Assessing AI's progress

Key Takeaways

[Music] welcome to the fireside chat I hope that you are enjoying the event so far I don't know about you but I've learned so much about Ai and cool applications and use cases as we could see AI is transforming business models by automating processes enabling data driven Innovation and inspiring the creation of a vibrant startup ecosystem in today's panel we'll talk about AI powered Innovation entrepreneurship and how leaders navigate AI but before we dive in I would like to start by introducing our fantastic panel it is my pleasure to welcome minu Gaba senior director of engineering at Google oeta Samson director of uex AI and computer enablement at Google and Jessica kamada managing part at szel ventures thank you all so much for joining us today I've prepared some questions for you but I would also like to encourage our audience to ask any questions on chat shall we get started we've been hearing a lot about how AI is sparking a new wave of creativity and opportunity the AI startup ecosystem is also rapidly expanding and mature businesses are increasingly leveraging AI to stay competitive and to create value in novel ways so oeta I would like to start with you how do you see AI reshaping traditional business models and what opportunities does this present for entrepreneurs looking to drive Innovation and launch new startups can you tell us a little bit more about that sure uh that's a lot I think I think one of the things that I think most about when I think about AI reshaping business models is obviously uh productivity um when you think about entrepreneurism or startups or even even um organizations and teams within large organizations or companies um you always think about those little details that kind of slow action down that have to be done but a lot of people just don't like to do right and I feel like with gener of AI and and AI especially with uh rapid process automation which has been around for a while um that those uh tasks that maybe we have to do but don't want to do are increasingly being shifted towards automation or AI agents or even through um uh chat or prompt engineering um so that it might have taken you that hour to write your first marketing plan and now it could take you 10 minutes right because you have an assistant that's helping you uh so I think definitely productivity increasing productivity within the individual realm but also at scale and then of course um increased personalization so the ability for businesses to really get to know their customers and users on a very intimate level and also Prov provide service at that same intimate level using the data that they collect um consent consensually of course uh to help better those experiences and dare I say a little bit of Anarchy um people might say agency but I do think that what AI allows uh individuals and employees to do is to have more agency over how they do their work so uh in the beginning it may feel a little chaotic where a lot of rules and policies may not even be in place uh uh right now to govern that type of AI us but uh will maybe in the in in in the future so right now may feel a little uh bit chaotic because individuals are using AI in in every in any way that they think that uh they can make their lives better and for the entrepreneur it means uh the one man or one one woman company is now a one woman company with multiple agents that could help them to do all the things that need to be done in a startup so it makes it even um more accessible for anyone to start their own business I remember just reading on LinkedIn about someone who was trying to go to this large tech company called afrotech and he needed a sponsor uh and so he used generative AI to create an app that organized all the conference information does that app exist yes but now he was able to do it uh just by himself uh and his uh AI assistant so uh who knows what the future will bring with uh creativity match with um labor and and Manpower thank you so much Aletta you touched on really important uh topics and some of them I I would like to Double Down a little bit more later in this conversation but you mentioned prompt engineering creativity productivity how agents can help solo entrepreneurs at least at the beginning going from idea to uh startup creation so let's start with talking by talking a little bit more about prompt engineering because it's increasing accessibility but also decreasing the barriers to interact with large language models so what trends do you see emerging in in in the intersection of uux and Ai and how are they shaping the future of human machine interaction I was interviewed uh by deoe uh earlier this year about what makes a developer and I think uh the definition of developer is definitely shifting and changing and we see this across tech companies as well as um uh in in in startups entreprene and in the entrepreneurial space so there used to be this very narrow and strict definition of the person who programs a computer and with generative Ai and prompt engineering and natural language programming we know now that we can speak to a computer and it can understand what we say and so that opens up a whole new realm of people who can program a computer prompted engineering is one of those um ways of doing that um giving Frameworks to the the computer to mimic or to introducing personalities to to the computer as well and so we have a a lot of ways where we can communicate with computers in a way that they can recognize our patterns and what we're asking for and requesting for um that we didn't before that that before there was only one narrow way and that's coding and now we can actually communicate and so I think that brings in a whole new realm of folks and what we call developer experience and that means a whole new tooling and platform to make that accessibility uh a utility so now that we can engage with computers in different ways we now need different tools and platforms to take advant AG of that thank you so much oeta and this is a perfect segue to my next question for minu um we talked about uh chaos of development in AI good chaos because uh development is so fast right now so minu from the perspective of an engineering leader and given the rapid advancements in AI how do you manage prioritization in your engineering teams while still encouraging a culture of in ination and wellbeing thank you Joanna this is such a good question I live and breathe this every single day yes AI is moving very fast and coming from a machine learning infrastructure team