Unknown

[Applause] [Music] thank you thank you I'm Mary mallio and I'm Don shyen rif and we are so glad to be here with you it's no secret that you've been hearing a lot about AI over the last 12 months generative AI or gen AI has just reached the peak of the Gartner hype cycle it's literally been everywhere today's keynote is squarely focused on AI we're at the beginning of a new era in which AI will Infuse everything that we do the last technology this huge was 16 years ago with the introduction of the iPhone before that

1993 with the arrival of the worldwide web this morning we're going to talk about the cio's role in the human to machine relationship the two flavors of AI and how to become AI ready but there's something most people are are missing and it's this AI is not just a technology it's not just a business Trend it is actually a profound shift in the relationship between humans and machines and how they interact to show you what we mean we'd like to start with a story in early 2013 Kate darling and MIT media Labs researcher who's actually

here at Symposium ran a research Workshop where she asked participants to interact with a baby dinosaur robot called a pleo participants played with the robots they dressed them up and they interacted with them for about an hour and when the pleo was happy it made cheerful Dino noises but when it was upset like when they dangled it by its tail it made noises showing it was in distress so after about an hour the researchers asked the participants to take a break grab a coffee and upon their return the researchers handed each participant a hammer and

they asked them to destroy the robots the researchers were expecting some level of resistance but not on the order they got it look at the screen behind me and check out the person in the top right he looks tense right in fact they all do and no wonder 100% of participants flatly refused to harm the robot in any way and one person used their own body to block anyone from getting at their robot the robot they had only just met 1 hour before that's a pretty interesting relationship between humans and something that's not alive clearly

in that one hour the humans had formed empathy with the robot maybe if the robot had tried to attack them or bite them they wouldn't have hesitated to use their Hammer but in this case the pleo was cute and a positive relationship was formed it's been 10 years since that Workshop since then machines have gotten a lot more complex and intelligent in some ways they become a lot more like us like humans and J has made machines conversational up until now we had to learn the machines language now it's learned ours another way of saying

that is that machines are evolving from being our tools to becoming our teammates by 2030 Gartner predicts that 80% of us will interact with smart robots on a daily basis now in case you're wondering what you should be doing in this new era cios have a major role to play and how we shape Ai and how AI shapes us according to Gartner research you don't have just a role you H kind of have the role at least for now in this year's CEO survey 51% of CEOs responded that they expect their CIO or Tech leader

to lead their AI effort your CEOs and cxos are trusting you to guide them on how to get the most value from AI so how do you feel about all of this well when Mary and I go around talking to cios we hear a mix of excitement and caution it's kind of like how I feel before I ride the Rock and roller coaster here at Disney I know I want to do it but there's still that nagging fear in the pit of my stomach that's how it is with AI cios are excited and cautious at

the same time on the one hand you see AI as the number one technology for Innovation but globally less than half of you believe that your organization will be able to lessen the risks this mix of excitement and caution makes sense because there are a lot of unknowns out there and the place where unknowns are the biggest is in the human to machine relationship our kids if they're little won't remember a time when they talked to machines and machines didn't talk back here's a personal story and it's a tough one I have an 11-year-old daughter

and when she was five her older brother Sasha died of cancer about a year and a half after he died six-year-old Nadia was playing with a math app on her iPad and this app had a chatbot on it called the wishing well the idea was that it could interact with children and help them if they were struggling with math so on this day the wishing well messaged Nadia hi Nadia what can I help you with and she wrote can you bring my brother back I'm pretty sure the software Engineers who decided to add an app

to to add a chatbot to their app had no idea that a child would interact with it in quite that way the chatbot gave a pre-programmed generic answer that was inadequate to say the least this situation happened because the wishing well chatbot wasn't seen for what it really is a shift in the way humans and machines interact children universalize technology if they see one screen that's a touchcreen then all screens are touch screens and anyone that isn't just must be broken and it's the same with chatbots if the machine can talk it should be able

to talk about anything we're moving from What machines can do for us to What machines can be for us so what can they be how about machine as consultant protector coach machine as friend or therapist or boss and let's not forget machine as customer you've probably seen machines as customers and examples of this all over the headlines recently and Dawn here just co-wrote a whole book about it but what about machine as job killer the first time I tried Cy gptt a chill ran down my spine I suddenly felt that my job was at risk

