Research Computing Roundtable: AI in 2024

I thought it was gonna be exactly all right Brian hey everyone Brian fan here I'm a Solutions architect here at ciq uh my background's in HPC Administration and architecture uh it's good be it's good to be back on the webinar yeah good to have you Mr Allen got find the button yeah so I'm Alan S and I did an absolutely miserable job of of flubbing my introduction last time so I will just point people to my LinkedIn profile which is linkedin.com in Allen d-t and if you go there you can actually find a link to uh developer Stories podcast just posted by Vanessa soat was

one of your recent guests uh so you can get lots more than you ever wanted to know about my background in high performance Computing okay so am I am I redeemed you are redeemed yes awesome and I'm sure someone in the background here is gonna grab your your little link and posted in there so easy access to Allen s he is an amazing man so thank you for being here and Mr Forest BT good morning everyone I'm Forest B I'm a Solutions architect here at CQ my background is in H BC especially applications and containerization uh especially in the academic and National Lab space so

very excited to be on and uh see we have a great panel here today yes great panel okay so we're gonna get into it because this is definitely a Hot Topic there is all kinds of opinions and ideas floating around we're gonna get to it but just real quick while you guys are kind of thinking about what it is you want to say about AI I do have to say everyone watching out there in YouTube land thank you thank you thank you as of today we're at 925 subscribers to the ciq YouTube channel so what we're doing is we're going to be doing a little

giveaway once we hit 1,000 we are going to um pick a lucky winner we should probably do that live that'd be so cool and one of those little like pick a random number between you know what I mean like have the AI do it right that's advanced stuff we got going on here guys we're going to picky look winner and we are going to give them a backpack so let's see if we can do a little presentation share screen share screen which screen I want this one that says carart back pack I'm doing it you guys I'm I'm already trying to share Rose oh you

are sorry I anticipated that you would ask me oh you got it right on you guys this this is the thing you know human beings can learn we we can learn and we can do cool things so when we hit 10,000 subscribers so like subscribe share this with your friends um and if you guys have topics that you want these amazing people to come together and kind of hash out you want to see certain demos on products that we that we provide you know werewolf obtainer Ascender Rocky anything that you guys want to know let us know and we'll do that so yes lucky winner

is going to win a beautiful car heart backpack I love it okay let's stop the screen share now right I like it we did it I feel very proud of myself okay so yeah oh fanda welcome please introduce yourself so glad you're here hey I'm Fernanda director of HBC of Ultron data woohoo okay so we are going to be calling on alen sill first because I know that you kind of gotta like run out the door to another meeting so I want to hear it man what have you got to say about what is happening in AI quickly wrap up what you're concerned about what

you're excited about where we're going okay so I've got a little bit of time but uh I won't take too long I hope uh everyone who has been on this podcast before this what a video cast knows that I have opinions about um well a number of things but uh uh first of all clouds let me just start with clouds public clouds there's no such thing as a public Cloud we have public parks we have public libraries we have commercial clouds to which you can gain access for a fee we do not have public clouds the closest we have is grid Computing you're welcome um

the um the second thing I want to say is there is no AI uh because actually it's a it's a masterful bit of relabeling of stuff now we covered all this before so I'm just recapping so um you know AI is a is a term that's come to be applied to a large variety of things that have been around a long time and the latest of which is large language models uh and I just want to point out which is said correctly many times on the internet every day and everyone ignores it that lar language models are not general purpose AI so what is AI

we talked about this in the last time and we came came to conclude that in terms of Technology AI is largely HPC relabeled right uh I I'm I'm willing to sign on to that opinion from the point of view that U that uh AI uses and builds on machine learning and large scale data processing but I think there's two levels to this discussion so the second level is the one that I've been focusing on lately because it affects my purchasing plans as a director of a high performance Computing Center and I think that's kind of what a lot of people want to know about in

in in steering their choice never mind whether you think any of the things I said up to now are correct the question is what do we do about it and I think that there is a very simple razor as in aam's razor as in something you can use to whittle down to the core of the problem that you could ask in your setting which is what is the source of the data so when we talk about large language models we talk about general purpose AI generally speaking you're talking about people U mining the entire concept since the internet which turns out to be just a