my strategy has been to create smaller building blogs so it is easy to adapt to The Changing Times you know this results in less throwaway work when the priorities or the users needs change and let's take a step by step the first step to figuring out what is important is Define your core business objectives what are you optimizing for you will always have more things to do than the time available and if you only focus on the Tactical task you will miss out on dedicated time for Innovation so my idea is fence out some time for Innovation for exploration based on what you're comfortable with it could be 2080 3070 you know choose your number but you need to dedicate some time for Innovation another thing I encourage is rapid experimentation that is the key to finding new things that you can focus on you know average success rate for experiments is around 12% and actively communicating the value of experimentation will actually help you achieve new ideas you know create new new you know you you'll be able to focus on things that you have not been able to focus on because you are dedicated time you're dedicating time for experimentation and create a safe space for failure Empower your teams to take risks good ideas can come from anywhere and of course you have to regularly review and adapt the priorities based on the emerging Technologies and changing Market you might want to ask yourselves regularly hey is there a better way to do this hey maybe the users need needs have changed what do I need to do differently to make sure I'm still being responsible I'm still innovating and I'm still delivering you know the time to Market is the key to success but also meeting your customers need is very very important thank you so much minu you touched upon some really interesting aspects of innovation and how we focus always on the user and how good ideas can come from anywhere but also it's becoming known that AI is generating vast amounts of code so how do you envision AI reshaping core engineering practices in the next 3 to 5 years let's say yes AI is reshaping um engineering practices and um you need to always be on the hook of for shipping new things one of the things we have learned is that iterative development is the key you have to be agile and flexible you have to prioritize iterative development you have to also create robust testing Frameworks so you can release fast you know making responsible AI audits as part of your development process which includes design data collection deployment while maintaining a rapid release cycle is is a key because if you don't do that you know at the first thought you think that oh I will do that in the end after my model is ready after my application is ready but we all know you know in software development finding a bug earlier in the life cycle is much cheaper to fix and faster to effects so one of the things we have to um basically remember all the time is responsible AI has to be part of every part of the development life cycle you cannot wait till the end I'm I'm so thankful there is so many mandates from the government to actually run some ethical reports responsible safety reports that every company needs to do you know we are using the power of AI for the advantage of the um society and not hurting it anyway thank you so much minu and let's move let's shift gears a little bit now in our conversation we talked about Innovation PR practices in product development in improving developer experience and decreasing barriers to enter into new aih engineering practices so let's talk a little bit more about entrepreneurship and Jessica turning this conversation now to you what are the highlevel trends in AI entrepreneurship that you believe will shape the next wave of successful startups I love that question because I feel like uh Trends and AI is all the Venture Community is talking about right now and actually Etta touched on one major Trend already in her answer earlier which is that uh basically AI is enabling Founders to do more with less and for Venture back startups money is power so AI is creating the ability for Founders to beet more capitally efficient and to move faster both ways so in super early safe startups when you only have like a tiny Capital um amount maybe you're funding it yourself you're trying to boost strap until you get your MVP you can do a lot more from a coding and development perspective than you ever could maybe you have ai helping you to write code and uh plugging into different apis that you you know we didn't have 10 years ago and you can do more with a Twan team um to be the equivalent of a five or six man engineering team and so what we're seeing now is the time between let's say zero to one products or um you know C companies that are coming to Market with their very first MVPs the time to be doing that is a lot more condensed and so um that's what I think the the largest trend is right now I'm primarily a consumer and so B Toc B to B Toc business model investor there is a ton that's happening in the B2B and adary space with AI um for our fund as Swizzle Ventures we're focusing on innovations that benefit women which I think is timely giving this as a webin in ml uh Symposium and that benefit women's lives so mostly in health care caregiving and Financial Health and I think from a trend perspective what we're seeing in the entire ecosystem is that AI is hitting B2B software first um it's helping us break down and translate calls and webinars like these um it's helping sales teams and Auto automating Sales processes and helping those teams move faster it's doing a bunch of things and what I'm super excited about is um for all the as for AI to really hit consumer in a big way and to help end users um and one thing that we're seeing already in companies that we're investing in is the ability for AI to be personalized to each individual user in one's case in this is in health care so women's health is uh is very robust like and part of that is because hormones um introduce crazy variables that have been very difficult to control for and research um and giving really pinpointed recommendations on that are specific to each person so now ai is uh enabling that to happen at scale which we could never do before I'm so want try really excited about other is I think kind of moving from access to editing and instead of like having people do research about a billion things it's saying okay this is the one thing that you should do thank you so much Jessica I I really loved how you touched upon different considerations that uh that you