okay let's get started folks it is a jam-packed day so I'm gonna I'm gonna get started here I'm not gonna wait for five after ten we're gonna start off here let me just share my slide okay so today we're gonna start off with a homework review then we're gonna discuss pitch deliverables and rubric and then talk about AI and health education and then more about AI startups because that's what your final project deals with and then research planning and funding demo we will definitely not have time I'll be sharing some YouTube videos that are required watching this week and then we are going to have some group time at the end of class today so thank you for everyone who finished the homework especially the interest form on time I have put the groups hopefully you guys have seen that these are the groups I may have switched from when you guys saw it last just as more people are signing up and I want to make sure that the groups are sort of equal so just something to keep in mind in terms of the homework so I'll stop screen sharing here for just a second for discover pressing social issues I got a lot of great answers a lot of detailed answers a lot of you use chat GPT which is fantastic like again the point of this course is always to explore new tools so make sure to try out new things if you already haven't some of you used deep-sea you use Google Gemini a lot of you use Microsoft Copilot Microsoft Copilot for me is my favorite because there's no knowledge cutoff for chat GPT you need to press that search button for it to be connected to internet search so that's a personal preference and just because it is connected to the internet search but I'm not arguing that Copilot's better than any of them if you watch the optional YouTube video which many of you did and many of you also tried with the same tools and some different tools you saw that there was no one clear winner I mean deep-seek was good for a lot of these things but again no one clear winner depending on what you're doing you might be one tool might be better than another and then what kinds of prompts so this is the part I wanted to go over obviously we didn't talk about prompt engineering in depth last time like how do you go about these questions and that's because that's something that I wanted you to explore and you to think about and so I'm gonna share some of the ones that I really really liked so so one person asked what are some pressing social issues and topic of climate change and pollution and so that was sort of where everyone generally started what are pressing social issues and what's interesting about this question is that this person said don't tell me the solutions in parentheses and then the second thing that you prompted was put it into a short paragraph and make it readable and then the third thing was recheck this so that I thought was very interesting and then some other people went into like very sequential questions and this is exactly the way prompt engineering works it's super nice if you go this way because chat GPT or whatever other tool you use it does have context it does know the previous questions you have asked so someone asked what are the important social issues with environment and health can you be more specific more on air pollution and respiratory disease more types of contamination and waste exposure more examples more on environmental stress and mental health how does environmental stress affect the human body in daily life right so just these series of questions and they're not even full questions they're literally just phrases honestly and it works and someone was even more specific to this course someone said I first prompted I'm interested in building an AI startup in the field of economics give me a list of problems that currently exist or are most pressing in that field and so that was all very cool stuff I loved all the observations you guys gave of the different tools it was I thought very interesting some people like chat GPT's UI then you guys told me about like what specialized in what things that you tried math problems for example in copilot but we're getting them wrong and deep-seek which is more specialized that also was getting it wrong so it was cool that you guys tried this for many different things the one that kind of made me laugh was and I'm kind of confused and I hope this is a joke someone wrote Claude he's so nice I love him so I don't know what's going on there but let's not get too attached to our chat bots please by the way if you want to turn on your cameras it's optional but it always helps me so that I know that what you guys are understanding or not understanding yeah any questions about the previous week's homework you can you're most welcome to use other AI chatbots there are so many right out there I generally stick to the the larger companies or the more talked about ones I myself you know I think I told you guys I'm a software engineer at Microsoft so a lot of the chatbots that I'm using always are Microsoft based so just something to keep in mind having said that you know there are obviously a lot of good ones out there okay now let us go over the pitch presentation I'm sure you guys are kind of curious about that so let me share my screen again okay AI for social impact pitch right we know that there is a demo date that is coming up in about a month's time what what are the expectations here right and I put you guys in groups right what was that for what