linear algebra problem and remixing it and giving you back pictures stolen from people plagiarism is a service I call it at Great expense to the planet right please stop doing that uh Jensen uh I know it made you billions of dollars but um you know it's really not uh something that's going to apply in most scientific data centers why not because you don't have that data set and you're not interested in that data set so what drives HPC and AI the interesting questions and I think there are plenty of them boil down to what is the source of the data so there are lots of

very legitimate and valid and interesting and fascinating applications of uh AI uh methods in science as in you know steering computations and selecting among optimization problems as in working with data but when you ask yourself the question what's the data set that's in my hands that I'm actually going to use it's not large language models and it's not uh image generation it's not that stuff it's applying these methods these machine learning and and inference and training methods to the data that that you have to work with and so that's that's uh thank you for giving me several minutes to talk about that uh and hopefully

um some of what I said is useful Gary I see that you're already off mute are are you like forming thoughts here and you want to have a a response or adding to it or you're just hanging out I you know I I think the uh I think the streamyard already did it for me I didn't press anything so it was the AI yeah yeah but but I I no I I'm just enjoying listening to to Allan speak I I I think he has some good words here so I guess my my question that comes up is like what is it that you think other

people are meaning when they say AI because it's a ve it's a very strong like let me explain to you and you know where I'm coming from but what is it where what is it that you think they're coming from so is it like a you know AI is some kind of like alien intelligence that like is doing something to us and you're like Hey listen you guys like that's not really what it is like we're we're this is has been growing Dave's laughing at me Dave you want to respond to that well I mean fin I I don't I'm not sure where you're where

you're going yet uh do you want to finish your uh your thought slash question there yeah okay so like for those of you that are in like actually been been doing this for a long time right this uh um uh these learning models and you understand that it really is just uh alen how did you call it it's a plagiarism right plagiarism is a service but but it's a great name and I was just going to point out that that Donald n the famous computer scientist uh observed once that 90% of a good piece of software was its name and that they had come up

with a very good name and he was trying to write some software to go with it to some degree is is a AI is a Triumph of labeling and just waiting out things until you could find something that um that uh that did uh uh you know meet the criteria and um and uh the large language models did significantly Advance the state-of-the-art of what people think is amazing if machines do it I I just have to point out that it's just linear algebra and that it just consists of mixing inputs there is no intelligence in what is what has finally brought us over the hump

of calling something AI there is uh it's built on a very large amount of other work that I think everybody here can discuss at length about machine learning and and uh training and uh and data science uh remember that as far as names go when we came up with big data I loved the definition that said the big data is everything we used to throw away right it's all the you know the records of the grocery store camera of how long you linger in the aisle in front of the frozen peas or something you know and you know it was only when we going back

to the argument that this whole group had made earlier that that AI is HBC it was only when we had sufficiently large scale Computing and data process and that we're able to use such data so big data was the previous Triumph of labeling AI is sort of one the one the game but it's just built on top of these other methods Alan I'm really curious what would you say to people who say that it's not exactly plagiarism it's learning in the same way that like learn because you can reduce all large language models to to linear algebra it's just mixing of inputs there's no learning

involved in other words there's nothing that it can in fact the biggest problem we have with this is it routinely generates garbage right and there's no fix for that there's algorithmically no fix for the ability of these codes to Simply take their predictive measures of and it's simple it says I'm at this point what is the next most probable word or phrase or picture or pixel you know you will never fix the the the capacity of that method to generate mistakes very interesting thank you Fernando I have a Counterpoint humans consistently generate garbage if you've ever listened to any politician out there oh not argued

yes worried about like next year or so as politicians I think I just did a I am far less worried about an AI becoming a politician than I am a politician using AI put it that way deep I don't want to cut you off if you're not done foran Fernanda but um no I'm I'm done but go ahead I think you are actually an expert in this field so please well I will say too so um Allan I I think that one of the points that she made is that um these models they're just big piles of linear algebra and so because of that they