take on when investing in AI startups especially how you mentioned that we're looking at women holistically in order to develop and personalized Solutions through AI because it's especially in healthcare is one of the most uh fundamental and challenging things to do is how to balance all the all the idiosyncrasies around women and hormones and come up with these personalized solutions for them so I love that your fund is thinking about that and is investing in that space You also mentioned investing in different Industries and different solutions so what are the most critical factors you consider when evaluating AI startups and how has this criteria evolved as the field mature is it do you still use the same thought process and framework um that you used at the beginning to make these bets and Investments or has it changed over time it's a great question I think the biggest thing that has changed from let's say late 2023 or 2023 in general to now is that once chat gbt launched and really the AI wave really started I think there was this whole Rush of startups that were trying to get in their next fundraising rounds to like add features that were AI just so that they could be in the AI conversation because maybe they were trying to increase their valuation or they just wanted to seem like they were in on it and for investors you really had to parse out okay is this a native AI product that's actually solving the customer's problems or were AI features tacked on to Simply say that the company was AI um and I think one way to parse that was to ask are these in advancements in AI um the reason that this solution could not have existed 10 years ago if yes then it is probably a you know naab base AI solution versus someone like kind of acting like they're trying to be that today I think that has decreased a lot because the brand new companies that were investing in um preed or seed that started this year uh they are native AI companies because they're just coming up in this world so my questions for them is okay how is the company leveraging AI processes and tools into the company building process um basically how Capital efficient are you given everything that exists today and are you leveraging it to your best ability the second thing like I stated before is that um is this a native AI product that is actually solving the core problem and the why now question is what is that secret sauce that's making it so this could not have been built 10 years ago or even 5 years ago and number three is is the AI and the IP around the AI proprietary um and if it's not proprietary then like what is that secret moat that is around that AI um that makes it defensible and that is very difficult for competitors to come in and build um the same thing and like what is going to make this a market winner so those are those are the top things that I look at when looking at specifically from an AI perspective I love that framework we always need to drive Innovation but stay close to our competitors and the to really understand what's happening and what trends are emerging and and previously you mentioned that teams have to do more with less so I think that's a perfect segue into my next question to minu because I always really like to discuss tradeoffs because making decisions is is a critical part of decision making and with that in mind um my next question is centered around exactly that uh so minu how do you balance the pressure to release new AI features very quickly with the need for rigorous testing and ethical considerations yeah that's a really good question Joanna as I said earlier that having a rigorous testing framework as part of your development process is a must you there are now White House mandates to make sure that the AI we are developing is responsible and is safe so that has to be part of your development process throughout you need to to make sure that every step of the process design data collection deployment is all you know safe is tested for responsibility and the way you balance is like Hey how do I ship features fast and versus also make sure that I'm doing all the things that are required all the things that are ethical all the things that I should be doing is making sure these processes are not heavyweight and they are part of your whole development life cycle doing it in the end actually does not help you actually it slows you down it might sound counterintuitive but it does slow you down so that's what we have been doing and again staying in touch with what is needed what you were developing last quot might not be needed now so stay in touch with your users make sure that whatever new ways are coming out to make sure that your product is responsible and safe you're adhering to those there are lots of advancements happening every every day in safety and responsibility I stay in touch with what's latest what's required what are we doing and where are the gaps how do we identify those gaps and fill those gaps and continuously make our product better we have ethical responsibility to the society to make sure our products are responsible I would I want to meet the users need but I want to you meet the users's need safely so safety is p zero for me it's not negotiable but shipping the featur is on time the way to do it is make sure it's part of your process it's it's a part of your process which is automated which is you don't have to think about it you know people don't make a mistake because of not knowing they are doing it because it's just built into the process they don't even have to know about it in Google that's what we strive for normally but I think most of the industry is moving towards that that hey safety is non- negotiable responsibility is nonnegotiable and let's make sure that we have processes and tools in our development process so this become this does not become a hindrance but this actually speeds up your release process keeping your release is Nimble having a continuous train of features where everything is always getting tested is the key I love that because safety is not negot not negotiable Innovation is not negotiable we have so so many things that need to happen in order to have these Innovations get to Market so so so Jessica how do you see the balance between funding research heavy projects and those that focus more on on on commercialization as a BC in a traditional Venture Capital firm this is typically what's true is that we have pretty tight 10year time windows that we expect our portfolio companies to return Capital Within so that we can return Capital to our