IBM did survey of 5000 businesses and 82% of them was to adopt AI but the number one hurdle with explained ability the executives and business unit owners in those companies didn't feel that AI was sufficiently transparent for them to for it to fit within their their workflows. The other one is is that modern techniques are overly reliant on a lot of labeled data so I think one of the biggest challenges is you know when any a I startup or even a incumbent company good in to work with an enterprise the first question they have is you know where's all your data is it clean is it labeled is there 50,000 or some cases 500,000 or a million examples that I can learn from and the answer that question is typically no so that's why you see a lot of data labeling companies doing a lot of work now labeling that data. So there are the two biggest hurdles is because with that label aspect of it it just takes so long to deploy these systems that you know but two years in you're not getting any value you haven't even label katraine businesses yet.

That's gonna be a very big industry right now is going through and relabeling all that data you figure years and years and years of collecting that with no actual way to have that be available as an asset for for artificial intelligence to even learn off of it I've also feel like a lot of companies are incorporating the name AI into things but you know back to that they don't really know what a is you know it's like oh yeah we got AI in that thing you know but what is AI yeah yeah there's there's and there's both two sides of that - because CEOs will tell their executives go find the AI yeah of course you know people are gonna say III have it for you I mean I think I think any I think AI a disability for machines to learn but also to read it and I think people a lot forget that next step because this current wave of AI that we're seeing these statistical machine learning techniques are phenomenal at learning from data but they don't read them very well and I'm talking about reasoning in the historic symbolic a I sense of inductive deductive a deductive analogical reasoning these kind of things that we can do is you would be so I think of AI is the disability to learn from data but there's also reason with that data or reason with the knowledge that you acquire from that data and put that knowledge together in new and novel ways and create ultimately new knowledge.

"WEBVTTKind: captionsLanguage: encontinuing back with more digital trends live artificial intelligence is something we love talking about here and increasingly it is such a part of technology no matter what facet you're in and we have an expert on artificial intelligence from kindy we're now joined by Ryan Welsh hello Ryan I don't think I get quite hear you ok and we are joined now by Ryan Welsh Ryan can you hear me I can hear you now there we go see we need the AI to fix this first let's talk so Ryan thanks for hopping on here today and I want to talk about a lot of things you know that your company is doing so let's first go to this why don't we talk about what kindy is and then maybe get into talking about the first explainable AI or whatever order you want to do with that yeah sure so first first the the company so we're an AI start-up based in San Mateo and we build explainable AI products for government financial services and healthcare and we exist because those industries can't use purely machine learning statistical machine learning techniques because those techniques are generally black boxes so effectively how do you build systems that can get past regular regulatory requirements and also just gain trust of users in those industries okay so it's just kind of helped everybody out there and you know in getting to understand this and that's the thing I think for AI for a lot of people it's such a broad concept that it is it is a difficult thing you know to really wrap your mind around what it is and where it's going how did how did this company come to be like how did you get this started yeah a graduate degree in quantitative finance I was working for a law firm during a financial crisis and effectively we had to read a bunch of information to help our clients get unwind a bunch of esoteric credit services and in three days I had to read amount of information there I left for business school three years later I was still reading so effectively how do you build machines that help us consume that information and also we make decisions faster instead of taking three years maybe take the actual three days so when you work with these systems they don't really work well on language and they can't explain to you while there why they're making recommendations why they're bringing back certain search results so kind of bringing all that together which is why I started kindy in June 2014 and why we've been going gangbusters ever since because the kind of hypothesis of you need explainable systems you need to work on natural language you know we've been able to bring that market that's that's actually a really great concept and that is something that I think it's gonna be necessary going forward you know the more complicated this gets if we can't understand it that there's something I think there's some kind of a maybe a human element in there where it's like we wanted to we want the AI to explain it to us in our language you know not just yeah here's yeah here's so interesting is the fact that you have me on and are asking the question why is exactly why we need it yeah right it is I you know I've been I've been doing a talk recently where I get up on stage and I say that for AI to thrive it needs to be explainable and then I walk offstage and everyone's like wait a minute network I would just walk off the stage and and there's something about being you know what makes us human is this ability to ask why and this desire to ask why and every time someone makes a statement your first reply is why it's because we want to interrogate and understand the thief the person's belief system and their their logic so that we can ultimately determine whether we believe them whether we adhere to their principles and ultimately we can we can gain trust with that individual we don't have machines that can provide that and it just really won't be able to fit within our you know workflow as human being yeah I mean you don't want to take things at face value you want to know exactly yeah how that happened that's that's a good point I mean I didn't even think about it that way but just human nature like okay cool you told me this how how did you find this out you know how did you determine this what was your process well what are some of the things that we wanted to talk about here too cuz you had brought this up as what are the top two issues for enterprises