What is the logic behind all the Voice Assistance?

Nowadays it doesn’t ask us to lock it automatically recognizes people in the photo and whatever tagging that he had done provided for the learning for this algorithm so I got them learned to use that data and now it has become proficient enough to start matching human-level accuracy so this algorithm for accuracy is now 99.1 4% while human accuracy is just tired ahead at ninety-nine point three zero percent right good.

So now obvious single Gotham’s which are mirroring human-level accuracy but when we say narrow eye it is more about doing a specific task and doing it well the second stage of AI is generally I like in the example here and more let’s say your voice assistance like Alexa Siri or Google Assistant right so Erin you can ask you can talk to your voice assistant and ask that voice assistant a plethora of things you can ask it about the weather about a school for a game you can ask you to set an alarm right and later you can also ask follow-up questions with algorithms.

If you say something in some context ask a question some context and follow up another question the algorithm will retain the context it will know exactly what is it what context are you talking about right so so this is more, so these algorithms are more purpose to handle a variety of tasks and use interlink knowledge from one domain to another so then so so that was the example that is the second stage is Istanbul AI and the third stage of AI is super AI right in which kind of now goes into the realm of science fiction because right now we don’t have algorithms which are which can be categorized as super AI.

What super a mean says that sentient machines which have their own intelligence which have the context right which can act independently and are inherently smarter than humans right because once you impart intelligence to machines like the kind of data reading rate that they can access the kind of memory that they have the kind of things that they can link even now far exceeds that of a human so if you add intelligence on top of that then these machines become smarter than humans, so this is again this is on the horizon nobody knows what’s the fine line to this right and this is now going to the realms of but right now it doesn’t anything like this exists and it is now going to realms of science fiction and you know you know movies like dominated, etc so I just want to quote hang here.

In doing as you all know is a leading you know researcher and faculty in AI and machine learning across the world, so his coat is I worry about super AI in the same way that I worry about the overpopulation of mass right so from this code we can understand that this is a very far-fetched idea at the moment although from the recent trends what observed is that been a belief whenever we think of an idea is far-fetched.

We have been surprised and ideas have come to fruition much much sooner than we ever thought they would for example the voices sustained you know these days there’s a voice assistant in all our all of her pockets you know in our phones in terms of either Google an assistant or City it is there for all of us to access but the whole idea about a wise assistant or up coherent product which was demoed was only in 2016 right in Google i/o conference.

So within three years from becoming Justin from being just an idea this kind of technology is now ubiquitous it just resides with everyone we can access it like it is deployed at production scale and all of that like then it is very very accurate as well right similarly that say when I heard personally of Amazon go stores so I’m not sure how many of you are familiar with Amazon goes towards promises on both stores are essentially stores which are cashier-less so again the idea was coined in.

I think two years ago and the idea was that you can just walk into a store to pick anything from the shelf and then just walk out and you’d be automatically billed right and imagine the kind of accuracy need to have the kind of algorithms at play to be able to decipher what have you bought right an accuracy that you need to correctly will the customer right because of you go wrong building you know the product like this can never succeed you cannot be billing customers with the wrong amount right.

I thought personally when it was the cool idea was coined two years ago that it will take another five-six years for it to be implemented right but do have a surprise and many people are aware that we didn’t do yours last year late last year Amazon actually opened some of these stores in San Francisco in the Silicon Valley area of the United States in San Diego and La right so so we are now constantly being challenged on how much can we think how far ahead of anything right and whatever we think is going to think to take X amount of time is now happening much much sooner.

Yeah, so it does super AI do seem like a thing of the future but again we may be surprised and it may just happen much sooner cool so from that context on what is and what are the stages of AI let’s also try to understand some of these terms which are now used by everyone and there is a lot of overlap in these terms right beef everyone is talking about artificial intelligence everyone is talking about machine learning about deep learning right but I’m not sure that everyone knows about what is the distinction between using AI or machine learning or deep learning, okay, so AI basically is the umbrella term for anything which fit which machine does which can be termed as intelligent right so anything with where it is independent and taking its own decisions and be called as artificial intelligence.

It can be very rudimentary can be rule-based as well right even if you let say you know program your microwave for you know fully to run differently for different kind of food that also has intelligence like automatically if it runs so it can be very rudimentary but any kind of behavior which a machine is displaying which you know does tasks in an automated fashion where some calls have to be taken of the machine is taking it independently that has coined as article intelligence within that realm right there is a subset of that which is machine learning.

Machines which are powered by ML Algar so data science algorithms which are a slightly more complex kind of algorithms right which involve an element of learning rate which and by learning what I mean is that if you run these algorithms multiple numbers of times every run these algorithms will become better and better either this and learning you there’s a feedback loop by so what the algorithm does is it looks at the output it is producing the output which is desired understand from what is the difference then and then self-corrects to improve its output as it goes along okay.

So that is all machine learning so all of these concepts of random for s and supervised learning ensemble techniques right all of these techniques are used under this umbrella of machine learning and a lot of organizations use machine learning techniques to make their presence decisions and business decisions can be let’s say forecasting sales or figuring out who to go and do them with who to market or should they enter a particular market or not in all of these business use cases nowadays what organizations use are fairly advanced machine learning techniques they’ve gone on the days when you know there used to be only one ad targeted at everyone in a newspaper around or on a TV channel right now.