stakeholders um and so for me as an early stage investor getting in in that really First Institutional round um that means that a company really has to go from like zero to 100 in in 10 years um which makes it really difficult to be funding research only companies versus is focusing on commercialization um so because of that tight time frame and needing to return that Capital we focus on investing in AI products that are commercializing or that will commercialize within a year or so of us investing um that being said I think that especially if it's a llm or there is an llm that's incorporated or native to the company um they're constantly learning right llms are only as as best as the data that's fed and that they're learning from um so if the assumption is that they're constantly learning and improving their models um we do expect there to be some level of ongoing Research In Parallel with the commercialization efforts of the company and that's so critically important because in order to be able to achieve that level in which the technology is mature enough to be commercialized we also have to have really good product Market fit so so that leads me to my next question for Etta how do you approach designing user experiences for AI driven products yeah I think the first thing that you have to do when you're designing user experiences for AI driven products is to focus on people's needs and solving problems uh I think there is a rush to put a tech first kind of um wardrobe on when new technologies come out at the end of the day products people's problems do not come in products right they the people experience them and so when we're designing products that can solve people's problems we know that those products become commercially viable so I think the first step is to um continue with the um foundational uh level of product design that we've all known and love and that is um how do we solve people's problems and then secondly I think when you're designing experiences with AI is to understand the relationship that the user and your products will have over time uh Ai and ml are Dynamic technology they're the first Dynamic technology that we've been able to design with um prior to ml and AI the technology has been stoic um or or or not very basically non- Dynamic and Ai and ml by its very nature is dynamic and so when we're designing experiences with ML andai we have to understand that the product that we design that the person is initially engaged with will change over time uh if it doesn't change over time it's not taking full advantage of AI or ml uh within its product scope so I think it's really important for us to realize that um a we're not the only people designing the products B the user has a lot to say about how that product will be engaged and and B over time that machine and person relationship will change and so this gives us what I call scenario building design rather than just product design um the scenario will change over time the context will change and the product itself will evolved and this is uh may sound like science fiction but it's it's definitely um here and now um especially when we think of things like autonomous vehicles which I used to work on as well the relationship with the machine and the person um evolves over time and so we need to think about that in designing those relationships um and but I do think that fundamentally if we're going to leverage AI ml for the goodness that it can do for um um the world is that we need to keep it focused on solving um user problems and not just on the technology itself I love that and what part of the process do you integrate user feedback into product design and product development as a developer relations lead I have to ask this question yeah I think um I think minu touched on it earlier um and again like ML and AI is a dynamic uh technology right so it evolves over time so the the sooner that you can get users involved in your product design process the better um leaving it to the end will just give you all kinds of problems I think uh I'm a big um I'm a big proponent of co- participatory and co- design uh co- participatory design meaning that the user basically designs the product with you uh as it evolves um you are doing you're moving from thoughts to things a lot faster you're having an iterative cycle and you're incorporating user feedback in that iterative cycle as you as you not only um collect data and build the model but also put the model into production so um users have kind of they should have a stake and a voice in what you're building um from uh from end to end as we as we say in the product design process I think there's this tendency to kind of wait um until you have something all the way built um and with uh ML and AI you can you can leave spaces and gaps for users to help you to visualize and see what the experience really should be and you can use ml andai through simulations in research and other ways concepting and validation um because you actually don't know how users are going to react to something that is automated uh you don't we don't know um and so putting users in the thick of our iterative product design process is kind of like a given or a must or what I call First principles for AI design as a developer relations lead I'm always looking to find new ways on how to bring in more feedback from our users because I truly believe the more we test the more we put the products and the services and these Technologies in the hands of the users the better feedback will will collect and the better the experience will be at the end of the day for everybody because these Technologies they are so transformative and they're going to be used by billions of people so the teams also need to have representation of the people who are going to build this Technologies which which is why we having this conversation today so so Jessica if I may ask you the next question how can women have a seat at a table or let me rephrase it what unique perspectives do you believe that women bring to leadership in AI one is I'll Echo AA a little bit here to say women in AI have a perspective to bring that we have the privilege to create products that solve problems for us specifically and for humans in general um versus just saying hey I'm G to like build another like tool or widget that does this for a corporation um that's number one and number two is I think we have the responsibility as women and and leaders in AI sorry I'm not technical I don't build products but for the folks that do to ensure their AI models are