Artificial Intelligence for Retail: Skynet level salespeople, God level reviews
While we are watching, Apple and Google are fighting for the creation of AI, and Mask and Hawking are afraid of it, somewhere near artificial intelligence is already literally entering our life from the side where we did not expect. Like me, reading a site about venture business and other start-ups that is widely known in relatively narrow circles, I did not expect to stumble upon an ad for the hackathon of the M-video trading network, which was not the most expected topic from the store selling irons and plasma: on the development of chat bots for retail and the use of artificial intelligence in retail . Being, to some extent, an enthusiast of the topic of AI, I already soaped out to read what application they were going to find in retail, but even on the hackathon site I could not find any detailed description of the idea. It is reported only that ten teams will fight for the prizes, the main one of which, as I understand it, is not even 150 000â‚˝ as such, but receiving a contract from M-Video to finalize their ideas before commercial implementation.
These girls already look like a specially designed shell for AI.
The topic of the hackathon is quite blurred: “Aggregation of reviews on the site” and “Seller’s assistant for the selection of goods in the store” - I did not find the details in the original article or on the site of the hackathon itself . From this it became only more interesting. What will the AI ​​assistant seller do - look for the models he needs in the database, which often takes so much time in the store when you are interested in a particular razor model? Why, for example, does AI need to aggregate reviews on a site when they are already hard cataloged? Just digging a little Google for the relevant requests, I found enough food for thought, which allowed me to accelerate my imagination is not a joke.
This is not an attempt to guess what will actually be developed on the M-Video hackathon - I am not aware of their plans, there is no information on the hackathon website about this. This is a free essay based on. What can be done by armed with AI to solve the tasks.
')
On the topic of "reviews" and "AI", there are already some materials on the Internet, based on the reading of which, I had two possible ideas:
Which store is not interested in the fact that under the product line there are always only positive reviews? Manual moderation of reviews costs too much in man-hours and man-money, and any mechanisms of user self-government are not controlled just in the sense of protecting against honest, but unpleasant feedback from disgruntled consumers. And then artificial intelligence comes to the rescue of the online store, which, analyzing the accumulated feedback base, learns to separate the positive feedback from the negative, no matter in what Aesop language the latter are written. By the way, the development would be useful: imagine Habr, who no longer suffers from experiments on fine-tuning self-regulation mechanisms, but simply hooked up the AI, who knows exactly when the post corresponds to the subject of the resource, even if it is in a corporate blog, and when it’s an advertisement or just chewing on chewed for light plus signs. Which will no longer be when the AI ​​takes over. He will just decide for himself what to show us in accordance with our own requests for what kind of Habr is a cake. And perhaps one of the teams on the hackathon will start writing something that, when adjusted for another task, will become the killer of karma.
An even richer topic, if you think that product reviews are left not only on the websites of various online stores, but also in aggregators like Yandex Market, and in blogs, and anywhere else. And if you recall the experiments of Google with the neural network in the translator, some unprecedented prospects are opening up: once having climbed the entire Internet, the AI ​​could collect all the relevant reviews for a particular product, making the most objective matrix of reviews for it - if it had such a task. Or the most complete matrix of positive reviews, if the task is different.
Further more. When I took the next topic of the hackathon, formulating the creation of a certain AI- "assistant seller for the selection of goods in the store," even my imagination went into a stupor for a while. Again, the description of the hackathon did not provide any information at all. And again I had to look for inspiration, staring into what others have already begun to do in this area. And somewhere among the traditional marketing bullshit about mobile first, big date and, of course, blockchain, I came across the following hint text:
The clue came, which is typical, from one software giant : Tatyana Delyagina, director of marketing data processing technologies and the cloud platform of Microsoft in Russia, voiced it. Who better not to know where to dig in this thread.
If you start from her words, the AI ​​will not be used as an assistant in the search for a directory, as it might seem from the category name. The AI ​​will tell the seller how to sell to a particular buyer. Again, letting go of the fantasy and, just in case, making a reservation that it doesn’t have any relation to the real intentions and the formulation of the M-Video hackathon task, which I simply don’t know,
The only hope to save our money (maybe by that time the unconditional basic income program arrives in our e-wallets — well, who knows?) For an extra second between the full distribution of the ideal sales machines, in this situation it will be their ideal competition — no matter how cool And even if Apple’s ideal AI seller somehow shares my ears with Samsung’s ideal AI seller, Xiaomi’s ideal AI seller will have to somehow balance them all - in the end, I only have my ear, the third smartphone will have to be applied somewhere else already.
In general, without running too far ahead - I am now very interested in what path the fantasies of the hackathon developers will take, what the winning team will have for the project and what they will develop for M-Video further. There are less than two weeks to the results.
These girls already look like a specially designed shell for AI.