was for the social impact pitch so your team is gonna comprise of four to five students that's what all the teams right now look like if someone else you know signs up on the form because there's quite a few of you still haven't signed up then the teams might change but I want to try not to change them now that class has started and we're gonna be meeting in groups today so there are so many deliverables for demo day but we're not gonna wait until last minute so rough drafts will be constantly do kind of like every week and the idea is that you are using different AI tools that we discuss in class to kind of come up with these things now something to keep in mind is even though this class is all about use AI that doesn't mean that oh use AI and keep your human brain aside and don't really even think about what you're doing and why you're doing right AI is supposed to augment human intelligence and not you know replace it because as we have seen time and again AI isn't always right it doesn't always have the best ideas and I would argue it also doesn't have that novelty right AI is trained on other things and it sort of mimics and mirrors that even generative AI I would argue is not not that novel in some ways so in terms of rough draft of deliverables obviously for your startup which has to be in one of the categories that I've assigned to you it has to be a startup that uses AI we need to come up with a company name and this is due week four we need to come up with the logo again we'll be using AI's help to come up with these things and then we're gonna create Instagram posts marketing related Instagram posts because that's kind of what like when you're trying to market your startup you do and then the one-page pitch that contains all this above information and other relevant information and I'll talk about that in just a second here as well a huge as a YouTube video that answers the question why should we fund your startup right that whole idea of pitch presentations essentially is like I have a startup and I want money and who's gonna get this money right so that's sort of the goal here and so there are some suggestions you guys can read through it more but but the idea is that it's the company name should be relevant to the AI solution it's nice if it's memorable or funny logo again should be creative Instagram post I've given some examples and we'll talk about this more in the coming weeks weeks and then yeah one-page pitch YouTube video things like that now the demo day rubric so again the pitch presentation is just gonna be one page how it's gonna work is that some of the teams will be in the breakout room for half the session some of the members of the team and then the other half of the team is going to be presenting and then we're gonna swap so that's sort of what I'm in this envisaging for demo day again we talked about the AI powered pitch so now what should the actual one-page pitch contain well company name business logo the team members who they are maybe what their expertise is problem statement AI solution that you are proposing the market size which means how many people are gonna be using your product how many people can you target for your product how do you stand apart from the competition there might be other startups AI or non AI related that are doing the same thing you're doing why are you better sort of and three Instagram posts and a YouTube video and you will how to submit this and Google Classroom again none of this is due today or even next week but I just want to give you an idea on what this all is obviously you guys are not creating a real startup this is sort of a fake startup but as if you were creating it you're gonna use all these AI tools to help you you'll be judged on creativity and aesthetics that is really important kind of your understanding of the material we have discussed in the class and using the material we have discussed in class so you need to know the technical aspects you need to use AI technology and then you need to clearly state what the problem is how you're gonna solve it the competition estimate customers where with the money be used yeah this will be posted on Google Classroom end of class today any quick questions okay Richard if you want to start sharing the slides now and I'm going to put you guys in breakout rooms in the meanwhile you guys will be in the main room right now but you'll be in breakout rooms at the end of the class okay Richard yeah thanks I'll share release all right so then let's not explore how AI is changing the healthcare industry so the medical field is becoming like increasingly digitalized thanks to innovations like electronic health records EHRs wearable devices new regulatory frameworks like HIPAA and high-tech and these advancements have led to like a huge increase in health care data this has made artificial intelligence become an increasingly important tool for analyzing that data and AI can use the data to diagnose diseases personalized treatments a drug development and more as these models develop AI becomes increasingly better for assisting in these tasks as well in diagnosis and treatment AI tools can detect early signs of disease and recommend treatment options by analyzing data like patient features such as age and gender medical images like x-rays and MRIs and clinical notes from electronic health records for example AI powered imaging systems can detect patterns and images that might indicate