can't be considered intelligent I will point out that your brain is just a big pile of linear algebra as well I mean just a little bit of nonlinear I mean you're just like a neuron is just a summation machine right and you just have um you know different different neurons and different layers and different areas of the brain that are that are you know filtering through and doing essentially exactly what these um these models are doing so and you know I'll also point out that we don't so I I want to make some statements but I'm afraid that I'm gonna drag this conversation into a

really philosophical drag it go for it okay um all right we're in for it all right my background is I'm a neuroscientist and the reason I'm a neuroscientist is because I got to a point in my life where I was just like I have this experience of being me um that's different from the experience that Rose has of being Rose and I we call this thing Consciousness and I can you know appreciate all my senses and I can take them inside and do stuff with them and have thoughts and y y y we all kind of know what Consciousness means even though it's it's kind

of hard to Define um so I jumped into Neuroscience being like I want to study Consciousness and I you know read a bunch of books and a bunch of papers and I studied with some of the people that wrote those books and you know did all that kind of stuff and reached kind of like the the the community the highest Community within Neuroscience which was studying Consciousness and a lot of those people had become disillusioned by that study by the I got there and started telling me you'll end your career trying to study this because there's no answer we know that collections of neurons firing

together produce Consciousness somehow because we know when they quit firing the Consciousness goes away right beyond that we don't know much and so with that very very limited understanding that we have that people who have spent their careers studying it still have I think it's maybe a little you know it it's it's kind of a little too easy to just say that you know these virtual neurons um summing information together and spreading it throughout a network and honing it on stuff um you know I mean it I how can we say that that's not producing some form of Consciousness when we don't understand how Consciousness

is produced and even if we don't go as far as Consciousness even if we stick with just the intelligence label I mean you know the uh you're taking uh information that's out there synthesizing it and um you know uh simulating which is a lot of what your brain does right you're producing kind of a different simulation of reality based on the reality that you've already seen so seems pretty intelligency to me I don't know that's my two cents Allan you're so bummed out you gotta go don't you yeah this I I have an IT division meeting but uh so far the CIO is just presenting

material on Microsoft multifactor authentication so no reason to jump immediately uh but you know maybe AI will replace that someday but um I guess my biggest criticism of of what we see labeled as AI is that it turned out to be so universally useless for anything except bad plagiarism so you know if it's conscious it's not doing a very good job of it um uh but I will concede for this point that neither do we and uh myself included but I'll leave you guys to explore I I I hope you have a good time it's no criticism that I'm leaving I just have to go

listen to the CIO yeah thanks for being here Alan thanks for starting the conversation always a pleasure so Brian you had a little message over on the side here what what what is this question here are we in a GPU bubble is that so I guess for our next topic uh AI predictions in 2024 and basically the hot the very hot thing right now is everyone's trying to Cha uh train their AI models right so uh I POS the question like at what point is the base model good enough where people are just fine-tuning it and the need for these gpus will eventually decrease and

we're g to be left with you know data centers full of gpus where we don't really know what to do with them so uh you know what and then as we as time goes on with these trained models inference will become more important so are we going to be developing more inference chips that are low power to actually leverage these models and uh what happens to these gpus uh that were used for training yeah good question Fernanda you go a you got a response here so are we in a GPU bubble not if I can help it is my first uh the concise version of

the answer uh it's the reason why I'm a Voltron data it's the reason why we're working on on uh moving um analytics and data PR processing to the GPU to you know do all the other workload things that you can't do uh on the GPU yet um I think we underestimate the amount of entertainment that comes through these things um we are over um we're probably putting and uh too much of a a weight and whether or not AI is going to be useful to society there are many things that we spend a lot of compute power on um you know there are there's a

lot of the internet traffic and network that is not useful and we all know what that traffic is and I'm not gonna talk about it here right there's a ton of storage that goes to a lot of videos uh for websites that are not safe for work so the internet is not uh necessarily here to be useful to humanity in in the sense that we scientists and we folks in HBC are here for so I think we underestimate the value of entertainment that AI is going to bring to people and generating art and generating music and everything else that comes with it that's one two

it's still young we had all of these methods developed right so these are not new methods but we never never had the compute power to do it so now we do and now we have to redevelop these methods we have to take the methods we developed we have to now do the compute like finally we can do the compute on it but now it's a next step now we redevelop these methods right we we refine them we take them in a different direction it's still very young um so we need to you know kind of give it a little yes it's chaotic yes it's wrong

