Extreme Networks CTO Nabil Bukhari: Can We Trust Generative AI?

hi I'm James Maguire and on today's webcast we're talking about a real hot button issue generative Ai and Trust does artificial intelligence have our trust should it have our trust that's a real issue what does it all mean for business to discuss that I'm joined by a major industry thought leader with me is nuil bukari Chief technology and product officer at extreme networks noil absolutely thrilled to have you with us today James it's always good to be here always was good to talk to you and you're a spot on that is the that is the topic these days that that is a that is a Hot Topic these days uh all right you You' said some things and I I will quote you you've said there's a lack of trust around generative AI the technology executives are forgetting in the rush to adopt AI the fundamentals of our human society and culture if a if AI doesn't have our trust it's dead in the water now I I agree with all that personally but what is what is your thought what what's the issue here yeah James uhu I did say that so that's it's a lot to say and some Executives don't want to say that because they want s of put that under the carpet like what AI not trustable we're trying to sell AI so I I I applaud you for saying it right and and James so so let's let's unpack that a little bit because there there's a lot more to it in there which let's start by talking about what do we mean by trust in a work environment right because on a personal level um you know on a side note let me just say this part and then we'll come back to trust sure um you know I was speaking about I speak about AI all the time and and what recently I was speaking somewhere and a person in the audience asked me and said like well why is AI different right and it really got me thinking and and surprisingly the answer was not that came up in my head that oh this is the fastest growing technology or is like amazing and stuff because there's a lot of other really amazing Tech that is out there and some of it is going really really fast sure but the unique thing about AI is the this is one of those perhaps the first tech that the audiences the end user got to know about it and got exposed to it way before the companies could do anything with it and that's the whole chat GD exposure like you know people in the first month and nobody in the Enterprise had any thoughts around it right and that's set the pace and the expectation to it but so I I always find that that's like really something very unique about AI wa it's I agree one small point I agree with you in that it's kind of confusing that there was Enterprise AI for years before chat GPT companies were pouring money into it and maybe not getting that many much results and then GPT you know debuted in November 2022 and it was a consumer technology so it's a funny thing about this this technology it is a consumer and an Enterprise technology where say for example Cloud you know like private cloud and and hybrid Cloud that's not really a consumer technology but AI is both a consumer and an Enterprise technology 100% And it's a tech that got exposed to the users first before anybody and this now leads into the the topic that we were talking about about the trust is that it has created this dichotomy of expectations from the users because you know they can sit in their private offices and they can open up any of the chat Bots there are so many of them available and they can ask them to you know design their travel or you know find recommendation write them an essay or whatever have you and they're using it on a day-to-day basis and then they come to the work environment and they expect that their work environment would have the same level of AI right but the trck is that people inherently trust that the Enterprise Technologies would be more secure would inherently compliant with your company's you know policies would be compliant so they don't have to think about it would be accurate there is so much data on the expectation of accuracy between a consumer technology and an Enterprise technology right so these are four or five things that kind of come together and now the users just expect that when they log into their company's AI it will inherently have all of these things but they want the company's AI to do more things than what the consumer one can do and this is the dichotomy that is is happening out there and it is forcing companies to actually adopt AI faster than they are actually capable off doing and that's really where I was coming from that there is a certain expectation of trust that includes accuracy compliance guard rails everything that is inherent in the people and right now on the Enterprise side I would tell you from my experience n out of 10 Enterprises will not be able to deliver that and then what will happen is that users will lose that trust so that was the background of that comment it makes perfect sense and I I could see someone going into work and they're they log into their their you know chat bot and think it's going to be at a higher level more secure but of course from their view but even the company's view it's actually less secure because if I enter oh you know my grocery budget every week into my you know personal GPT it's no big deal but suddenly if I'm in the Enterprise setting and I enter my you know divisions budget into GPT that's proprietary information that's really a problem um so even you know I think the company may be more worried than the staffers are worried like what are how are our staffers using this we're nervous about you know what what the replications are 100% And if you remember early days of Chad gbd there were like multiple different news items out there that oh such and such company their IP got exposed through Chad GPD multiple different variations on that theme and now we're like three four years three years down the road but but still what we're seeing is that there is this rush to push out an AI strategy by every company out there and quite frankly people like myself and people sitting in the boards they are the worst you know offenders in that because