early-stage diseases and natural language processing algorithms can review unstructured clinical notes to gain insights and support evidence-based decision-making and then this with integration into things like clinical decision support systems also helps physicians receive real-time data-backed recommendations for patient care so this is a case study of how image recognition AI is used in breast cancer diagnosis Google Health's AI system analyzes mammograms which are x-ray images of the breast to detect early signs of cancer and by training on large data sets of mammogram images the system can identify possible signs of disease and images that might normally be missed by like a human eye thus enhancing the diagnostic accuracy and reducing the workload of radiologists aside from mammograms other like image recognition techniques can also be applied to other contexts like MRIs for prostate cancer detection and more so yeah another application of AI in healthcare is natural language processing NLP and clinical decision support systems as kind of mentioned so a lot of like the data generated in healthcare is mostly unstructured data which means it doesn't have like a format there's no like input fields it's not like numbers and computers can't really just like easily understand unstructured data but NLP algorithms can process like the unstructured like clinical notes to understand details about patient symptoms history and other factors and then these details are then used in like a clinical decision support system to categorize symptoms and suggest treatment plans appropriate medication dosages and more and then with techniques like named entity recognition NER these systems can just accurately identify important clinical elements ensuring that the recommendations are both precise and evidence-based so this is a like example of how reinforcement learning is used in treating sepsis sepsis is a life-threatening condition where the immune system overreacts to infection leading to widespread inflammation and organ damage in sepsis the lung and kidney are two organs that commonly fail and clinicians often support lung function with mechanical ventilators and vasopressors and use intravenous fluids to support kidney performance currently there's no consensus as to exactly when and what amount should be administered to best treat this disease but reinforcement learning models can optimize treatment by recommending like dosages and timing for intervention and the rewards that are used to train the RL model can be patient survival but doctors often choose to optimize the function of specific organs like the lung and kidney so yeah AI also has great uses in drug discovery one example is the in silico medicine where they use generative AI to design drugs for rare lung diseases like fibrosis completing the early stage development in far less time and with way less cost as well this uses AI to identify a target which are molecules typically proteins associated with the disease then design molecules targeting that protein the molecules generated will meet a specific criteria such as metabolic stability just how fast the drug is metabolized and in that case that would ensure the drug like would remain effective for a certain period of time and then by predicting molecular interactions and biological activity and I can accelerate the discovery and optimization of compounds with desired pharmacological properties and then another application of AI in health care is personalized or precision medicine this is where treatment plans are tailored to individual patients based on information like genetic makeup clinical history lifestyle environmental factors and more so these AI systems analyze genomics just DNA and proteomics protein data to identify biomarkers that is specific molecules that signal disease and predict how a patient might respond to a particular treatment option the ultimate goal is to ensure that each patient receives a customized treatment plan with maximal effectiveness and minimize side effects for them and just improve overall patient outcomes despite all of the beneficial applications of AI in health care there are some limitations and challenges firstly AI models are only as good as the data that they're trained on so issues like inadequate or non representative data can lead to biases and reduce overall accuracy and generalizability of the models which is something that you don't want to happen and then also due to issues like interoperability and user trust the integration of AI into existing clinical workflows hasn't been like the most smooth so far and then regulatory and ethical considerations are also important where a patient privacy data security and the ethical use of AI are all concerns that need to be addressed and finally AI algorithms need external validation and reproductibility studies to ensure that they perform reliability in real-world settings before they can be used all right yeah next let's see how AI is used in education so AI can make the learning experience more personalized and efficient AI technologies are being applied to enhance learning teaching and even administrative tasks for instance natural language processing allows for intelligent tutoring systems that analyze student input be it text or speech to then detect their understanding identify knowledge gaps and provide personalized feedback for example a student might analyze a system might analyze a student's essay