Can Amazon AI voice replace customer reviews? It’s starting to try

Anyone who is looking for a product review in Amazon knows how valuable the experience of other shopping to be the experience of the other shopping and how easy it is to fall from the rabbit hole to the customer’s comments from five-star praise to lifting from one-star broadcast. As Amazon continues to offer AI features, it will facilitate shopping. Audio descriptions created by AI The products are in a mixture and this customer interpretation is scratching-perhaps settled in a position that will change it as a source of purchasing information.

The AI-VOED product descriptions, called “Hear the Emphasis”, rely on a large language model in order to transform customer reviews and convert the content into short-form sound clips by pulling Amazon’s product catalog from Amazon’s product catalog, and then converting content into short-form audio clips.. Summaries began to be released in a subset of US customers in selected products in the summer months and now reached As a button in the mobile shopping application, all US customers scares more than one million products to cover.

A unique attractiveness of Amazon and other e-commerce options has been the ability to get information from real users, not only product descriptions. Of course, although false investigations have been a problem for Amazon for a long time Usage of prohibited wage promotions and other forms of inspection writing. Nevertheless, you need to list various efforts to play the system, and in recent years, the latest review writers show that chatbots like chatgpt. However, the real people who share their unique experience with a product – customers as a source of information and informed decision -making – from clothing and shoe sizes to security and narrower questions has become an important part of the learning process: when buying a new bread frying machine, the simit setting is correct?

Can AI improve this? Investigations are eager to nature in mass individuality, but human readers are very resourceful to distillation that they need human chaos. In a sense, the average Amazon Review reader stripes the brain, keywords and key information and is a very good large language model that chooses, so an AI must at least have to do so and must add customer experience as promised.

AI and Cognitive overload

AI has advantages. The first will not experience cognitive overload With more than 32,000 comments, an electric tea in Amazon is faced with a kettle. Can you scan the data, but can he only give us what we need or what we want to know? This can still be difficult for AI to be correct, and it can be naturally problematic to confuse product catalogs, customer reviews and web information into a single distillation – the sources of this information comes with its own intention.

“It is important to recognize that AI is still strong as it is still inadequate, such as automation and pattern recognition and is still inadequate as in judicial tasks.” He said. “A key question is whether it is a way to take into account as a input in the context of the customer when producing this summary.” He said.

According to EDKI, the value of AI is to find the appropriate problem of adequacy compliance. If this is not achieved, the sense of cheating will probably enter through a gate that is probably left open. AI fatigueHe says that consumers probably lived so far.

Amazon, AI-Aumaio is currently the same for each user.

“The AI ​​summary must capture the nuance and context.

The ability to take into account the account and the examination process into more dialogue between the consumer and the machine is the place where technology is directed – in other words, in other words, the AI’s actively added to the AI’s intermediary means to the AI’s intermediary means to identify your own words and then encouraged to select the products. It is separate from the bar.

Chat and sound summarys will remain between ways of interaction, but it is what will shift the experience from one-way delivery to two-way discovery from one-way delivery, according to Edkie, to ask for real-time interviews with an AI audio agent-to ask for certain questions, to explain concerns and to get more deeper information from investigations. “Currently, you can interact with Rufus through text or sound input, but Rufus cannot speak again, their answers are only text, “he said. You can make a two -way speech with a boat talking to you with sound agents, “he added.

For now, a customer segment that is likely to see urgent benefits Visually Impaired ShoppingAccessibility makes it an interesting aspect of the feature, but the sound should be high quality and present the content correctly.

Brian Numinville, the director of the consumer research firm feedback group, says that such features have the potential to make exchange more accessible and convert the detailed text into sound summarys that are easily consumed. However, the feature should be designed in a thoughtful way to really benefit people with visual impairment. According to Numinville, this screen involves providing full compatibility with the readers and keyboard navigation, providing open, structured and concise summary, and avoiding extremely long or confusing sound presentations. The quality and clarity of the sound produced by AI will significantly affect the usability.

“The transition from different human investigations to summaries produced by AI may mean losing important nuances, context and personal touches.” He said.

The risk of losing unique shopping insights

The tendency of artificial intelligence to focus on common themes can dilute their answers even when distilled them.

Human tends to include investigations Pile stories And the details of the extremely specific usage situations-imagine what motivates another to write a review in the first place, and how the shoppers who can help make the shopping experience more personal and understanding.

“AI may ignore unique information or niche needs that are not compatible with the majority of responses.” He said. He continued: “In addition, investigations – such as identifying prejudices or relying on certain reviews – critical interpretation ability is decreasing with AI summaries.”

He sacrificed the additional explanation that can distinguish between product descriptions, customer reviews and web information, scratching a large content to create summaries.

“It is not 100% clear how much of the product descriptions in the current application are 100% clear – Amazon did not fully detail this mixture.” He said.

Amazon refused to comment and directed CNBC to public information about the feature.

“Product descriptions are typically marketing guided and emphasizes positive, while investigations reflect real user experiences and include both pros and cons.” He said. Without referring to confusion, it may be difficult for the cultivates to distinguish Ad-Speak from real investigations, the results may be like local advertising.

“This blending, if it occurs, can mislead consumers by lending the factual authority by hiding subjective views or promotional materials as impartial investigations.” He said.

Research, customers Confidence sound As a form of information distribution, regardless of how balanced the information given is.

“In my opinion, Amazon’s desire to sell products seems to aggravate all critical perspectives in a product,” Australia Curtin University Professor of Internet Studies. ” He said.

Another concern, “AI can aggravate the average and overall scores, even if most of the receivers, even if not a few negative examination – one star – is looking,” he said.

He is a research scientist at Zoom Communications. In his personal opinion, Nauman Dawalatabad said that technology is progressing in the direction of better customer experience. “I take it as a technology that helps us to make knowledgeable decisions,” the time spent was wasted when it may come from reading through cognitive fatigue and customer investigations. He said.

If a description of the ahli leads to impulses and regrets, that is, on the buyer, he says, and is not always different from the way things work – the same fine coercion against all marketing efforts can be made. As the sound -based agent continues to develop and starts to talk to a AI agent (instead of writing and searching), and what they want, and what they want, “He will get exactly what you need.”