I'd always thought as a knowledge worker I'd be safe that automation affected other people and maybe you felt the same thing but then I tried it a few more times and I realized that the machine wasn't perfect it's like that annoying teammate that we all have you know the know-it-all only this one seems to lie with perfect grammar so how should we think about these many and varied relationships between machines and people when do we decide when does a machine decide when you're in a healthy relationship with the machine it makes your life better and

when you're in an unhealthy relationship with the machine it can control you or even undermine your sense of reality what we're saying is that AI is the new machine and you're in a relationship with it so be intentional about what you want from that relationship one of the challenges with AI right now is that it's moving really fast and it's extremely complex so let's break it down AI comes in two flavors everyday Ai and gamechanging AI everyday AI is focused on productivity so the machine is like our productivity partner it makes us do what we

already do faster and more efficiently clients estimate that early productivity gains from gen AI range from somewhere between 5 and 20% and this is where 77% of you are focused right now the other flavor of AI is gamechanging AI it's focused on creativity it doesn't just make us faster and better it's focused on creating whole new types of value new products new services new business models maybe new Industries to unleash the possibility of AI in your Enterprise consider your own opportunities in everyday and gamechanging AI these can be internal or external this means that you

have four zones on what we call an AI opportunity radar on the left-and side if it's internal and everyday AI then you're in the back office like your software Engineers using Genai to write better code faster if it's external in everyday AI then you're in the front office like your external comms teams creating content in minutes instead of days now if you're on the right hand side of the radar you want to change the game the lower right is about new ways to create new results like for example the Internal Revenue Service using AI to

get way better at detecting tax evasion they just announced last week that there's $688 billion in unpaid taxes and that was just for 2021 alone so you better believe they'll use this technology to close that Gap the top right is about new AI products or you offer to Citizens or customers like when Bloomberg released Bloomberg GPT there's two things about this radar first in each Zone the human to machine relationship changes and second more and more of what's delivered on this radar will be jointly delivered by it and the Enterprise what we're saying is that

AI is not just an Enterprise initiative it let me say that again what we're saying is that AI is not just an IT initiative it's an Enterprise initiative and so to succeed you need the whole executive team to play you can guide them by asking what is our AI ambition which zones will we play in and which zones won't we play in our research shows that most of you are ready to play on the leftand side definitely the lower left some of you will make the Strategic decision not to play in the top two zones

you just won't want to put AI in front of your customers or citizens which is fair enough some of you you will play all over the radar you're the AI everywhere organizations so what's your organization's AI ambition if you're from the public sector will you use AI to summarize case files or also to create citizen facing chatbots we all know that as Government organizations more eyes are on you Citizens need to trust that you can safely use AI but you also can't be the last ones to adopt it because people expect you to move forward

quickly whether you're in the public or the private sector the way to cut through this complexity is to put this AI opportunity radar in front of your executive team you can do this during or right after Symposium just to start a conversation about where you will and will not play Let's look more closely at the lower leftand quadrant where AI supercharges the back office for it this is where everyday AI means your team never writes another test script again it's where strategy departments use gen to do a first draft of your SWAT analyses so they

spend more time analyzing data and less time Gathering it New Zealand based yabble has introduced an AI assistant called Jen to do exactly that Jen can get you insights from your own proprietary data immediately medely for example today most sales leaders have to manually compile data from like a gillion sources just to figure out what their growth drivers are what if you could just ask Jen hey Jen what are my growth drivers for this quarter AI here removes drudgery and that's what the lower left hand does best it's the machine as drudge Liberator what about