and offer suggestions on grammar style and argumentation additionally generative AI can also create content including lesson plans quizzes and study guides varying in difficulty and customized to a student's needs it can also work through a problem step-by-step to help students understand how to solve that problem so yeah one example of AI for personalized Mingo for learning is con Mingo by Khan Academy has anyone like used con Mingo before or you hmm okay well yeah well con Mingo is an AI powered virtual tutor that can provide one-on-one tutoring by identifying knowledge gaps and providing personalized feedback it guides students through problem-solving processes without simply just handing out answers it can also review essays and even suggest additional practice questions to help for a test preparation while con Mingo isn't an replacement for a human tutor it is a powerful tool for students as it can give them immediate and individualized assistance so another example of usage in lesson planning are tools like the fit if it can generate adapted reading passages summaries and key vocabulary words that are aligned with the students reading level and curriculum standards it also creates multiple choice short answer questions and open-ended prompts to support learning objectives one case of this being used is like in a New York district there were two sixth grade teachers who used the fit with Canva to design a lesson on ancient Greek bases kind of in and yeah these AI driven tools are helpful as they can free up valuable teacher time allowing educators to focus more on like in-class learning and more personalized support so grading is another area that AI can help with by using optical character recognition OCR along with NLP AI systems can convert handwritten answers into digital text and evaluate them against like a predefined rubric whether it's like objective answers such as multiple choice questions or student responses like essays and math solutions these systems can provide immediate feedback and detailed analysis of student progress this can speed up the grading process which is incredibly time-consuming for teachers especially as like they have look for the low as they have larger classes and for question types such as essays or handwritten answers and yeah these systems are also helpful to students as it can give them instant feedback on their performance, allowing them to know what they need and improve on. For educational administration, predictive AI models can predict course demand while agentic AI can provide personalized schedule recommendations based on a student's academic history and goals. Some of these chatbots might be powered by Model Context Protocol, MCP, which is essentially connecting the chatbot with external servers and then calling external tools to execute their commands. An example of this might be when a chatbot uses a Python script to solve a complex math problem it normally can't. And with MCP, an AI system might be able to call one tool to learn about classes, and then another tool, maybe even another AI tool, to review a student's transcript and understand what the student excels in, what their goals are, such, and then suggest the best classes for them to take. These tools can also answer questions that you might normally ask to a human counselor. And if students were to ask some more generalized questions to AI instead of counselors, along with other tasks that AI can assist in, then the administrative burden on staff would be greatly reduced, and students can also receive faster and more personalized support. All right, next, let's go over a little about AI startups, since you'll be creating a pitch for one as your project. So an AI startup is just a company that develops new AI technologies or uses existing ones to solve real-world problems and improve processes in various industries. There are a lot of different AI startups out there. Yeah. So for each AI startup, there are a few basic steps that I'll go through. So first, a problem needs to be identified, and then this might be investigated for market demand, and you might think about how AI can offer a solution to determine the feasibility of that problem. And then relevant data would be gathered through various techniques. Then the data would be used to train an AI model to solve that problem. And finally, the model is tested, released, and continuously refined to improve its performance, since it's not just like static once it's out there. You'll want to improve it to make sure it can keep up. And then building an actual AI startup would involve many costs across different areas, such as you'll need hardware and cloud computing, which are just for running any sort of like AI algorithm, like especially like more advanced ones. And then for data acquisition and preparation, since again, high-quality data is necessary for any successful AI model, and high-quality is like unbiased, relevant, and all of that. And then you'll need talent acquisition, which is like recruiting skilled data scientists, engineers, and AI specialists to like help run your startup, get it going. And then software licenses and developmental tools, which are just the necessary tools and platforms to develop the AI solution. And then a lot of costs for research and development, since you'll need to continuously improve to stay ahead and remain relevant in the rapidly changing field. And then for legal and regulatory compliance, which