Jeff Berman is one of a rare breed of AV industry writers who focuses on the business side of the market. In addition to a rich history of working in retail, he has written for M&E Daily, Smart Content News, Smart Screen News, and CDSA Cyber Security News, and also worked for six years as a contributing editor for the Consumer Technology Association's annual Digital America publication.
By now, you probably know that artificial intelligence (AI) is being employed to one degree or another in digital voice assistants like Amazon Alexa, Google Assistant, and others. You may even know that machine learning is being used by online streaming services like Netflix to deliver the best-quality stream your network can support and try to better figure out what TV show or movie you might want to watch next after wasting eight hours of your life binging a full season of Diva Brides. But did you know that at least some devices that are part of your home theater system, including your new AV receiver or Ultra HD TV, may be using AI as well?
You can only expect a growing number of manufacturers and service providers to start using AI in an increasing number of their products and services in the months and years to come, whether you want it not. And you can also expect those manufacturers and service providers to keep finding new applications for AI.
So far, however, many consumers have already raised two big concerns with AI. One: How much of my personal information is being collected by the AI used in these devices and services? Two: When will the AI used in these devices take control of my house, my city, my country, and/or the entire world as Skynet did in the Terminator flicks?
Perhaps an even more fundamental question is this: Why the hell do we even need AI to begin with?
To answer the first question, at least some--but not all--of those devices and probably all of those services using AI are indeed collecting at least some of your personal information. After all, these devices and services would not be able to offer you the personalization you're receiving without knowing at least some information about you, especially your viewing preferences. In other words, just because you're paranoid doesn't mean corporations aren't keeping close tabs on all aspects of your home entertainment listening and viewing habits. However, anybody worried about the AI gathering more significant information about you, including your address and credit card information, should consider that the makers of these devices and providers of these services likely already have your payment information and other personal details about you, including where you live.
Based on comments made by analysts we interviewed for this story, you probably don't have to worry about the AI in your devices and services taking complete control of your house and the world--at least not anytime soon anyway. As for why--or whether--AV needs AI at all? Hopefully we'll get to the bottom of that, as well.
Devices Using AI
In one of the more prominent consumer-electronics AI announcements of the past two years, Samsung Electronics introduced an 85-inch UHD at CES in 2018 that it said can upscale any content to 8K automatically using AI. Samsung didn't respond to requests for comment to discuss the TV's AI or potential consumer concerns.
In promotional online videos, Samsung has touted how its AI based on machine learning--a subset of AI--analyzes the TV content and automatically upscales low-resolution images into 8K picture quality. You can get an idea of what it looks like in the videos below.
Even though much of the talk around AI has been that it's "this very futuristic kind of thing that's doing all these crazy, wild things," said Bob O'Donnell, president and chief analyst at TECHnalysis Research, "some of what AI does is actually very subtle," such as the way Samsung is using the technology in the TV. "There have always been scalers," he said, adding: "There's been upscaling on TVs since the early days of SD--standard-def to high-def and then high-def to 4K.... It's just a question of how you do the scaling and how good it is on lots of different content, because inevitably what [we've] seen in the past is people have done scaling that works really well on some content but not so well on other content. It might work great for slow-moving movies, but it's terrible for fast-moving sports, for example." If AI can make upscaling "consistently better" and if it's "smart enough to know" that the device it's controlling must switch the type of scaling being done based on the content, "I don't think anybody would complain about that--that's clearly a great thing," he said, adding that goes for how Yamaha's using AI also.
Yamaha introduced AI in its audio/video equipment in spring 2018 under the name Surround:AI, Phil Shea, content marketing manager for the AV division of Yamaha Corporation of America, pointed out. Currently, the AI is found in its top three RX-A 80 Series AVENTAGE receivers, as well as the CX-A5200 preamp. "The AI technology analyzes a scene's soundtrack in real time to calibrate the perfect sound field for the listener, making the speakers disappear into the room," Shea explained.
According to Shea, consumers have no reason to be concerned about any privacy issues with the AI being used in the Yamaha devices. "Being that Surround:AI is a closed system and not cloud-based, there is no data that is captured and sent to a server," he said.
So, is Surround:AI necessary for the average consumer? Probably not. But so far, consumers who own the Yamaha devices featuring the technology "love the convenience that Surround:AI provides; instead of manually having to find the best surround mode on their remote control, Surround:AI finds it for them," Shea said, noting that's "especially helpful when an action scene quickly switches to a quiet, intimate setting."
Yamaha went on to integrate its MusicCast-enabled AV receivers into Josh.ai, a voice-controlled home automation system designed for residential custom integrators. Announcing it, Yamaha said custom integrators can add MusicCast-enabled receivers as a "controlled component of a home entertainment system using the Josh.ai browser-based setup interface," explaining that the platform "auto-discovers all AV devices on a network and natively understands multiple sources and destinations." Via Josh.ai's proprietary natural language processing (NLP), "users can conversationally request their home to execute any number of tasks," Yamaha noted. As an example, it said, the room awareness feature of the Josh.ai microphones allow users to give commands like, 'OK Josh, dim the lights, listen to 'Paint it Black' by the Rolling Stones, and watch 'Black Mirror' season two, episode three.'"