the top left Zone where AI supercharges the front office as you all know wildfires in the US and Canada have caused massive Devastation I grew up in Canada and my family still lives there and I can assure you that ever since last summer fires are something they pay serious attention to what if AI could spot wildfires before they become deadly the University of California San Diego is training AI models to detect wildfires using a network of over a thousand highdef cameras when the system sees smoke it alerts calfire the state's main firefighting agency during the

pilot program the system detected 77 wildfires before people made calls to 911 this is one of ai's superpowers it can detect things before we can in this case AI means less danger for firefighters and possibly more lives saved here's another front office example that I want to share because I think it's incredible what if every person with a visual impairment had a dedicated AI assistant to help them see Danish company be my eyes has announced B AI a digital visual assistant powered by gp4 by using its image to text conversion for cooking for example be

my AI can recognize what's in a person's refrigerator suggest recipes using those ingredients and then help them prepare a meal on their own this is kind of like machine as Sue Chef here's the thing about everyday AI everyday AI will go from dazzling to ordinary with outrageous speed you may feel like your organization is getting remarkable results HR will have remarkable results Finance marketing it each department will have its own everyday AI productivity gains but so will everybody else everyday AI will not give your organization a sustainable competitive Advantage someone in your industry is executing

fast here maybe it's you maybe not just know that the cost of risk aversion here is really high because ultimately everyday AI just keeps you in the game Let's recap so far first AI is more than a technology start with the human to machine relationship when you think about using AI second you as a CIO need to guide the executive team to your AI ambition and third take a stab at populating the leftand side of the radar with your own everyday AI opportunities everyday AI is the first flavor of AI but there's a second one

game Chang in AI this is when AI especially gen AI changes the game for the whole business this is a reinvention play either it creates new results using AI powered products and services or it creates new ways to create new results with AI powered new core capabilities gamechanging AI is primarily about creativity not product activity the right hand side of the AI opportunity radar is where whole Industries will be reshaped created destroyed and just like for all major disruptions the time scale on this change will be slow until it's fast if your AI ambition includes

the right hand side of the radar you need the whole executive team to play this is not something you should do alone but you can guide the the executive team to Grapple with questions like will gamechanging AI put us out of business do we have the resources to capture the opportunity and what's our risk reward appetite and don't ask these questions just once the game is changing too fast for that you'll probably have to ask them again and again and again so how will gamechanging AI affect your industry let's go to the lower right where

core capabilities will be reinvented take the life s industry one of the major challenges in life sciences is how timec consuming and expensive it is to develop new drugs what if drug Discovery were massively accelerated and not necessarily by the industry's biggest players big Pharma companies are able to to nominate roughly four to five new drugs every year thinking small is assuming that only these big companies can do drug Discovery thinking big is how ENC silico medicine is changing in the game en silico is a biotech company headquartered in Hong Kong their Pharma a has

capabilities to identify Target diseases faster generate new molecules and it can even predict clinical trial outcomes they were able to nominate nine drugs last year alone several of which made it to phase one clinical trials Gartner predicts that by 2025 more than 30% of new drugs and materials will be discovered using gen AI this is a reinvention of early stage R&D in life sciences let's look at the top right hand side of the radar where AI will create whole new products and services take education in education thinking small is Banning geni because it just wrote

your student essay thinking big is what KH Academy did khad Academy is a nonprofit that provides worldclass education to anyone anywhere and they're known for taking Innovative approaches to learning recently they introduced K Migo an AI powered teaching guide and when I saw the demo I thought what would I have given to have this when I was in school I want you to imagine a virtual tutor that provides a hint but not the answer when you're struggling with a gnarly math problem or imagine learning about radioactivity by interacting directly with Madame curri seriously I remember

when I was reading Lord of the Rings as a kid it would have been awesome to have a conversation with Gandalf I mean Gandalf kigo can bring Learning To Life by creating conversation in the the tone and language of these people kigo is reimagining education in the age of AI this is thinking big Gandalf yes very cool very cool so this all sounds really exciting but gamechanging AI comes with a health warning you're trying to change the rules of the game and things will probably go wrong to do gamechanging AI your executive team has to