no matter which executive I talk to the story goes something like this like oh we walked into our board meeting and the board said we must have an AI strategy right you know and you can TR you can replace the board with a CEO or a TX or whatever the case might be mhm the reality is James that the best way to talk about AI is to not talk about AI right when was the last time we got up and we said like hey you know I really want to build a lot of buildings so I must have a concrete strategy like how am I GNA next concrete right we don't do things like that or yeah you know I really want to create a data center so I'm really going to start with the fossil fuel strategy right so sure this is that it's kind of upside down so then you could ask like okay so where should we start from and this is something that we talk about quite a bit is that look AI at the minimum is an accelerator and and on the higher end it is something magical so it it goes between an accelerator to being completely magical but the thing is that it doesn't do anything on its own what you want it to do must come from you so what are you going to do do with the AI right so right now if you think about most of the companies they say okay let's start with a Chad bot Chad Bots are already commoditized right and most of them actually do a lot more damage than benefit um but the reality is what are the experiences that you want to enable and and I like to talk about it in an arc framework and and stay with me for for a few minutes I'm throwing another acronym in there okay good no I love acon because we are we are tech people right so the arc really Arc so look at you as at a company right as a leader if I am said like hey you know I I want to do something with AI I should look at the experiences of my people my customer my users whoever and I should segregate them out into three major buckets experiences that I would like to accelerate because the experiences are fine but it takes too long think about like sitting on a phone waiting for a customer support person like that's an experience I would like to accelerate any of the automations you know right you know copy and this and that and stuff then there are experiences that we say oh my God this experience absolutely sucks I wish I could replace it all right that's the second and then the third category is like hey I wish I could create a brand new experience for it so accelerate replace and create and that should be the first conversation companies should have and then see if AI can help you accelerate replace or create and in some cases it would and in some cases is there are much better other Technologies to help you accomplish that so I'll stop there and this is like that is it's chaotic out there people it's a rat race or whatever other better term there is uh people are just dumping money into it without really considering the ROI and that doesn't that's doesn't bod well for the tech itself I think you're very right I mean a rat race I would use the word confusion there's Mass confusion out there and I I like your idea of the Ark as a sort of a found to build an AI strategy I'm wondering actually at this point if the idea of an AI strategy for the Enterprise is sort of a myth not because the the managers don't care is that I don't think they fully understand it and In fairness to them it's so new and it's changing so quickly how can they fully understand it because it's capable of something like this in June and by December it's capable of something else altogether so I'm going to build a strategy for this thing that changes every three or four months yeah um I want to use ARC I like that idea but it's like okay it's going to accelerate well accelerate versus replace versus create it gets kind of blurry it it seems like it's a very hard thing to make a strategy for it because it changes so fast it is absolutely and James now think about it this way if if the strategy and you're rightly saying so that the strategy around the arc will move really fast now imagine instead building a strategy on their teets upon the what I'm sorry around the tech itself or the technology Tech okay the technology itself okay yeah I've heard people kind of like talk about like oh I'm I'm going to build all of my AI on such and such large language model and and I always laugh with them is that by the time you set that sentence there are three new ones that popped up and then there are all these architectural like people get wrapped up into this conversation around like oh am I going to am I going to build my own language model am I going to fine tune it I'm going to use rag architectures am I going use you know complex ai composable ai like there's 18,000 things are moving so fast so what I talk to companies about is that look what experiences do you have and you want to deliver that is a lot more manageable problem set than trying to go after the latest technology in this field so look the reality is this is moving much faster than what we are used to and there's no escaping that we we can't bring it down to the speed that we are used to right but it's a lot easier to lock onto and build a strategy that we actually understand a lot better than the tech itself um and that's really the whole angle around that look AI is really there to accomplish something so that let's build that strategy on that something rather than AI itself right okay so you're saying uh don't don't think exactly about everything that AI can do take a take a specific task that you want to improve and then apply AI to that it's a very practical approach so you know so a lot of companies lot of companies they say like okay we are going to use AI so let's think about where all the different places that we can apply AI now I am very categorically saying that's the wrong approach and most of the companies that take that they will end up spending a lot of money creating a lot of different silos and creating all sorts of chaos inside their company that's just a bad place to start with now imagine that you come in and you say like hey what is what are my biggest problems what do I want to solve like oh you know as an