specific to AI yeah well number one is explained ability there an article recently where IBM did survey of 5000 businesses and 82% of them was to adopt AI but the number one hurdle with explained ability the executives and business unit owners in those companies didn't feel that AI was sufficiently transparent for them to for it to fit within their their workflows but then the other one is is that modern techniques are overly reliant on a lot of labeled data so I think one of the biggest challenges is you know when any a I startup or even a incumbent company good in to work with an enterprise the first question they have is you know where's all your data is it clean is it labeled is there 50,000 or some cases 500,000 or a million examples that I can learn from and the answer that question is typically no so that's why you see a lot of data labeling companies doing a lot of work now labeling that data so there are the two biggest biggest hurdles is because with that label aspect of it it just takes so long to deploy these systems that you know but two years in you're not getting any value you haven't even label katraine businesses yet that's it yeah that's gonna be a very big industry right now is going through and relabeling all that data you figure years and years and years of collecting that with no actual way to have that be available as an asset for for artificial intelligence to even learn off of it I've also feel like a lot of companies are incorporating the name AI into things but you know back to that they don't really know what a is you know it's like oh yeah we got AI in that thing you know but what is AI yeah yeah there's there's and there's both two sides of that - because CEOs will tell their executives go find the AI yeah of course you know people are gonna say III have it for you I mean I think I think any I think AI a disability for machines to learn but also to read it and I think people a lot forget that next step because this current wave of AI that we're seeing these statistical machine learning techniques are phenomenal at learning from data but they don't read them very well and I'm talking about reasoning in the historic symbolic a I sense of inductive deductive a deductive analogical reasoning these kind of things that we can do is you would be so I think of AI is the disability to learn from data but there's also reason with that data or reason with the knowledge that you acquire from that data and put that knowledge together in new and novel ways and create ultimately new knowledge how far away do you think we are from from that exact kind of system we're we're very far away from artificial general intelligence you know I there was a article where where he was Kurzweil say in 2029 event Brooks it was saying twenty two twenty two hundred that III put it out there they're pretty far I mean if you're in the industry and you work with AI systems you understand how limited they are specifically around sensory motor and natural language understanding or comprehension of language systems are very good at parsing sentences but not really good at understanding the semantics and the pragmatics of language that's a that's an interesting perspective especially coming from somebody you know in the industry because I feel like a lot of people are worried you know we're gonna reach singularity and the AI is gonna take us over in a few years but you're saying that's not gonna happen for a while yeah and I think this is where the industry gets gets in trouble is because I think the breakthroughs that we have have been phenomenal yet compared to the height that we're throwing behind it it's just it's just not there so we're like the the hype is exceeding the actual applications of these these technologies and that kind of mismatch I think will ultimately send certain techniques into a AI winter and I think that the people that are going to prevail are going to be the the companies that realize that AI is a feature of a product not the product itself so you gotta go in you gotta solve real business problems or people problems and hopefully we have a idea feature of the product not the product so very good point well Ryan I want to thank you you know for having a for hopping on the show here today with us and where can people follow can and and follow everything that you do as a company yeah yeah Kimi comm is a great place to check in on our AI research white papers that we have out kindiy Tech Twitter and of course on on LinkedIn as well so that's that's everything thank you yeah Thank You Ryan Ryan wells joining us right here on digital trends live have a great daycontinuing back with more digital trends live artificial intelligence is something we love talking about here and increasingly it is such a part of technology no matter what facet you're in and we have an expert on artificial intelligence from kindy we're now joined by Ryan Welsh hello Ryan I don't think I get quite hear you ok and we are joined now by Ryan Welsh Ryan can you hear me I can hear you now there we go see we need the AI to fix this first let's talk so Ryan thanks for hopping on here today and I want to talk about a lot of things you know that your company is doing so let's first go to this why don't we talk about what kindy is and then maybe get into talking about the first explainable AI or whatever order you want to do with that yeah sure so first first the the company so we're an AI start-up based in San Mateo and we build explainable AI products for government financial services and healthcare and we exist because those industries can't use purely machine learning statistical machine learning techniques because those techniques are generally black boxes so effectively how do you build systems that can get past regular regulatory requirements and also just gain trust of users in those industries okay so it's just kind of helped everybody out there and you know in getting to understand this and that's the thing I think for AI for a lot of people it's such a broad concept that it is it is a difficult thing you know to really wrap your mind around what it is and where it's going how did how did this company come to be like how did you get this started yeah a graduate degree in quantitative finance I was working for a law firm during a financial crisis and effectively we had to read a bunch of information to help our clients get unwind a bunch of esoteric credit services and in three days I had to read amount of information there I left for business school three years later I was still reading so effectively how do you build machines that help us consume that information and also we make decisions faster instead of taking three years maybe take the actual three days so when you work with these systems they don't really work well on language and