is just ensuring that all of the data and processes meet industry standards and regulations, which you might need to like update as time goes on as well. Then yeah, to just get an idea of some current AI startups, let's just look at some successful examples. First one I'm sure you all have heard of and know is OpenAI, which focuses on AI infrastructure and model development. Its value is like 157 billion. There's 21.9 billion funding and 172.8 million monthly visitors. While these are just like numbers and aren't all that important, it gives you an idea of how like, how much each startup is worth, the investment that's needed, and its popularity, so current usages, such. And one thing to note about these numbers also is it's not like the most current, but rather it's for like 2024-ish. And yeah, the same applies for all of the statistics on this slide. I think the most recent value of OpenAI has gone to 300 billion. So you can imagine that the actual value should be a little bit higher than what's displayed here. Yeah, the next example, PathAI, it's more similar to what you'll be doing since it's focused on solving a problem. In their case, it's improving pathology, the study of disease and precision diagnosis. It still has a high value of 780 million, 576 million funding and 48,000 monthly visitors. And the last example we'll look at is DeepL, which is another example of using AI to complete a task. They have made a language translation tool that uses neuromachine translation to translate text from one language to another. Value, 2 billion funding, 400 million monthly visitors, 52.4 million. And yeah, these examples just show the applications of AI, its growth and investments for this industry. And as AI continues to improve, we can expect more startups to emerge and just improve as well. Some of them might be made by people like you who are learning AI and might apply it in future careers. So yeah, we've already started the research process in last week assignment. So now let's go more in depth and talk about more AI tools that can help with research. First to kind of define research, it's the process of studying a topic or problem to gain new knowledge and develop solutions. It involves the collection, analysis and interpretation of data, which is important for advancing AI technologies among other things and addressing challenges in various fields. Research is required to develop and improve AI. The AI tools can also help with the research process by in ways like looking through information and data and summarizing it for the researcher. Another thing that's needed in making your own startup is planning, which is the process of setting clear objectives, developing strategies and managing risks while continuously adapting to changing conditions. Good planning is important for successful AI development and deployment and going both ways, AI can help you with the planning process by analyzing data, predicting trends, assessing risks, simulating ideas and helping to make informed decisions. So whether you're building an AI startup or working on a different project like Science Fair, good planning can help ensure that you're successful in both of those areas. Like Science Fair, good planning can help ensure that your ideas are doable and sustainable and AI tools can help simulate startup ideas for your category, be it like health, education or whatever else. And then funding is the financial support for any project. As like we've seen, AI has many associated costs. So funding is very important for any sort of AI research, development or deployment. AI can also optimize the funding process again by analyzing data from multiple sources such as historical funding records, project proposals, market trends, scientific publications and news articles. And other sources maybe to yeah, then identify promising projects, assess risks and predict potential returns on investment. We'll be using AI specifically chatbots to do research for funding that is to determine what we need money for. And then yeah, next turn it to Ms. Serpia to go over the AI tools. Okay. Awesome. First of all, before we move forward, does anyone have any questions on anything we have covered so far? Also, while I am talking through the next few slides, if you can change your name to whatever this, let me just share. Whatever this sheet, whatever your name is in this sheet, then that would be great because I have mostly put you all out in breakout rooms, but there are a few I still have not because I couldn't kind of tell who you were. Okay. Let us talk about AI tools. So since we're now going to be formally starting teamwork and stuff, we want to make sure we are working efficiently. So the first couple of tools that I am going to discuss today are AI tools to boost productivity and efficiency, especially in the context of meetings. So Otter AI is an AI tool that converts speech to text to transcribe meetings. It helps you capture every detail of conversations. It collaborates in real time during meetings. And so the nice thing about this is you can connect Otter AI with whatever you use normally for video calls, be it Google Chat, Google Meetings, Zoom, anything else, Teams. Obviously, most of us usually work with Zoom, so we can connect it with that. And what it does is it joins the Zoom call as a separate sort of bot, if you will. So it will say, like, And it will join as another individual, kind of, and it's going to listen