The topic of the hackathon is quite blurred: “Aggregation of reviews on the site” and “Seller’s assistant for the selection of goods in the store” - I did not find the details in the original article or on the site of the hackathon itself . From this it became only more interesting. What will the AI ​​assistant seller do - look for the models he needs in the database, which often takes so much time in the store when you are interested in a particular razor model? Why, for example, does AI need to aggregate reviews on a site when they are already hard cataloged? Just digging a little Google for the relevant requests, I found enough food for thought, which allowed me to accelerate my imagination is not a joke.
This is not an attempt to guess what will actually be developed on the M-Video hackathon - I am not aware of their plans, there is no information on the hackathon website about this. This is a free essay based on. What can be done by armed with AI to solve the tasks.
')
Aggregation of site reviews
On the topic of "reviews" and "AI", there are already some materials on the Internet, based on the reading of which, I had two possible ideas:
Analysis of reviews divided into positive and negative in automatic mode
Which store is not interested in the fact that under the product line there are always only positive reviews? Manual moderation of reviews costs too much in man-hours and man-money, and any mechanisms of user self-government are not controlled just in the sense of protecting against honest, but unpleasant feedback from disgruntled consumers. And then artificial intelligence comes to the rescue of the online store, which, analyzing the accumulated feedback base, learns to separate the positive feedback from the negative, no matter in what Aesop language the latter are written. By the way, the development would be useful: imagine Habr, who no longer suffers from experiments on fine-tuning self-regulation mechanisms, but simply hooked up the AI, who knows exactly when the post corresponds to the subject of the resource, even if it is in a corporate blog, and when it’s an advertisement or just chewing on chewed for light plus signs. Which will no longer be when the AI ​​takes over. He will just decide for himself what to show us in accordance with our own requests for what kind of Habr is a cake. And perhaps one of the teams on the hackathon will start writing something that, when adjusted for another task, will become the killer of karma.
Aggregation of reviews by external sources
An even richer topic, if you think that product reviews are left not only on the websites of various online stores, but also in aggregators like Yandex Market, and in blogs, and anywhere else. And if you recall the experiments of Google with the neural network in the translator, some unprecedented prospects are opening up: once having climbed the entire Internet, the AI ​​could collect all the relevant reviews for a particular product, making the most objective matrix of reviews for it - if it had such a task. Or the most complete matrix of positive reviews, if the task is different.
AI assistant - the seller, whose offers to anyone can not be refused
Further more. When I took the next topic of the hackathon, formulating the creation of a certain AI- "assistant seller for the selection of goods in the store," even my imagination went into a stupor for a while. Again, the description of the hackathon did not provide any information at all. And again I had to look for inspiration, staring into what others have already begun to do in this area. And somewhere among the traditional marketing bullshit about mobile first, big date and, of course, blockchain, I came across the following hint text:
Cloud services of artificial intelligence and machine learning make it possible to communicate in a new way with customers: to make interaction with them more targeted and personalized. Fresh ideas of the participants of the competition, implemented on the basis of the intelligent cloud platform Microsoft Azure, will help M.Video on the path to digital transformation.
The clue came, which is typical, from one software giant : Tatyana Delyagina, director of marketing data processing technologies and the cloud platform of Microsoft in Russia, voiced it. Who better not to know where to dig in this thread.
If you start from her words, the AI ​​will not be used as an assistant in the search for a directory, as it might seem from the category name. The AI ​​will tell the seller how to sell to a particular buyer. Again, letting go of the fantasy and, just in case, making a reservation that it doesn’t have any relation to the real intentions and the formulation of the M-Video hackathon task, which I simply don’t know,
- more or less understandable, like teaching AI all the techniques of a qualified sales person - from assessing the viability, interests, suggestibility of a person with subsequent methods of “pairing” that no sales person will ever master 100%, but the AI ​​does not interfere with wielding a full arsenal of all human notions strictly by the book, giving a sales result close to one hundred percent;
- and very much in the style of "Black Mirror", when AI will analyze the background of the buyer from the transport on which he arrived and got into surveillance cameras, to the history of publications in social networks, history of purchases, billing and anything. In other words, an AI assistant seller can be turned into an ideal sales machine, which neither the spontaneous nor the most conscious buyers will be able to oppose with anything.
The only hope to save our money (maybe by that time the unconditional basic income program arrives in our e-wallets — well, who knows?) For an extra second between the full distribution of the ideal sales machines, in this situation it will be their ideal competition — no matter how cool And even if Apple’s ideal AI seller somehow shares my ears with Samsung’s ideal AI seller, Xiaomi’s ideal AI seller will have to somehow balance them all - in the end, I only have my ear, the third smartphone will have to be applied somewhere else already.
In general, without running too far ahead - I am now very interested in what path the fantasies of the hackathon developers will take, what the winning team will have for the project and what they will develop for M-Video further. There are less than two weeks to the results.
Source: https://habr.com/ru/post/337516/
All Articles