they can't explain to you while there why they're making recommendations why they're bringing back certain search results so kind of bringing all that together which is why I started kindy in June 2014 and why we've been going gangbusters ever since because the kind of hypothesis of you need explainable systems you need to work on natural language you know we've been able to bring that market that's that's actually a really great concept and that is something that I think it's gonna be necessary going forward you know the more complicated this gets if we can't understand it that there's something I think there's some kind of a maybe a human element in there where it's like we wanted to we want the AI to explain it to us in our language you know not just yeah here's yeah here's so interesting is the fact that you have me on and are asking the question why is exactly why we need it yeah right it is I you know I've been I've been doing a talk recently where I get up on stage and I say that for AI to thrive it needs to be explainable and then I walk offstage and everyone's like wait a minute network I would just walk off the stage and and there's something about being you know what makes us human is this ability to ask why and this desire to ask why and every time someone makes a statement your first reply is why it's because we want to interrogate and understand the thief the person's belief system and their their logic so that we can ultimately determine whether we believe them whether we adhere to their principles and ultimately we can we can gain trust with that individual we don't have machines that can provide that and it just really won't be able to fit within our you know workflow as human being yeah I mean you don't want to take things at face value you want to know exactly yeah how that happened that's that's a good point I mean I didn't even think about it that way but just human nature like okay cool you told me this how how did you find this out you know how did you determine this what was your process well what are some of the things that we wanted to talk about here too cuz you had brought this up as what are the top two issues for enterprises specific to AI yeah well number one is explained ability there an article recently where IBM did survey of 5000 businesses and 82% of them was to adopt AI but the number one hurdle with explained ability the executives and business unit owners in those companies didn't feel that AI was sufficiently transparent for them to for it to fit within their their workflows but then the other one is is that modern techniques are overly reliant on a lot of labeled data so I think one of the biggest challenges is you know when any a I startup or even a incumbent company good in to work with an enterprise the first question they have is you know where's all your data is it clean is it labeled is there 50,000 or some cases 500,000 or a million examples that I can learn from and the answer that question is typically no so that's why you see a lot of data labeling companies doing a lot of work now labeling that data so there are the two biggest biggest hurdles is because with that label aspect of it it just takes so long to deploy these systems that you know but two years in you're not getting any value you haven't even label katraine businesses yet that's it yeah that's gonna be a very big industry right now is going through and relabeling all that data you figure years and years and years of collecting that with no actual way to have that be available as an asset for for artificial intelligence to even learn off of it I've also feel like a lot of companies are incorporating the name AI into things but you know back to that they don't really know what a is you know it's like oh yeah we got AI in that thing you know but what is AI yeah yeah there's there's and there's both two sides of that - because CEOs will tell their executives go find the AI yeah of course you know people are gonna say III have it for you I mean I think I think any I think AI a disability for machines to learn but also to read it and I think people a lot forget that next step because this current wave of AI that we're seeing these statistical machine learning techniques are phenomenal at learning from data but they don't read them very well and I'm talking about reasoning in the historic symbolic a I sense of inductive deductive a deductive analogical reasoning these kind of things that we can do is you would be so I think of AI is the disability to learn from data but there's also reason with that data or reason with the knowledge that you acquire from that data and put that knowledge together in new and novel ways and create ultimately new knowledge how far away do you think we are from from that exact kind of system we're we're very far away from artificial general intelligence you know I there was a article where where he was Kurzweil say in 2029 event Brooks it was saying twenty two twenty two hundred that III put it out there they're pretty far I mean if you're in the industry and you work with AI systems you understand how limited they are specifically around sensory motor and natural language understanding or comprehension of language systems are very good at parsing sentences but not really good at understanding the semantics and the pragmatics of language that's a that's an interesting perspective especially coming from somebody you know in the industry because I feel like a lot of people are worried you know we're gonna reach singularity and the AI is gonna take us over in a few years but you're saying that's not gonna happen for a while yeah and I think this is where the industry gets gets in trouble is because I think the breakthroughs that we have have been phenomenal yet compared to the height that we're throwing behind it it's just it's just not there so we're like the the hype is exceeding the actual applications of these these technologies and that kind of mismatch I think will ultimately send certain techniques into a AI winter and I think that the people that are going to prevail are going to be the the companies that realize that AI is a feature of a product not the product itself so you gotta go in you gotta solve real business problems or people problems and hopefully we have a idea feature of the product not the product so very good point well Ryan I want to thank you you know for having a for hopping on the show here today with us and where can people follow can and and follow everything that you do as a company yeah yeah Kimi comm is a great place to check in on our AI research white papers that we have out kindiy Tech Twitter and of course on on LinkedIn as well so that's that's everything thank you yeah Thank You Ryan Ryan wells joining us right here on digital trends live have a great day\n"