in on the meeting. It's going to provide this transcription. It can provide summaries. You can talk with the chat bot and say, hey, what was covered in the meeting? Like, for example, if we were to connect it with this meeting, which I haven't, then it might say, like, oh, you can ask, like, what was covered in the meeting? And it will say, like, then it might say, like, oh, you can ask, like, so what was Richard saying about applications of AI in health? So if you, let's say, wanted to review this meeting afterwards, you can do it very quickly. It, yeah, allows for collaboration, and it works on many different platforms. Now, one thing I will say is I do have a demo video for the things we're going to talk about. I'm going to be posting it, like, today and tomorrow, essentially. With AutoAI, you have to be careful. First of all, don't, if you can help it, don't use it with a school email address or work email address or anything like that. Use it with your own personal email address. I have read so many reviews online where it's kind of hard to disconnect AutoAI, and so it joins in on meetings that you haven't given it permission to, and it informs everyone in the meeting that, hey, I'm using AutoAI and stuff like that, and so we don't want that. So also don't connect it with your calendar. Don't automatically connect it with all the meetings. So I can, I'll show you in the video how that all works, but I can just show you the website quickly. AutoAI, this is how it looks like, and then you can sign up for free. Obviously, for these things, many of them have different pricing, but we'll always just use the free versions for our class, and so you can see that there are some limits. But yeah, it's pretty cool. I think it can be super useful for even like school meetings and stuff, but obviously you would need your teacher's permission of, can I enable this? Because all these things are obviously listening in, they're recording things, and in many states in the US, you need permission before recording. I see a couple of questions. Okay, I see. Shoya, your name is not on the sheet. Can you just fill out the form again? I could not, I think, find your name or something. Anson, sorry about that. I will remove you from one of the groups. Okay, so that's AutoAI. No, none of these are required, but this can be helpful. The idea is for you to learn new AI tools, especially non-Chat GPT tools. But again, if you have issues about privacy and concerns and all those kinds of things, then I would say, maybe don't use it. In fact, the version of Zoom I'm using right now for these classes, this is obviously the business version. So Zoom itself also has its own AI, like transcription and summaries also, I believe. But this is, if you don't have the paid version of Zoom, then AutoAI is useful. A lot of times, like when I'm doing admissions consulting with my students, they're always like, oh, can I enable AutoAI so that I know I can look back in and watch the recording? But if I don't have time to watch the recording, I can just look at AutoAI notes of like, what did you tell me to fix on my essays, right, for college admissions? Yeah, yeah, Anson, I can do that. Do you log in with school email, own email? I would say own email. Don't mess with the school email because if it accidentally gets joined with your calendar or anything, it's gonna send invites to everyone, say, oh yeah, I'm gonna record this meeting, and that's gonna be really awkward. And people have had issues kind of disconnecting, so that's why just be very careful with this tool. The other one that I wanted to share was CRISP. So this one is also similar to AutoAI. It has some of the similar features of transcription and things like that, but most people have said this online, and this was my experience as well, that the transcription feature of AutoAI is most definitely better. It can come up with action items better than CRISP does, at least when I tried it out. But the thing that this one is really, really good for is that it removes the background noise from calls. And you'll see this in the YouTube videos that I assign you, is that I literally am tapping my table, I'm doing all these sorts of weird things, just banging the table, honestly, just to see if those noises come through. When I don't have CRISP enabled, all of those noises are coming through from my webcam microphone. However, if I have CRISP enabled, none of those extra noises are coming through. And so I would, if you want to test out CRISP, this is really good. It's really nice. I actually have CRISP enabled right now for this meeting. So I will get transcriptions and I will get stuff, and I can share that with you guys later. You can also send it over Slack or other different sorts of things so you can share it with other people. But yeah, it is, I think, really good. If you want to try it out, don't try it out with any fancy microphones you have. Try it with really bad microphones, like your webcam microphone or something, because fancy microphones anyway filter out this extra noise. And so then the way this works is that basically CRISP software, it processes after your microphone listens to you, it processes over that sound, and then it erases that background noise. This one does not join as a bot or anything like that in the Zoom call, so there's really no way of knowing that I have enabled CRISP or not. Yeah, how do you enable CRISP in Zoom? So the video is going to cover