yes it's horrible but like give it a little time everybody calm down like it doesn't have to turn around usefulness immediately let the novelty chill down a bit and then lastly we don't there's I wanted to get a thousand gpus if somebody knows out there where there's a thousand gpus that I can run a benchmark for our internal system let me know because we can't find it we can't find it on the cloud we can find it through National Labs we can't find it you know in in universities there are people out there using gpus Beyond mlna I and that's ever growing as tools continue

to move on thanks to you know the vision that Jensen had to create an entire ecosystem called cuda to make this possible in the future we're going to have an entire ecosystem probably called hip and in the future we're going to have another one for pun Veo this is not going away if anything I don't think it's a bubble that's going to go away Nvidia might go away because other things will you know pop up so they're their grip on the on the economy of gpus might go away but I don't think that that the GPU itself is going to you know disappear Fernando you

touched on something there that I wanted to touch on and that's the dominance of Cuda in the market what are your thoughts on some of these efforts that are out there to kind of make some of this you know Cuda is very entrenched everything has been built on it for you know 10 years or so now what do you think of some of these efforts that are out there to like more generally like I just saw the other day this software effort where they wanted basically do translation of native CUA calls into like rock calls on AMD gpus to kind of get out of the

Nvidia Silo have you seen any kind of Market push for some of the or like Intel sickle that type of thing have you seen any Market push for those software solutions to get around Cuda yet or is it still mostly just Nvidia dominance so the only reason two things uh two thoughts that I had as you were asking the question first we've tried many of these um versions of platforms that claim to run on many other platforms right opencl being the first one that came out when when GP GPU had come out the problem is is that when you try to build a a framework

for a lot of things you end up being bad at a lot of things yeah you know performance was really hard and then and then you had vendor games Nvidia was one of them that that you know would um not really put any effort into making opencl run because they wanted to establish their Aruda and know uh platform which you know it's a vender game vendors are going to vendor right um the only drive to replace Cuda Cuda is an amazing platform the only dri to replace it is economic right and it's because this this one company has a hold on the market um and

so what other companies the best way that we would get out of this is if other companies can start building the same primitive libraries that Cuda has put effort to and that costs a ton of money I mean nvidia's got 10 years on this right um and and then that will give them an edge or at least a slice of the pie uh where you know Nvidia has pioneered it they don't have to do everything Nvidia does they just have to do enough to be able to eat a slice and a chunk of that pie and and build these libraries so you know like a

simple one for for uh uh for you know starters in any platform is you know things like mathematical libraries right um another one is like things like Shuffle and how you manage memory right those sorts of things so you don't have to do everything but you have to do a lot to be able to even like take a bite and a chunk out of what Nvidia is built very interesting thank you go ahead Dave yeah back to your original question Brian is if if we're in a GPU bubble and there's there's multiple different aspects of that I mean one of them that is kind of

coming up in chat a little bit is um whether or not uh you know I mean we might be in a GPU bubble just because the technology might shift to something else it's not technically a GPU but still serves the same purpose like a TPU or something but um I I think more to your point is like that's a really interesting idea that at some point we will have performed all the training that we need to to have a model which is good enough that it can perform whatever inference we need I mean I guess to do that we would have to step away from

specialized models and we'd have to have a more general purpose kind of like intelligence which is able to do a lot of different things which you know I I think we're kind of far away from that still um the other thing too is and this is kind of Michael Young has brought up an interesting uh uh point in chat and I think it's kind of similar to that so what happens when you start so so what's the end of learning right so um we as humans uh continuously are sort of learning and building upon our knowledge and stuff like that what happens when you start

to take uh inference that comes out of that end of it and you start to feed that back in to start building the models you know um I think that's already been done to some extent but you know I mean when you start to do that when you start to actually use the the technology to build itself you could see how you never get to a point where uh you don't need to do training anymore because the models can always be improved I guess my big point you know where does the model end in the immediate kind of practical societal implications of that is you