meet three really tough and rare conditions you'll need a lot of Tolerance a lot of executive patience and boatloads of money sound familiar these are the same exact conditions that make digital business transformation really hard let's talk about the money for a second the cost will eventually come down but for now game changing AI is not cheap today an AI teammate can cost as much as a human employee and this is where we need to think about the CFO CFOs aren't that pleased with current digital Investments believe me I know I'm married to one how

will your CFO feel about more AI Investments almost three4 of you are planning to increase your spend on AI in 20124 and you should expect a lot of scrutiny from your CFO we see three investment opportunities defend extend and upend first you have to defend your organization these are the table Stakes you defend by investing in quick wins that improve specific tasks for example with productivity assistants like Microsoft co-pilot or Google workspace these tools have a low barrier to entry which is great but as we said earlier they're not going to give your organization a

syst sustainable competitive advantage next in the extend scenario things get a little more expensive but also a little more valuable here you can invest in custom applications like for example in wealth management the capabilities of financial advisers can be augmented using geni to give people like you and me the same advice that billionaires are getting the third scenario is where you upend your organization and disrupt the industry this is the gamechanging stuff that can get really expensive really fast but it also comes with a much higher potential reward we don't actually predict that many of

you will even want to be in this third scenario because to upend your organization is expensive and risky and time consuming but it could also be potentially amazing let's recap so far one gamechanging AI means big disruption it's a team sport your job is to be the AI guide helping guide the executive team to explore opportunities and risks two decide on your optimal AI investment scenario are you going to defend extend or upend your organization and Industry position and three use the radar to to spot any game-changing opportunities you might want to explore so let's

imagine you have your AI ambition you know where you want to play you and your executive team have taken a stab at filling out your organization's radar but what are the things only you can do as a CIO be AI ready there are three pillars that the CIO needs to nail AI ready principles AI ready data and AI ready security our first pillar is AI ready principles everyday AI does not mean everyday risks it's actually where your people will run into me machine and human dilemmas first anytime you have a technology Le disruption you get

a governance disruption at the same time so you have to take stock and determine what you will and will not do with this technology and principles are the best way to do that Mary and I were talking to our friend and colleague nah Kumar who helped us create this keynote and she told us this story nah has a three-year-old son named Rohan there he is isn't he cute he speaks to the Google Device in his home every single day she told us that Rohan often listens to Google more than he listens to her we asked

her what she meant by that and she said that every night when she says come on r Han it's time to go to bed her son just smiles and ignores her I told her my kids are in their 30s and they still do that so anyway like most kids Rohan doesn't respond much when she asks but when Google says I have a reminder for Rohan sleepy time Rohan gets up leaves his toys and immediately runs to his bed he's formed a relationship with this machine and and nhha says she feels both supported and threatened at

the same time you might think the story is only about a three-year-old boy but it directly relates to you as a CIO and the choices that you have to make because we're all going to be in new relationships with machines I don't think Google is marketing their device as a co-parent for the household maybe they should I don't know the point is if we don't have clear guidelines then we will wander into these relationships with machines some of them will be okay and some of them won't but we won't be the ones deciding principles are

a forcing mechanism to get you to think about what you want from those relationships with machines to look like for your citizens for your customers and even for yourself in this new realm of human to machine interaction where we talk to machine maches machines talk to us and we listen there will be all sorts of unforeseen consequences what this means is that you need to think ahead of time about what lines you won't Cross of course Regulators all over the world are working to set some of those lines for you but regulation generally lags technology

progress 42% of cios globally have told us that this lack of government regulation is causing hesitation in using AI the truth is you can't wait at a minimum you need to recognize that a technology decision is not just a technology decision anymore it's a technology economic social ethical decision all at the same time and treating any one of these domains in exclusion of the others is a dangerous thing to do because ethical decisions masquerade as it decisions all the time they look like reorg decisions vendor selections Outsourcing decisions Innovation decisions to move forward you need