that. Essentially, we download the CRISP app, and then it will ask you to pick the CRISP microphone, which is a virtual microphone. It's not a real microphone, but basically you're just saying, yeah, CRISP process over my things. And there are some couple of steps that you have to take in Zoom, such as changing to low background noise for CRISP to really do its magic. But this one is more of like a desktop app. Yeah, but you'll see all of this in YouTube that I'm gonna upload those videos soon. So you'll see how all that works, and I would encourage you to try. If you're not comfortable with AutoAI, fine, try at least CRISP. CRISP is, I would say, still pretty good. And for background noise, it's amazing. Rohan, yeah, if you can fill out the survey, please. Yeah, thank you. And that goes for anyone else. Please fill out surveys. There's a bunch of you not in breakout rooms, so. So that was meetings. Now, research. Obviously, we need to continue our research for especially today's assignment, which I'll talk about in just a few minutes. So the first research sort of thing is scholarship. Scholarship has limits, but you can use it for free for, I think it's like three articles. I'm not sure, is it per month or per week or something, but you can use it. So you can upload a PDF to this thing. And by the way, let me just show you the websites. And the reason I'm showing you the websites is sometimes you have ads on Google that links to other websites. So just so that you know in general how the website looks like, so you know you're downloading the right thing. So this is what CRISP AI looks like. And then Scholarcy looks like this. So it's this purple website. And again, you can see the pricing, but we're gonna use the free version. So this one is really nice because you can upload a PDF of a research paper and a really long research paper. I tried a really, really long one and it works and it essentially summarizes it. It creates references. You can, again, there's a chat bot, so you can ask questions and it'll give you some answers. So yeah, it is fantastic. So I definitely recommend you trying it. Yeah, how do you connect Otter.ai to Zoom? That I'm gonna cover on the video. There's a very particular way I want you to do that. And that's like copying the Zoom link and pasting it in Otter.ai. The reason I want you to do that is because that way it won't take over your entire calendar. But for Otter.ai, please watch, please wait until those YouTube videos are posted and watch that one before you do anything just because I don't want it to completely take over your life essentially. We ought to be very careful with that tool. So for Scholarly, you're welcome to even before I upload the video, try it out. In my video, I'm gonna talk about how can you write a history research paper and how can you do that sort of ground research before you actually start writing using Scholarly. And then the other sort of tools you can use is again, chat bots. So Perplexity, has any of you used Perplexity? Yeah. Yeah, yeah, it's a pretty good thing. Yeah. Yeah, it's nice. So for Perplexity, and this one again, I'm gonna show you a video of how you can come up with a science fair project with Perplexity. But essentially you can even before I upload that video, you can play around with this. All you need to do is click this deep research thing. And same thing for Google Gemini, you can click deep research. Microsoft Copilot has a lot of these, a lot of these chat bots have this deep research functionality. Essentially, once you click on that, what it will do is it will create a research plan first of like, here's what I need to do, here's what I need to find out. And then it's going to like do this internet search and find all these articles and then use them and cite them. And this is again, up to date because it is connected to these internet searches and they're cited. So you can kind of double check your work. So for research, that's what I want you to do. All of these AI resources that I'm sharing with you, there is a free tier. So you are only going to use free version. There are limitations though. So before you actually start questioning anything, just make sure you are kind of checking stuff. Okay, demonstration that I'm going to be posting the YouTube videos. Let's go on to the homework. So the homework for today is writing a paragraph. So here, I'll just open it up. It's using scholarship perplexedly and using the deep research functionality on any of the chatbots. It's up to you to answer the following questions. And that's what's the objective of your startup? You need to finalize an objective by next week. Why would you need funding for your startup? What tools did you use? And if you had any observations. And I only need one submission per team, okay? So with that, I am going to be opening breakout rooms. I'll post the homework assignment so you can start talking about it. The most important thing is get each other's contact information because you will have to be working it on this project, not only during class, but even like during the week. So make sure you have each other's contact information. For people who aren't in a group or I couldn't identify who you are, you'll still be in the main room, but I will try to place you. So let me open up all rooms.

AI startups

prompt engineering

social impact

healthcare and education

productivity tools

group collaboration

pitch presentation

AI resources