know what end percentage of people do you want to automate you know out of whatever we're doing and that's you know neither here nor there you know if we're prepared to deal with that as a society but you know as our models get better and we you know you know Sam Altman at you know at Al would have you believe AGI is not as far off as you perhaps some of us would like as far off as I would like to imagine it is that you know um and so you know for me I think the I definitely see that you know from the philosophical

angle when does learning stop well never stops we always have new things that we can learn but for in the immediate you know we already have so many concerns about Automation and stuff like that it really just comes down to me of how much can you know how far can we take you know generalized models um before we start to have you broader societal effects because we're automating you know millions and millions of people out of you know their jobs with no real you know plan for what those people are then going to go and do so be interesting to see kind of the intersection

between how models get better and more Specialized or how they get more generalized how kind of that tradeoff with how it starts to affect automation broadly and how people are working H yeah go ahead Fida so I know that for sure AI is not going to pre you know cure my cold that I have right now I know that AI for sure is not going to replace my plumber it's not going to replace my car mechanic right yep there there are things that that are still going to need um human touch and but they can be argumented by AI you know uh there is a

noise in my engine mechanic and we can we can probably train some you know some AI to figure it out and say it's likely this without having to have 20 years of experience having listened to a bazillion engines to figure that out um but we've also gone through these major periods of inventing things like the cotton Jin and other things that put you know thousands of people out of work the the difference here in this time in place is that there are eight billion people in the world right now and putting a million people you know out of work 10 million people out of work

a 100 million the impact of that will be much more destabilizing than the impact of you know a few thousand or t of thousands of people right uh and that's I think my worry more is the idea that we can put 10 million out of work and now we have no plan for them that's going to be destabilizing to many countries yeah exactly and agree completely that um K mechanics plumbers anything that requires robotics will probably be some of the last if text based systems can't eat it alive and it has to somehow coordinate something like that it'll probably be a significant distance farther down

down the line than more you know computer typing sitting at the terminal all day based careers I imagine but yeah exactly and that's a good point with like the cotton genin and stuff like that there have been periods where things have suddenly changed it's just interesting how highly specialized the changes are starting to become There Was You Know basic technologies that really kind of set that kind of moved things from a very very basic level of technology to a very you know still primitive to where we are today but revolutionized it it's interesting to see how AI changes you know it's much more of a

threat it seems like to these more specialized fields in the end but anyway I think if governments can figure out as long as people are fed and housed and they have water to drink for the most part things are going to be okay uh it's it's if profit driven you know incentives not only put people out of work but also make them hungry cold and thirsty right and those are going to be very destabilizing wow we got dark really quick yeah I know I mean the reality of it is you know there is no automation talks you know I've always heard people say oh it's

Universal basic income that'll save this type of thing you know based on an automation tax I see the AI advancing I don't see the you know safety net for the automated people advancing at the same amount so we'll see how it you know it turns out it's to quote myself again from like 20 minutes ago is I am much more worried about a politician using AI than an AI becoming a politician exactly I just this is kind of a side thought too but I I just want to um kind of acknowledge I don't think either of you meant this in this way but the cotton

Jin is kind of a loaded example it didn't really put people out of work maybe so much as it just shifted the the labor force I don't know just want to kind of acknowledge that you know no you're correct you're correct call it industrial production of texy right I was actually thinking of the mailman like when we got email right it was like oh we we don't need you know the United States Postal Service anymore but quite frankly I think they're still kicking it they might be struggling a little bit but they're still kicking and they're bringing me packages they love to bring me packages

I don't really get mail like letters anymore you know other than spammy type stuff um but I think that it it's also possible that and this is this is the case I think for for all of us and all different circumstances is that we're looking at our now and projecting things into the future of you know if this changes then you know this is going to change but forgetting that there's a whole bunch of other things that are going to change along with it and so we just can't really be sure but one of the things I'm just putting my little two cense out here

one of the things that I I really like in terms of like you know should we do this or shouldn't we do this is that you know if whatever it is that you're doing like is a good idea for all people to be doing then it's a good idea and if it's really only a good idea for you and maybe 15 people you know maybe that's not such a great idea to move forward with whatever the the new idea is throwing that out there yeah I mean I I feel like um there was a period of time during which a lot of folks me included