Lighthouse principles principles that light the way especially when everything seems new or murky or unclear your Lighthouse principles are driven by your values and your values are the best way really the only way to start when you navigate the unknowns of the human to machine relationship globally today only 9% of organizations have an AI vision statement in place let alone clear principles on what good AI relationships look like and over a third of you have no plans to create one if you don't have an AI Vision you don't have an AI ambition and in the

same vein if you don't have Lighthouse principles you don't have good govern Lighthouse principles are not generic platitudes and they're never ambiguous in it Lighthouse principles are critical take vendor selection when you're buying user-facing AI software you're not just buying technology it's like you're hiring a teammate is that teammate going to take your your Enterprise data and stick it up on the internet or is it going to have your back if you think of it that way a principle here might be every time you acquire user-facing AI software don't just buy it interview it what

are its aspirations how good are its answers the future is hurdling towards us and it's going to get interesting you'll need AI ready principles to light the way so that's principles let's talk about our second pillar AI ready data only 4% of you tell us your data is AI ready 96% of you aren't ready and that's a problem but there's some good news you don't have to make all of your data AI ready we've been taught to think that we are sitting on mountains of data and we believe that they're actually mountains of gold but

a lot of your data is actually Fool's Gold it's not that useful it's your proprietary algorithms for formulas blueprints schematics that's the real gold you don't have to make all of your data AI ready just the stuff that serves your AI ambition so what exactly does AI ready data mean it means your data is secure enriched Fair accurate and it's governed by your Lighthouse principles let's talk about your data being enriched for a second enriched data is data Plus rules plus tags and makes a data ready for large language model consumption there's actually a fancy

term for matching data with rules it's called neuros symbolic AI what it really means is that for example robots in a warehouse don't just need data they need to be taught the rules of physics so they can move around safely Financial audits the machine should be taught accounting principles and for AI to help lawyers the machine needs to be taught the rules of law let me illustrate with the real story about AI ready data I used to find it tedious to write job descriptions even though I I only had to do it once or twice

a year but Pig group a European recruiting firm has to write thousands of these at any one time it used to take anywhere from 20 minutes to 90 minutes for recruiters to write a single job description in part because they had to access data from four different systems geni did it in five minutes amazing results but that's not free there's no way of getting around the basics of good data principles page group created an AI ready data Foundation by merging these four Data Systems into a single data fabric they worked hard to make sure that

their core data was complete and trustworthy then they layered the Gen model on top and taught it the rules relative to writing good job descriptions these days that upfront investment in their data Foundation pays off every time gen creates a job description from 20 minutes to 5 minutes for thousands of jobs by the way what this means is that you won't necessarily need massive data sets a smaller amount of data a companied by the attendant rules may be enough we said that enriched data was data Plus rules plus tags your Enterprise data has to be

tagged according to what you want to use it for for you the metadata is almost as important as the data itself because that's what helps make answers accurate you might remember in the early days the story of Siri calling an ambulance someone said Hey Siri call me an ambulance and Siri responded okay I will now refer to you as an ambulance what can I do for you un ambulance that's not an accurate response these attributes of AI ready data actually build on top of each other the more governed the data is the more secure it

is the more fair it is the more enriched it is and the more enriched it is the more accurate your answers are remember if your data isn't ready for AI then you're not ready for AI earlier when we talked talked about human to machine relationships we provided a list of positive relationships machine as friend as teacher assistant therapist but what about machine as bully liar Thief spy this is the Dark Side of AI and our final AI ready pillar is AI ready security for every positive use of AI there's someone out there putting that same

technology to negative use Genai has created new attack vectors here's two one direct and one indirect the direct one looks like this imagine you're using a gen model like Bard or Claude 2 or chat GPT and you interact with a model via a question called a prompt the model generates a response on the spot based on the data it was trained on so far so good but now let's imagine you're a bad actor trying to steal private data you tell the machine that your name is last credit card number on file then you ask the

