In an interview with Gadgets 360, Rajeev Rastogi, Vice President, Machine Learning, Amazon India, said, “Machine learning is ubiquitous on Amazon today.
One of the most basic examples of how Amazon uses machine learning (ML) is to mistype a query in its search bar. The e-commerce site, Rastogi points out, is that instead of looking at the text distance between typed spelling queries and accurate queries instead of providing accurate results it’s okay if you misspell something.
For example, if you type “geyser” on Amazon to search for available geyser options, the market will automatically correct the spelling and show you the corresponding results. Amazon uses ML models to translate the content on its site into now supported Indian languages.
Of course, these types of applications are now commonplace, and most of us do not think about the terms artificial intelligence (AI) or machine learning.
Rastogi said his team is currently working on a seed initiative aimed at bringing in a conversational shopping experience. Aimed at first-time online shoppers who are very familiar with interacting with offline shoppers by ordering through the e-commerce site.
Conversation trading through chatpots with smart assistants like Amazon’s own Alexa is one of the ideas that comes back once in a few years as technology improves, and Rastogi talks about how it starts with text in English, but develops into other languages. , And give voice.
“A machine can read a document and answer any question about the document. It is difficult today.
AI is used to analyze text and speech at different levels. But computer engineers and data scientists have not yet found the right combination to use AI and machine learning to make accurate estimates, such as movie or product reviews. In a research paper by Gerit Wagner, Roman Lukyanenko and Guy Barre, researchers in the field of information technology at HEC Montreal, on how AI can be used in the literary review process, information from sources that sometimes use vague, confusing language and presentation Will have difficulty evaluating.
McKinsey Global Institute (MGI) partners Michael Chui, James Manyika and Mehdi Miremadi pointed out in an article that AI models “have difficulty transporting their experiences from one situation to another” and should train models even when used by companies. Very similar. This adds additional resource requirements.
Shreyas Sehgar, Assistant Professor of Operations Management at the Department of Management at the University of Toronto Scarborough and the Roadman School of Management, said it was uncertain whether the AI-based bot would communicate with humans and deliver relevant results, especially in markets including India. . Seker has done extensive research on how e-commerce sites use machine learning in consumer fronts and warehouses to enhance their functionality.
“When you hear these chats, no, is it going to rain tomorrow? Or can I play the song for this movie? They do a great job. But as you start to get more and more complicated questions, hey, can you help me find a good shoe for my trek? I think it’s very difficult to even clearly distinguish your purpose from this question on Chatbot or Alexa? What do you do as a person and how do you differ from others? What products would suit you? ”He said.
Dealing with bias and errors
One of the biggest challenges in using AI and ML nowadays is controlling bias and errors. Companies ranging from Google and Facebook to Microsoft continue to deal with these mistakes. Amazon is not fooling around on that front.
Seker and the Rodman School of Management at the University of Toronto Scarborough noted that Amazon’s AI rankings already include many dependencies that the company is already aware of and that are clearly involved in resolving them, but it is not clear how successfully it has achieved the desired results.
“For example, historically, users may have clicked on a particular brand of earphones, and then what happens is that in the future, I’ll replicate that exact brand again and again.
However, Rastogi strongly disagreed, saying that Amazon’s goal was not to change them completely, but to help human workers.
Who does this help?
The use of AI and ML helps Amazon deliver what you need by understanding your buying behavior and purchase history. However, this can sometimes lead to buying impulses and convincing you to buy something you don’t really need. Experts hope this will grow further with more conversational shopping experience.
“I think AI and ML can definitely boost the idea of turning window shoppers into regular shoppers,” Seker said. “I think this is definitely a great way to think of Amazon as a very credible seller.”
Consumers can deal with this behavior on their own by understanding how ethics can affect their choices.
“We are the ones who are guided in the shopping funnel by the algorithm in different places, whether it is the last one to go and click to buy an item, whether it is a recommendation or a review,” Seker said.
Ankur Bisen, senior consultant and chairman of the consumer, food and retail divisions of management consulting firm Technopak, said Amazon uses its methods such as advertising, marketing and discounts to help consumers buy more. Did retail shop.
“Amazon does it very precisely because it is limited,” he said. “Conversation AI is not just close to Amazon’s monopoly domain. Yes, they are very good at it because of Alexa. But you will find that dialog AI emerges in different forms offered by other technology sites.