when the internet first you know started to impact Our Lives um that some of us sort of was like oh the internet that's a cool thing sort of a fad probably be popular for a little while it was called a fad you remember that yeah and I feel like um you know maybe um AI is kind of similar to that I think that down the road it's going to uh you know it's G to have an impact on everybody's life like right now it doesn't matter I mean there's there's folks in our society who don't have computers who don't get on the internet who don't

have email addresses doesn't matter the internet still affects their lives right it keeps the electricity on in their house or something it's there's there are ways in which they're affected I think AI is going to be the same way you don't have to interact with it directly at some point it's going to have an effect on everybody's life it's going to be like internet level of changing okay so let me look at my little list here Ai and automation oh that's interesting we're kind of at 11:36 so let's actually Okay so we've talked about cautions and concerns just real quick because Fernanda I I think

you're right you know maybe we can just kind of like zip it around because of all things there's there's kind of a balance so what are you actually excited for when it comes to AI oh I think you're still on mute sorry about that um I'm excited for a lot so one of the things that I'm excited about is um the the level of of productivity and just overall um bringing joy to a lot of sort of uh drudgery that comes with work especially in development right in development work uh yesterday I was having dinner with some folks and um and the person I was

having dinner with said you know it' be really great if AI could create input files for my simulations because every time I have to create an input file for the simulation it's it where where is it I know I had built something similar I have to kind of start from scratch I'm like oh my God that's a genius like we can just train a whole bunch of them to learn input files and you can just say you know I was a molecular Dynamics like it took me seven months in grad school to build an input an input file for uh material that was like a

crystallin type of Styrofoam called you know isotactic polystyrene is the the technical term and it took me forever to build that uh the boundary conditions were wrong the the entire thing would explode in my computer and I'd have to go back to it it took me seven months so I think it would be great if we could um automate some of the science and I and I'm really I'm really excited for the possibility that this automation might actually lead us to a point where we can help the climate that's the one thing that I think we could really uh and and to you know to

to make another comment from last night's dinner I I was dining with somebody who's a climate scientist and for the first time I heard a scientist say I'm really excited for AI because I think it'll really help with climate models and because AI is probabilistic and my models are probabilistic also so you know usually scientists are kind of anti- AI because well they live in a deterministic world and their simulations are deterministic and so a probabilistic thing coming in giving them wrong answers makes them very ansy but in the world of climate science a lot of the models are probabilistic so can we can we

figure out ways that if we could just turn off you know let's say we turn off the impact of 30 million cars on the road uh let's say we you know tell governments here look if you just incentivize if you could take 15% of the cars on the road and turn them into bicycles or you know e Vehicles that's going to slow this whole thing down and then the whole science of like batteries and uh and even carbon uh you know capture carbon SE sequestration would be uh you know sped up by the possibility of AI just because it can crunch through data much faster

than we could uh by hand so those are the areas that I think I feel very hopeful that AI can help speed up something that's very imminent uh that's you know that's really uh affecting millions of people today uh it's also I'm I'm not gonna lie it's kind of fun to create you know aski to like create I don't know uh you know baby dinosaurs riding a bicycle like you know it's it's fun it's it's a joy that's a good point thank you Brian what are you excited for in the AI world uh mainly the I what I'm most Ed for is mainly the application

in medicine and you know the types of you know diseases that we're going to be able to cure in the future um this probably mostly in the genomic space and yeah yeah I think making people have a better quality of life is overall a very good thing and yeah excited to see where this goes awesome thank you Forest well firstly I'm excited I no longer have to spend hours on prototyp typing said O command Combos and that type of thing it like yeah like Fernanda said the ability to iterate much faster on development just on a day-to-day basis because you can you know prototype scripts

that type of stuff much faster is very cool so I'm just like a day-to-day basis it you know I have I have been resistant to gradually implementing it in my workflows but I am the point where I'm like yes it is nice to not have to do that type of thing for hours um on a serious note like everyone touches on here I think we've really just only scratched the surface of you know even outside of some of these you know flashier use cases what we can do with this I was just on I think it was a Microsoft presentation yesterday that I was listening