trained on diverse data so especially llms they're only as accurate as the data that they're trained on and historically a vast majority of the information across the internet in general was created by men and so sometimes what Chachi or another um bot will spit out is from a man's perspective um and this makes llms implicitly biased so there's this gender data Gap in general that exists on the internet that I think if we're not careful um AI can proliferate and we have the opportunity to bring it back and make sure that that data set is Diversified um that it's being trained on so this is just a quick example because I was reading this article yesterday about um in the New York Times or j a journalist offloaded all of her decisions to AI tools for one week and when she asked what she should wear at home after her workday the chat bot advised her to wear stylish clothes with light makeup on and like I don't know about you but that's not what I'm doing when I get home from work um so I think you know the issue is that chatbots give advice especially when asked about decision making based on the average of what everyone wants but if that average is that it's trained on information based on what men want then that's a huge issue for women um so I think that we just have a huge opportunity to improve that Integrity of data um and one of our portfolio companies actually that's called DM is building a solution to close this J gender data Gap um and they're training their proprietary llm on both reputable research in women's health that has been data uh like clinician backed um and also real conversations from their 100% female Community um with the goal of making women feel safe because they're getting kind of accurate accurate data back from from their llm so I I'll end with this and just say that if AI is meant to be good for Humanity which as an optimist I think it can be um then the people and the teams building it need to be representative to the humanity and population that they're building for um with women being obviously 50% of that yes 100% so maybe it's time to ask muu uh what's her vision on on this topic as well since she leads development of AI products minu um how do you think these unique female or diverse perspectives will influence the direction of AI development oh God Jessica said it all I 100% agree with Jessica you know the involvement of women in AI is not limited to just technological framework yes we have great women Engineers so you know there are lots of examples where are leading the technology but it is also essential to have diverse viewpoints into the domain and diversity is crucial so that we are reflective of the complex nature of our Global Society our our society is not like oh everybody thinks the same way as jessicaa pointed out we are about 50% of the population so we need to represent our point of view and there is like so I don't want to stereotype women there are some things that women bring to the table lot more than men I'm not saying men don't have that besides it technical expertise they they also bring empathy they bridge the gap between Technical and Society societal understanding they bring collaboration to the table and today the models need to be built in a collaborative model um mode otherwise they will not represent the diverse point of views sensitivity they know what impact can technology have on the society I mean men know it too but women have little bit more of that so we bring so many of these things to the to the table and as Jessica pointed out a model is only as good as the data it trains on you know why when I ask to generate an image of a nurse it's always a woman when I ask to generate an image of an engineer it's always a man because it's training over the data that is over internet and internet mostly traditionally saw men as Engineers women as nurses it's no longer true but is somebody updating that internet and these models are getting trained on that so we need to make sure that we scrap scrap our data for biases inherent biases that exist on the internet which with all the models are getting trained on so data cleanliness data uh augmenting all this is important in which coures input to the model so that we can actually build AI that helps everyone including 50% of the population here why love that I I would love to continue this conversation I could spend hours and hours just talking about uh innovation in AI the diversity the need to bring different perspectives just because if we look around this room we all have different backgrounds we come from different countries we speak different languages and that's why I think it's so important that we bring all these different backgrounds and experiences into AI development um because this technology will literally trans form the the way societies operate so um so but we need to wrap up our conversation we are almost at time and I'm I'm just learning so much from you but if you allow me to ask just one last question so muu today we have so many women watching us that come from different fields and different Industries maybe they are experts in public health in finance in retail coming from all these different fields so what are some easy maybe a bit more tactical first easy uh steps for someone wanting to use AI in their own industry and how can they begin their Journey towards implementation so the first step would be identify a clear well defined problem within your business within your domain research the existing AI tools that are relevant to your field there is ton of research in this area and there are lots and lots of tools tools a lot of them are free to use find some relevant existing AI tools gather and analyze your existing data to assess the potential of the AI applications and sometimes you might need to augment that data as I as we said you know data is a key part in any machine learning application then educate yourself and not just yourself educate your team also about the basic AI Concepts you can do by do this by learn doing online courses reading industry articles potentially Consulting with some AI experts that can give you some implementation strategy and sometimes you know when you are like okay I still don't know how I can help my indust help my business with u using AI sometimes you need to see other seed ideas there are lots of people exploring how to use AI in their fields so it can sometimes help if you just do a little bit of search online and see what other people are doing for me it works better when I'm brainstorming IDE with other people when I see