Amazon’s AI-powered Alexa is as big a privacy red flag as the old Alexa – Ars Technica

Zoom in / Alexa uses generative AI to create a taco poem.

Amazon is trying to make Alexa simpler and more intuitive for users by using a new large language model (LLM). During its annual hardware event on Wednesday, Amazon showed off a demo of AI-powered Alexa that users can soon preview on Echo devices. But in all its talk of new features and an AI-fueled future, Amazon has barely acknowledged the long-standing elephant in the room: privacy.

Amazon’s hardware event showcased the new Echo Show 8, updated Ring devices, and new Fire TV sticks. But what’s even more interesting is a look at how the company is trying to cut through the generative AI hype and uncertainty about the future of voice assistants. Amazon said users will be able to start previewing the new Alexa features via any Echo device, including the original, in A few weeks.

The added Alexa features are enabled by a new LLM that Amazon says is fine-tuned for voice conversations and uses algorithms to recognize body language and intonation. The company has been clear that Alexa will focus on generative AI in the future. But the new features are still in their early stages, Amazon noted, so bumps and bugs should be expected at first.

“Alexa, can I finally stop saying Alexa?”

One development that has had an immediately noticeable impact is Alexa learning to listen without the user needing to say “Alexa” first. The device will be able to use its camera, a visual ID previously created for users, and previous setup with Alexa to determine when someone is speaking to it.

In the demo, David Lemp, senior vice president of devices and services at Amazon, was able to have a conversation with Alexa, step away for a moment, then return to the device for further conversation without saying “Alexa” again.

“This is made possible through on-device visual processing and audio models that work in concert to try to determine whether you are addressing the device or another person in the room,” Rohit Prasad, Amazon’s senior vice president and chief AI scientist, explained on stage. At an Amazon event.

Amazon is also working to reduce Alexa’s response time and allow users to pause or use fillers like “um” without interrupting the interaction. Alexa can do this by using a massive matching model with billions of parameters, Prasad said.

All of this suggests that Alexa is listening and watching with more intent than ever before. But Amazon’s presentation didn’t detail any new privacy or security capabilities to ensure this new power isn’t used maliciously or in a way users don’t agree with.

When reached for comment, an Amazon spokesperson noted what had previously been determined Alexa privacy featuresincluding an indicator light on Echo devices that indicates that Alexa is listening and allows users to review and manage their voice history.

See also  Microsoft plans to secure Windows DNS like never before. Here's how.

The spokesperson added that Amazon uses “many tactics and features” to secure devices and customer data, including “rigorous security reviews during development, encryption to protect data, and regular software security updates,” and that Amazon “will continue to take steps to harden” device security.

Amazon says the Echo Hub for smart home control will be released later this year for $180.
Zoom in / Amazon says the Echo Hub for smart home control will be released later this year for $180.

In an effort to reduce the tedium of controlling multiple smart home devices, Amazon is also updating Alexa to simplify smart home commands. For example, Alexa could eventually know how to “turn on the new light in the living room,” even if you don’t know the official “name” of that smart bulb. Amazon also discussed the ability to control multiple devices and set up routines with a single sentence instead of having to say “Alexa” and a device name for each instruction. Other developments include making Alexa infer what you want, such as speeding up the robovac when you tell it the house is dirty. Such features, if reliable, would make voice control of smart devices significantly more attractive and intuitive. The features will be previewable “in the coming months” from Amazon He said.

Artificial intelligence and fire tv

Amazon announced the Fire TV Stick 4K Max (2nd generation) with Wi-Fi 6E on Wednesday for $60.
Zoom in / Amazon announced the Fire TV Stick 4K Max (2nd generation) with Wi-Fi 6E on Wednesday for $60.

Amazon also showcased ways Fire TV devices with generative AI technology can help users find digital content. It was previously reported that Amazon was exploring some of these features, such as having Alexa make movie and TV show recommendations based on prompts like “Find the movie with that guy who played the lawyer in it.” Too badOr search through options using filters like “Show me action movies,” followed by “Show me movies I don’t have to pay for,” and “Show me movies I haven’t seen yet.” To provide these recommendations, the Fire TV devices pull from various sources of movie and TV content, including IMDB, and from users’ Fire TV profile and viewing habits.

Amazon said new Fire TV AI features, including the ability to upload family photos and add filters to them, will arrive via an over-the-air update later this year.

These planned features make sense in and of themselves. Finding content amid today’s streaming wars is a hassle, and I’ve previously pointed to Fire TV devices as a potential lifeline for Alexa’s revenue. But I’m resistant to handing over more of my viewing habits to Amazon. Amazon has always claimed not to sell customer data, but this data can still be used to make business decisions and disclosed Non-user-specific trends With third parties, and for targeted marketing.