to where they were talking about discovering new tuberculosis drugs by feeding in the smiles representation which is like this text based representation Fernand I see ning I think you probably know what I'm talking about yeah actually I did it during covid trying to find candidates yeah exactly there you go yep precisely yeah so they you know if it's text based llms really have incredible stuff that they can do with it yeah um and so yeah and and also like Fernand says it is fun to just kind of you know sometimes you're just sitting there you know the dolly terminal on opening eye and you're just

like wow you know boy I I could just create anything you know logos it's it's you know it is kind of fun to play with but so it's you know very interesting applications we'll see how it plays out you know for those of us who lackluster artistic abilities right like I have this idea in my mind but can't draw it and so I know I love artists and I I know that this is fraught with like conflict and controversy but for the first time I can imagine something and it pops up right and it's so cool it's pretty cool I agree Gary what are you

excited for an AI oh you know I I take a very pragmatic view of AI I it's a great tool and as people have mentioned uh you know it can help you you know generate your data sets a lot of people use it for coding all the time now see like all the all the students coming out of college you know this is stuff they know and it makes me a little bit worried you know when we think about like uh you know how people are getting replaced I I am worried there there's like a whole you know the majority of the world's population has

no clue how to use any of these tools you know and so um uh but uh you know right now we're we're still at the point where you know as you if you ask AI to write you anything you know it'll put out something that'll give you a start but um you know you still have to kind of go through it and I think that's the same going to be the same for a lot of other things I mean you can see it with like self-driving cars you just can't make the final step you know the uh as fernandoa was saying about like having probabilistic

probabilistic outcome um suggestions you know it's not really all that you know that can't get you the final step to like a science outcome and so it'll just say oh these might be possibilities but you know you it's nothing you could publish in a paper and so um and then and then same for like uh like the uh like even an auto mechanic or or a doctor uh you know they're they're going to do things based on experience like touching feeling for something looking for something out of place that may be really hard to uh replicate uh so um I think but I'm I'm really

looking forward to this awesome thank you Gary Mr God love I Echo forest and that I am pleased that I will never actually have to learn regular expressions like on a deep level because I can just ask AI to give me something that'll detect this and it'll do it for me um um and that's a you know admission that I've never really learned regular Expressions on a deep level sorry um oh I don't know Dave I think you're pretty expressive well thank you but not regular um um yeah I guess um I guess what I'm most kind of excited about as far as AI goes

is watching the interplay like as it like as AI gets to the point where it actually starts to have serious impacts on society and humanity and stuff like that I mean society and humanity and all that kind of stuff are going to react to that and the two are going to develop in parallel to some extent I mean you can see exactly what happened with the internet right if you look back to at the early 90s you look how how different and how the ways in which society has like um kind of reacted to and grown in parallel to and in some in some ways

really in kind of not so great and scary ways but in some ways really awesome ways too I feel like AI is going to do the same thing um to society as well and you know I'm excited and a little scared also to see how um Ai and Society kind of develop in parallel and interact with one another yeah awesome thanks you guys H fernandoa as you were talking and one of the things that you said was um you know I have a little bit of a cold so by the way also than you for showing up anyway and and hanging out with us even

if you're not 100% but you said it's not going to help me with this cold and like my first reaction was I mean maybe it could right if there was some kind of Scan they could do if it was because of some kind of nutritional deficiency or like maybe you need a little more sunlight I don't know like maybe it could tell you that anyway that got that is what got me excited about um any you know possibilities of AI and how it can benefit our lives like on the the health aspect I me the the thing about the thing about diseases in nature is

that as soon as you figure that out something else will come along and so it might help me with this particular cold but future colds it won't help me absolutely not oh I hear you how fun all right you guys AI at automation W if you guys want to continue this conversation uh you know ciq is well-versed in the underground supporting all of this kind of cool AI work you know we support Rocky Linux we got werewolf we got Apper we got Ascender automation platform we got lots of things to talk to you about to help you U pushing your AI initiatives uh in your company so reach out to us cq.

comom we would love to chat with you um you fill out any form there God bless you Forest um you fill out any uh form on there is going to come to me we'd love to chat with you again thank you so much everybody on the panel blessings have a wonderful day and we'll see you next week same time same place bye thank you all bye [Music] everyone