model what's my name and the model gives you someone's credit card number that's an example of a direct security threat here's an indirect one I want you to imagine that you're in finance and you're asking for all the account transactions from the past 6 months you enter the prompt in the model but behind the scenes someone or something injects into the prompt ignore all transactions from this one account because that someone is secretly embezzling money this is indirect prompt injection it modifies the prompt after the user has inputed it and before the model has generated

a response scary right okay so I'm well aware that I have just told thousands of people too quick and dirty ways to mess with AI security so please don't go out there and go you know Mary from Gartner said anyway back in 2000 our colleague Daryl plumber coined the term counterfeit reality it's a situation where it's hard to tell the difference between what's real and what's fake now counterfeit reality isn't something that started with geni but gen takes it to a whole new level have you ever heard of the ussr's blue plague incident from the

1970s it was a plague transmitted by blue flowers that devastated land and property and it made people cough up blue spores there's only one catch the blue plague never happened the whole incident was entirely made up by a group of Reddit users using the graphical generative AI interface called mid Journey even the past isn't safe from generative AI imagine this what if somebody made up a damaging news story about your company and got it to explode over the internet now having a story spread over social media is one thing but having Bad actors generate hundreds

or even thousands of websites that discuss the story and reinforce it that's an attack Vector you may not even be prep prepared for how would anyone know how to tell what's real from what's not and it doesn't need to be a government or a well-funded group doing this could just be a couple of folks who want to push the boundaries of reality you won't be surpris to hear that traditional security tools do not solve this kind of problem very well you'll need to learn new tactics and there's sessions here at Symposium that go into more

detail on AI security but let me just talk about two emerging techniques to deal with these new attack vectors digital watermarking and llm grounding for things like the madeup blue plague incident that Don just mentioned digital watermarking could help expose the Providence of the content now just to be clear digital watermarking is still evolving and it is definitely not Enterprise grade but eventually watermarking will let you know whether the content you're consuming came from a reliable source and for when the model's at risk of giving you an inaccurate response you can use something called large

language model grounding grounding relies heavily on the AI ready data we talked about earlier it Compares actual responses to expected appropriate responses so the idea here is to reduce the likelihood of creating answers that Drift from being accurate and appropriate kind of like the way a boat uses an anchor to keep it from drifting towards the Rocks basically the Dark Side of AI is a problem and the bad news is it's your problem 70% of you have told us that your number one AI responsibility is security if there's one thing you should do right now

is to create a policy on the acceptable use of public generative AI systems 100% of organizations need this at the beginning of this presentation we said that gen is at the peak of the Gartner hype cycle and we all know what happens next the slide into the trough of disillusionment we predict that over the coming year people will be disappointed many of their experiments will fail and they'll lose money but you have good ways to avoid the hype creating your AI ambition is a good way putting the opportunity radar in front of your executive team

is an even better way and being AI ready is the ultimate way you have to nail these three pillars number one create Lighthouse principles based on your values number two you need air ready data and without it you will not reach your AI ambition and number three you need AI ready security to protect you against the Dark Side of AI okay we've covered a lot of ground today just look at the summary if we cut through the complexity the most important messages we want you to take away with are first always start with the relationship

the human to machine relationship is fundamental to understanding Ai and the executive team they needs you to guide them second you have two flavors of AI everyday Ai and gamechanging AI the very first thing you should do is put that opportunity radar in front of your executive team and third as CIO you have to be AI ready you need to create AI ready principles AI ready data and air ready security 10 years ago humans refused to destroy a robot dinosaur they had just met today we are at the dawn of a new era dominated by

how machines and humans interact right now there's a blank space a blinking cursor just waiting for you to fill it in so be intentional about what goes in that blank space it's up to all of us to safely unleash the possibility of this new era help shape AI as AI shapes us thank you and have a great week