See also  Nintendo Switch system update 18.0.0 is now available, and here are the full patch notes

Alexa chatbot

Alexa is also evolving into a chatbot that uses voice input, camera, and presence detection sensors to chat with users and provide information quickly. There’s a lot of focus on Alexa having personality through a more expressive voice and the chatbot having opinions like a favorite sports team. He must also be able to carry on back-and-forth conversations and create content, such as poems, with specific or even variable gestures.

“[Alexa has] “We’ve always been a little more transactional than we’d like, but that was a limit of the technology, not the vision,” Limp said on stage.

Amazon announced the Echo Show 8 yesterday.  Starting at $150, it will ship in October.

Amazon announced the Echo Show 8 yesterday. Starting at $150, it will ship in October.

Amazon

The Limp demo appears to show Alexa a draft of a note inviting people to visit and including details (such as that there will be barbecue chicken) from her previous conversation with Limp. He then appears to have sent the note to Limp’s phone, likely to the Alexa app.

Alexa’s chatbot features will be available as an “early preview” in the coming weeks. It is triggered when the user says, “Alexa, let’s talk” and ends when they say “exit.”

New Alexa, same privacy concerns

Over the past year, there have been questions about the future of voice assistants like Alexa and Google Assistant, as they have failed to become reliable revenue streams.

Historically, many of Amazon’s smart devices, such as Echos, have been sold at cost, with hopes of making money through Alexa-based transactions and through access to digital content (however, a recent report suggested that Amazon is turning to the prospect of monetizing the devices which are cheaper to produce).

But as Amazon seeks to put Alexa in more rooms in people’s homes (I can’t count the number of times Amazon has hinted that Alexa is like a family member), the company has done little to improve confidence that its devices will be safe and non-invasive. .

As Limp said during Wednesday’s presentation, Amazon wants Alexa to be able to remember recently played music, enable personalized reminders, and recommend recipes based on grocery purchases. To do this, you’ll need to leverage “the preferences you share, the services you’ve interacted with, and information about the environment in and around your home.” Providing all this information to Alexa may get easier with this AI product update, however trust Alexa hasn’t necessarily gotten easier.

The use of visual IDs to enable use of Alexa without a wake word increases reliance on cameras and microphones, yet Amazon has not unveiled any renewed approaches to customer privacy. The company has previously been caught keeping recordings, including childrenForever, Amazon workers were caught listening to Alexa’s voice and spying on Ring users. Alexa’s voice has been used in criminal trials. Amazon says it doesn’t send photos or videos to the cloud and stresses the Echo Show’s microphone/camera off button and physical camera shutter are built-in.

See also  Apple Silicon is an inconvenient truth

Wednesday’s event highlighted ways kids can have interactive conversations with Alexa by playing trivia games and learning facts. this Explore with Alexa The feature should be released before the holiday season. But Amazon did not mention any new and improved methods that will ensure that all content shared with children is safe and reliable.

The Echo Pop Kids smart speaker will ship next month for $50.
Zoom in / The Echo Pop Kids smart speaker will ship next month for $50.

Alexa has failed to protect children in the past, including in 2021, when she advised a 10-year-old to touch a penny with a partially exposed plug. Naturally, Amazon has addressed this specific error and will likely try to avoid similar situations in the future. But the company hasn’t done enough to convince me that there won’t be a similar event, especially with generative AI.

When I contacted an Amazon spokesperson for comment, he told me that since the penny incident, the company has improved its systems for preemptively detecting and blocking “offensive or potentially harmful content.”

However, Alexa’s generative AI update builds on “existing internal content policies and guardrails,” according to an Amazon representative. We know that these policies and guardrails have failed users in the past, even if Amazon says it’s working to improve. When it comes to children’s safety, there is less room for error.

An Amazon rep added:

We’ve also based LLM on a cognitive foundation so that if a model lacks the information needed to respond, it will automatically consult licensed data sources and the Knowledge Graph that helps Alexa accurately answer today’s customer questions. LLM is able to gather the most accurate and up-to-date information before responding to reduce inaccurate responses.

Limp described trust as a key pillar of modernizing generative AI, saying:

You have to trust this. It’s in your house. We believe there is absolutely no trade-off between trust and performance…which is why we will continue to provide customers with information and controls so they can take control of their experience, even with generative AI powering it in the background.

Privacy and security aren’t the flashiest things at live presentations, but since Amazon’s event ended, the company hasn’t provided any follow-up information about any new ways Amazon is working to ensure it properly handles customer data and privacy.

Amazon wants Alexa to be easier to work with, but also smarter about where you are, how you live, what you like, what you’ve done and watched, and what your kids are interested in. Some Alexa developments could improve its functionality and utility, but their appeal remains limited until Amazon can prove that it prioritizes security and privacy as much as it makes Alexa more competitive.

Leave a Reply

Your email address will not be published. Required fields are marked *