ai.bythebay.io: Mike Tamir Interview

[Music] [Music] so AI has gone through a little bit of so AI has gone through a little bit of it you should usage transformation last it you should usage transformation last 12 months historically I think it's been 12 months historically I think it's been integrated AI that means a machine that integrated AI that means a machine that can do several different human-like can do several different human-like things tasks and it has those abilities things tasks and it has those abilities so being able to chat being one of them so being able to chat being one of them until recently I think I have thought it until recently I think I have thought it as I've thought of it in terms of deep as I've thought of it in terms of deep learning and the reason for that is that learning and the reason for that is that deep learning has really helped crack deep learning has really helped crack open a lot of those capabilities open a lot of those capabilities nowadays it's becoming quickly nowadays it's becoming quickly synonymous with just machinery so in synonymous with just machinery so in order to an integrated AI you to create order to an integrated AI you to create a machine that can do a lot of the a machine that can do a lot of the things core things that human could do things core things that human could do you would need to be able to sense you would need to be able to sense things so maybe look at images and things so maybe look at images and recognize them you'd be able to to hear recognize them you'd be able to to hear things so so voice voice to speech or things so so voice voice to speech or speech to text and maybe you need to be speech to text and maybe you need to be able to comprehend so natural language able to comprehend so natural language understanding and you may even need to understanding and you may even need to be able to physically interact there be able to physically interact there have been huge advances in each of those have been huge advances in each of those directions recently so Q learning for directions recently so Q learning for human interaction combinations of human interaction combinations of advanced neural embedding for natural advanced neural embedding for natural language understanding convolutional language understanding convolutional neural nets have really driven the image neural nets have really driven the image recognition pretty significantly for the recognition pretty significantly for the last several years and for 1d last several years and for 1d convolutions also especially some work convolutions also especially some work that Baidu's been doing recently for for that Baidu's been doing recently for for speech to text so we've made a lot of speech to text so we've made a lot of advances in all of those directions advances in all of those directions nothing's perfect editors there I least nothing's perfect editors there I least I haven't seen any good integrated I haven't seen any good integrated a project that do all of those a project that do all of those simultaneously so I I work in several simultaneously so I I work in several different areas you know intact we do different areas you know intact we do the backend AI for for large enterprises the backend AI for for large enterprises so using neural networks for so using neural networks for recommenders for you as our recommenders for you as our understanding for making sure that your understanding for making sure that your that a company can interact with a with that a company can interact with a with a a customer the right way so not just a a customer the right way so not just trying to get that customer to buy trying to get that customer to buy another bag of chips but creating a more another bag of chips but creating a more long-term reinforcement learning type long-term reinforcement learning type relationship is where a lot of my focus relationship is where a lot of my focus has been recently so so the by the bay has been recently so so the by the bay conferences are great I love conferences are great I love participating in these and I always participating in these and I always learn something new I see I see folk learn something new I see I see folk that that that I enjoy meeting with that that that I enjoy meeting with regularly and the Lexx is great also regularly and the Lexx is great also [Music]