example I have customer uh support people I don't have enough people so the call time is too long for responding or you come in and you say like look I do Global business in like 72 countries I don't have enough marketing people that can translate every document in every language and the list goes on and on and on and on sure now you didn't start with AI you started with the problems that you have and then you go and say like oh is there an AI tool that can help translate well all of a sudden you'll find like 6 you can choose one of them is there one that can you know um calls before they come in so now what you're doing is you have specific problems use cases experiences I like to call them experiences MH and then you go looking for an answer for them and in that it's a lot more manageable problem than trying to go and say here I have ai in my hand where every place that I'm going to apply it right I mean I like I like the practicality that approach it sort of kills the magic of it because you talked about magic earlier and I like the idea that we can just bring ai ai in sprinkle it over the software and it kind of makes everything you know work faster maybe boil an egg at the same time yeah uh but you're killing the magic Nal I mean what about that I know but James as they say even the best wizards in the world they started by learning how to hold the wand I don't know if anybody says that but you get my you get you get my drift on that right it's just it there's a lot of froth in the market when it comes to Ai and a lot of it will settle down and at the end of it unfortunately there'll be a lot of companies who would be like wow we spend millions and millions and millions of dollars in it and what did we get out of it right and my aim as I go talk to our partners our customers and just industry as a whole is to help them get to a point where when we look they look back in the last 18 months they say like yeah we only applied it for two use cases but this is the value that we got out of it and while doing that we also learned all the things that not to do right and now we can go and take that cultural you know movement that I've created and then replicate it in different portions of my company it is pragmatic perhaps not as sexy as everybody wants but trust me in the end we are responsible or or answerable to our shareholders one way or another and to our people and our users and I think it's important that we take a pragmatic approach in this good all right so I and I think you may have answered this question but I like this question so much that I want to bring it up just to get your answer so if AI does not have our trust what can managers do to earn this trust and I think part of the answer there is to use it in a strategic practical way but is there is there another answer you know and so so I have two kids one is fouryear old and one is seveny old and you'll be like why am I going on that tangent because because in the end trust is about humans so we should think about how do we learn to trust when we were kids I just I'm teaching them how to ride a bike right and then so think about this way so my my young one who's foury old he wants me to hold on to the bike and run next to him because he trust that he will not fall sure now you do it three four five times and then slowly I start letting the bike go a little bit early I think we all learned that way we learn Tross by reputation the by repetition right so more and more we do it after a little while we trust that it will happen so the best way to learn trust or gain trust or develop trust around a technology inside as an organization is to give people time to deal with that multiple different times from a smaller thing to a larger thing and that's how we learn trust as humans and and Trust around AI is not going to go any other way because this is not a problem or a question about AI this is a concern and a thing that we have to do around humans and while AI is magical we are the same good old humans right so what I believe is that managers must remember must absolutely remember it is about the humans in the end it is not about the air it is about the humans so optimize it for humans how do they trust things how do they uh experience things how do they get used to things right and like the same thing I'll give you an example coming back to the trust part every time I speak one of the questions that everybody asks me is like oh will AI take my job well that's a big one sure that's a big one and I always tell them that 50 years 100 years from now I don't know maybe but 50 years are 50 years away right but I can tell you one thing right now it will not be AI that will take your job tomorrow it will be somebody using AI that will take your job tomorrow right right that's really the key how do you build and how do you apply AI to use cases where we can make our people more um productive more useful give them a little bit more time mental space to think about it people think that as businesses you know it's all about the bottom line I do agree it is all about the bottom line and the only way to improve the bottom line is to really focus on our humans rather than just on the a all all makes perfect sense for sure all right so if you if you look to the the the future of AI in the Enterprise and we don't need to look 50 years out but a few years out um it amazes me it seems like the adoption rate is already remarkably fast uh certainly on a personal level I look around me I see friends and family using AI in many different ways surprisingly what do you see looking ahead the the future of AI and the Enterprise and perhaps in our lives as well yeah and look I'm I'm going to bring it back to you know the arc framework and and it's not just the way we descri this is also how we are deciding our investments you know on AI and not in the last 3 years but in the last like 12 years because as you mentioned Ai and Enterprise has been around for a little while and we were one of the Pioneers in that space yes where we are today most of the use cases that I see out there are all really about accelerating something that we already know