some idea that Sparks oh yeah I could do that for my industry too oh I could do that for my business too I could do that for my team too I could do that for my home project I mean we have been at home also we so many times think about like oh it would be cool to use a machine learning model to do this solve this problem you know it could come from anywhere so brainstorm talk to people read online I'm sure you can find a problem that machine learning can help uh solve these days machine learning is becoming cheap and more accessible so it is in the hands of you know we are putting machine learning in hands of everybody so no problem is too big or too small for AI so I would just say start with a clear well- defined problem that you know that you want to be solved and then see you know what what tools exist to help you and if you are running out of ideas then I would just brainstorm with somebody either in your team either in your company or your family or Internet there are so many forums where you can say like hey yes you know do you have any ideas on how I can use AI for solving I'm in this business how do I help myself and the kagle community is also a great place to find this inspiration on how to get ideas and which use cases um people can can start using AI for so very similar a similar format Jessica very tactical advice for our audience today what are the the key steps that someone should take to turn their idea into a successful startup I think the whether it's a AI company or any other company I think if you want to build a venture back aable startup the first thing is to a ensure your idea is truly unique and that it has a large enough target market um because remember VCS are looking for companies that can get to it at least 100 million in annual revenue um so make sure that your idea can do that and that from an AI modeling perspective that it's defensible from a competitive landscape um you know before we even start building and I think the the second thing is to start building um whether that's you as a solo founder getting different AI tools together and just like muu was saying I think she laid out a brilliant way to actually explore those tools um and I think in a lot of different cities there are now a lot of support groups for um startup Founders to either link with AI investors or to link with folks to help you build I'm based in Seattle and we're a pretty large AI Hub um and I just want to put a shout out to the Allen Institute for AI uh the ai2 incubator here in Seattle that is a huge um support system for AI Founders here and I know that that exists in uh in other cities as well so I would recommend kind of reaching out to your local um startup ecosystems and start getting involved finding a co-founder Etc that's a great recommendation and of ITA we're wrapping up this panel and you will have the last word so what what advice would you give to women currently in the field so women already in thei who aspire to take on more leadership positions yeah I think Mino and Jessica definitely hit the Tactical um um activities that you want to do to get started I think one of the things and I'm just gonna speak philosophically here is um ML and AI cannot solve every problem so my first principles on designing with AI is to make sure that ML and AI is both applicable and desirable to your problem space and that means you want to think about shifting your idea or your Paradigm on technology from uh the capabilities to also the limitations of ML and AI um I think it's really important that we understand that this technology is not a a bomb and Gilead for every problem that's out there um we really want to Define our our problem space and make sure that ML and AI is applicable as well as desirable is it desired by your user group or your target market to um automate something that maybe they take pleasure in um these are things that you really want to think about before you do it I've talked with so many entrepreneurs and startups who thought that AI was the way to go and then once they started testing their product they found a huge um kind of almost resentment among their target users of um automating things that uh that were distinctly human um people people are quirky they decide to do things um maybe in a laborious way for a reason and so you want to make sure that you think about your problems faced and whether ML and AI is actually applicable to your problem um and once you decide if it is then you want to think about the limitations of the technology that you want to apply uh Ai and ml is is only as good as it's design in its data and so it's not going to solve every problem out there and you can't just let loose on its own think about a toddler that has learned the whole of the internet would you release that toddler onto anybody un suspecting person no so there are some things that you really have to think about and that includes guard rails uh that includes trust and safety but also the limitations of this technology um you can't just have ai to fix something so I think philosophically you want to think through all of these kinds of um um conundrums um when you're deciding to apply ml or AI to your product design thank you so much oeta that's the perfect ending for this session I really love how you said AI is only as good as its design and its data because that's that's fundamental fundamentally right thank you all so much for joining us today I really enjoyed our conversation I could could spend hours just continuing our conversation I still have so many questions so so maybe we'll have another edition of this panel at some point simply because the field just keeps evolving so rapidly Jessica minu OA thank you so much for joining this session and for for bringing your perspectives on AI development and Innovation thank you all so much for watching us and back to the main program we've learned so much today from new skills to New Perspectives to new ways to solve problems what did you find most useful what surprised you share your single most impactful or surprising thing you learned about AI today in the chat or reach out to us on the Google AI Forum at discuss. a.google DOD as you begin to use all you've learned in your personal and professional work Google is here to support you a great next step is to go to Google AI studio and start using the Gemini API or the Gemma open models family join a kagle competition and engage with the community thank you so much for joining us today let's keep Building Together Google is your best partner I'll see you next time [Music]