You can expect a growing number of manufacturers to start adding similar functions to their devices. Although Yamaha isn't currently using AI in any additional home theater devices, Shea pointed out his company has also been integrating AI into its Disklavier player pianos.
Click over to page two to for a discussion about the economics of AI, as well as reasons to be cautiously optimistic...
What's the Cost?
Incorporating AI into the Yamaha products hasn't exactly been breaking the bank for the company, apparently. Although Shea provided no specifics on how much it's costing his company to develop this sort of artificial intelligence: "The processing that is required to make Surround:AI possible is already available in previous, higher-end AVENTAGE models. In fact, prices for the models that incorporate artificial intelligence are no more expensive than older models in the same class."
Like Yamaha, Netflix declined to provide any specifics on how much cost was involved with AI. It seems obvious, however, that it's a lot easier for a large company like either of those two (or Samsung) to afford either the creation of its own AI algorithms--which typically involves, at the very least, AI specialists to be employed by the company--or to pay another company for AI. Yamaha noted that Surround:AI was created internally and not based on another company's AI.
Although AI project costs vary based on the specific work being done, custom software developer Azati Software says on its blog that typical prototype development costs are about $25,000 and it then costs a minimum of $35,000 to $100,000 to develop a viable product with functional features based on the prototype. It adds: "While a few years ago only the likes of Google, Microsoft and Facebook could afford to develop ML-powered software, now a large number of companies can do this as well [due to] the emergence of various tools, libraries, and frameworks for building ML-based software."
Guarded Optimism About AI
Based on the results of an online AI Consumer Sentiments study that the Consumer Technology Association (CTA) conducted in fall 2018 of about 2,000 adults of all age ranges, consumers had mixed feelings about the technology, according to CTA senior analyst Sayon Deb. The CTA found that three in four of the people surveyed indicated they had at least seen or heard "a little" about AI developments and there was a "high degree of awareness" among the respondents about whether or not products used AI.
When "we dug a little bit deeper" to get the impressions of the respondents, however, "the first thing that came up in their minds is robots," he noted, adding: "When you say AI, people think of robots, so the biggest misconceptions" about AI have been the technology's association with science-fiction movies including The Terminator and the apocalyptic scenarios that are presented in them. About 71 percent of those surveyed said AI could create unintended problems and about the same number said it could be used for malicious reasons, he said.
Despite such fears, overall, consumers surveyed seemed to have "guarded optimism" about AI and seemed to "understand the value-add from these sort of intelligent services can improve and enhance their lives, as well as their productivity," he said, adding that many consumers seemed to be aware that AI could potentially lead to new discoveries and free up their time to do more creative work and can also create new types of jobs. At the same time, he said, many consumers were still waiting to see if devices using AI would improve, so they were waiting to buy such devices.
Although the impression of killer robots or fears over losing jobs from AI implementation were "definitely overblown," he warned it's "critical for companies that are currently developing AI solutions to really pay attention to consumer concerns." What we have now are still "very narrow uses of AI" and more complex AI likely won't arrive until "far into the future," he predicted. But consumers "have the right to sort of worry about privacy and that is something that continues to be an issue that is still evolving, and ultimately consumers will have to choose where on the spectrum of the privacy do they feel the most comfortable," he said, noting CTA recommends that its members companies and companies designing AI solutions "really be transparent about AI applications in consumer devices and services."
O'Donnell at TECHnalysis Research, meanwhile, classified AI as a "mixed bag" so far. He praised its use in the Samsung and Yamaha devices and said it's good that AI is being used to help consumers find more of the content they like and might not know about. But, he said, "the problem is, in order to do that, it wants to basically kind of see everything else you've already been watching, for example, in the case of content, and then compare that against everybody else." Similar to CTA's Deb, he predicted that "what you're going to see eventually start to happen is you're going to be able to get some of the personalization without having to share all of your data to some cloud-based service because, over time, what's going to develop, I believe, is recommendation engines that can run locally on your smart TV or on your other connected device and, from that, be able to ... give you the benefits without some of the current downsides around privacy and security."
Much of this trepidation is fueled by the fact that certain early smart TV apps were monitoring what consumers were watching and not letting them know that they were doing that, he said. Some consumers heard about that and "turned some of these capabilities off because obviously most of them have the capability to be turned off--but it's frustrating for people to have them turned on by default," he noted, adding: "You should have to opt in to something like that as opposed to opt out."
How many of you have devices that use AI and what concerns, if any, do you have about the technology? Let us know in the comments section below.