that can give us an idea of how the solutions fit into real life. Another tool that I love working with is Essence.io. This tool actually has the ability to synthesize user feedback across multiple channels into actionable insights. In this case, we input information from one of our products. This includes feedback from, say, the app, storm, and multiple resources where we are able to gather feedback from NPS surveys or any additional surveys that are performed by the platform or the product or the feature or the company itself. All this information once it's uploaded, we can see that it can provide insights on kind of like a SWOT analysis on what things are working, what things aren't working, what are some of the requests that customers are asking for change, what are some of the defects and issues or bugs that our platform could potentially have. Another platform that I work with on a daily basis is Perplexity. And a little bit different to how chat activity works, which has similar capabilities. This one is focused on providing answers that have a lot of research. I use this specifically to quickly get up to speed on new domains or to validate my assumptions during the ideation phase. Another tool that I really enjoy using is the Giga Brain. And this one is based on Reddit community. And I really appreciate this one because we've all gone into Reddit to kind of find information about our products or similar products. And this tool actually condenses everything into a very easy to read format. An example of an input or a prompt that I used is analyze user sentiment and feedback on Barn and what are some of the capabilities most users are talking about. Its response actually segmented the information into consolidated user sentiment, followed by direct feedback and some of the capabilities that have been launched by Barn and what they actually translate into. Finally, the other tool that I use a lot is Fathom. And this one is Notetaker that helps to transcribe and summarize meetings. And as we can see in the screen, it actually moves into takeaways, topics and next steps. As a product manager, I'm doing a lot of testing and user interviews on a weekly basis. Having a tool like this is powerful because it allows me to actually focus on asking the right questions versus trying to notetake. As it is important to explore AI capabilities, it's also important to look at the other side of the coin which are AI limitations. One of its biggest is inherent biases, a significant limitation of AI that naturally comes from the reliance on data. The adage garbage in garbage outcomes, particularly into place and relevant here. AI systems can inadvertently perpetuate and even amplify biases present in their training data. We can continue to reduce and mitigate this by insurance diverse data sets. This means that as we actively source for data, we need to include a wide spectrum of inputs that not only contemplate diverse information, but also demographics to minimize blind spots. The second one is algorithmic audits, which means continuously monitoring and examining the information and the algorithms by diverse teams. And finally, monitoring and updating as we continue to feed information, we need to ensure that the new data and the social changes are also part of this information. Another big aspect of the AI limitation is the complexity and explainability aspect. Another critical limitation is the black box nature of many AI models, which ultimately means that a lot of the decision making processes are not transparent and they're not easy to explain. The lack of explainability actually is a barrier to trust and to accountability, especially in high stake decision making. As we transition into explainable AI models, we need to develop algorithms that are transparent and provide understandable decisions. This are crucial for broader acceptance and ethical applications. Additional include the lack of contextual understanding as there's advancing NLP or natural language processing at a higher rate than ever. There's still limitations in how we are able to, they are able to interpret complex information like scenarios or ambiguous settings. And this can result unfortunately in misinterpretations or even overly literal representations or responses. This underscores the importance of human interaction for overseeing the applications. Another important aspect or limitation is the over-reliance of AI. As new products come to market, new users are transitioning into applying this tools into their daily activities. However, it is important to understand that becoming overly reliant on AI can actually decrease the, can stiffen creativity and critical thinking. And finally, the high resource requirements to actually power AI. It is required extensive resources and expertise, and this becomes particularly difficult for smaller teams or organizations that might not necessarily have the right structure or the technical capabilities. There are many ways to actually mitigate the impact of AI limitations, introducing new strategies that can move us away and navigate those limitations. An example of this includes the actual continuous integration of human insights to monitor and catch errors. Introducing ethical frameworks to guide to guide AI, specifically in product management to audit AI systems and decisions to invest in training to understand AI capabilities and its limitations as we progress this tools and specially to promote literacy for stakeholders. To continue to drive infrastructure that adapts to high data handling practices that is responsible, that is agile and scalable. And finally, in community collaboration to ensure that we engage with AI audiences that are able to share insights, challenges and solutions that come along with the technology. The final section that we'll cover today is the ethical and privacy considerations that come along with AI. It's important to divide them into four specific pillars. The first one has to do with data privacy. As we collect and use data from our users, we need to provide resources for them to understand how, why and when we're collecting that information. And we need to commit to secure protocols that are able to confidentially handle data in a secure manner. Second, biases and fairness. We've touched upon this earlier on in the presentation, but it is also ensuring that we have an AI, an equitable AI with diverse data sets and constant evolution. Transparency and accountability. This specifically refers to AI decisions and making sure that we move away from that black box into a box that provides insight that is clear, justifiable. And trustworthy. And finally, user control and management for content. We need to provide our users the tools to be able to have clear choices on how they manage and use their data. We've gone through a lot of information in this webinar, but most importantly, AI has the ability to not only serve as a tool, but also help in decision making and creativity and predictive analytics, especially as product managers. We've seen a tool showcase that covers chat, GBT, assets, perplexity and other ones that are able to leverage our product management role and use those tools to complement our role. We've also explored limitations of AI. The powerful collaborator. AI is the usage and ethical considerations we need to continue to drive as we use AI. And finally, how it requires a mindful adoption that balances both its strengths and ethical uses. Thank you so much for joining today. And I hope this webinar wasn't insightful on how pms can continue to lever to AI. My name is Margarita and it was a pleasure. Thank you.