how to do right and like okay I know how to write an essay but I can ask the AI to write it better for me right I can it's really an automation use case right so it's like okay I know how to do it I can accelerate that through Ai and AI can accelerate it way better than perhaps the previous set of technologies that were out there I think that that's the phase that we are in we will slowly move in my opinion in another two years we will really start moving into the replace where it would be like okay these and this is where most of the alarmism comes in that oh maybe and even some big reports that are out there like who was I read the other day I don't want to quote somebody wrong but but somebody like that that in the next three years it will have a significant impact on 600 million jobs globally yeah I believe it yes right so will it yeah most probably and that's really the second phase in which we'll go and in that replace phase we have to figure out whether those jobs need to be there or not or Define in the way in which we Define them today we have to think about what is the relationship between a human you know employee and an AI employee working together like AI is a coworker concept that is out there AI as an agent worker you know that is out there but that's something that I believe in an Enterprise world is mainstream it's like about two to two and a half years out we are squarely in the accelerate phase and I believe that companies that do this accelerate phase really well using the arc framework focusing on the humans they will be very well p position for the replace phas companies that do it halfhazard it's going to be a shock to the system so that's how I view that we will have a lot of they made it and a lot more of oh they couldn't figure it out and they they are either not in business anymore or whatever the case might be so I do believe that that's where we are headed on the Enterprise side there is no company that can avoid it but just following it based on the latest you know Tech news out there that's not the way either so this will force companies to really go back to the basics and understand what is the outcome that they're trying to deliver and then apply AI in a very deliberate fashion I know probably a message that's contrary to what everybody wants to hear right well I I think there's also between the arc there could be a certain blurring of functions if you're going to accelerate something or create I'm thinking of a friend of mine who's working on architect ual project thing he just used the AI software to kind of create this frame for his house and there it was it was really done it was actually really well done and of course he could have hired an architect and so he accelerated it you could say but in a sense he actually created it so it's like the difference between what what's accelerate and what what's create gets really blurry as much as I want it to be clear no no it absolutely does and that's where every company has to run it for themselves because you know if I sit here and say these are my these are experiences I want to accelerate and these are the ones that I want to create there's no universal truth or Universal definition about it right assign whatever you know letter you want to assign to it but you must assign a letter like a r or C because that will allow you to categorize because in the end no matter who you are as an Enterprise especially uh and no matter how big you are you simply do not have enough resources to go and start doing everything you know right so that's where so however you categorize there's a lot of L in how you categorize it but the key goal there is to force you to categorize your use cases or experiences up front and then go apply AI to it I will also add one more thing because um you know a lot of our U lot of the people out in the industry then fall into the second issue in there first is not categorizing not deliberately selecting and then just going you know Peanut buttering it the second one is that they say like oh let us start with a use case that is the lowest risk low hanging fruit yes there's a lot of that sure one of the things that I would through you if I can advise to everybody please do not pick the law hanging fruit with AI really okay tell me why I'm curious why not so because this is how it goes It goes like you're like oh you know I'm going to use AI you put like oh this is a use case which is low risk and and you know it's easy to do and everything but what you forget is that low risk easy to do things are also low value as well right so then what happens is that you're still going to spend millions of dollars doing it and then you turn around and say like aha look I did this use case and you look around and nobody gives apologize for my language there nobody really cares because because this was not a high value usee and then you say like okay now I need the extra next million to do the next thing and everybody looks at it and it's like well sorry sorry you know there's no Roi for that right so pick the use case that people care about pick one but pick the one that people care about so when you are able to either accelerate it or replace it or create it and then you are successful people be like wow this is amazing and then you can show the return on investment you can show the value and stuff a lot of managers out there are going by the same way as we used to in previous Technologies let's just use it in a place where there's little risk AI has to be done completely opposite it's counter it's counterintuitive it scares the hell out of everybody but that's the nature of AI right one or two use cases that matter that that that is an interesting thought and a provocative thought and uh noil I think we talk about this another hour I feel we better say goodbye just to get back to our day but wow I would love to talk to you again sometime a lot of fascinating stuff thank you so much for coming on today James thank you so much it's always a pleasure talking to you uh and thank you for having me here

This transcript was generated automatically from the video's captions and may contain errors.