{"id":365784,"date":"2025-12-23T16:08:10","date_gmt":"2025-12-23T16:08:10","guid":{"rendered":"https:\/\/www.newsbeep.com\/us\/365784\/"},"modified":"2025-12-23T16:08:10","modified_gmt":"2025-12-23T16:08:10","slug":"how-ai-broke-the-smart-home-in-2025","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/us\/365784\/","title":{"rendered":"How AI broke the smart home in 2025\u00a0"},"content":{"rendered":"

This morning, I asked my Alexa-enabled Bosch coffee machin<\/a>e to make me a coffee. Instead of running my routine<\/a>, it told me it couldn\u2019t do that. Ever since I upgraded to Alexa Plus,<\/a> Amazon\u2019s generative-AI-powered voice assistant, it has failed to reliably run my coffee routine, coming up with a different excuse almost every time I ask.<\/p>\n

It\u2019s 2025, and AI still can\u2019t reliably control my smart home. I\u2019m beginning to wonder if it ever will.<\/p>\n

The potential for generative AI<\/a> and large language models to take the complexity out of the smart home, making it easier to set up, use, and manage connected devices, is compelling. So is the promise of a \u201cnew intelligence layer<\/a>\u201d that could unlock a proactive, ambient home.<\/p>\n

But this year has shown me that we are a long way from any of that. Instead, our reliable but limited voice assistants have been replaced with \u201csmarter\u201d versions that, while better conversationalists, can\u2019t consistently do basic tasks like operating appliances and turning on the lights. I want to know why.<\/p>\n

\"I\u2019m<\/a><\/p>\n

I\u2019m still waiting on the promise of voice assistants that can seamlessly control my smart home. Photo by Jennifer Pattison Tuohy \/ The Verge<\/p>\n

This wasn\u2019t the future we were promised.<\/p>\n

It was back in 2023, during an interview with Dave Limp<\/a>, that I first became intrigued by the possibilities of generative AI and large language models for improving the smart home experience. Limp, then the head of Amazon\u2019s Devices & Services division that oversees Alexa, was describing the capabilities of the new Alexa <\/a>they were soon to launch (spoiler alert: it wasn\u2019t soon<\/a>).<\/p>\n

Along with a more conversational assistant that could actually understand what you said no matter how you said it, what stood out to me was the promise that this new Alexa could use its knowledge of the devices in your smart home, combined with the hundreds of APIs they plugged into it, to give the assistant the context it needed to make your smart home easier to use.<\/p>\n

From setting up devices to controlling them, unlocking all their features, and managing how they can interact with other devices, a smarter smart home assistant seemed to hold the potential to not only make it easier for enthusiasts to manage their gadgets but also make it easier for everyone to enjoy the benefits of the smart home.<\/p>\n

Fast-forward three years, and the most useful smart home AI upgrade we have is AI-powered descriptions for security camera notifications<\/a>. It\u2019s handy, but it\u2019s hardly the sea change I had hoped for.<\/p>\n

It\u2019s not that these new smart home assistants are a complete failure. There\u2019s a lot I like about Alexa Plus; I even named it as my smart home software pick of the year<\/a>. It is more conversational, understands natural language<\/a>, and can answer many more random questions than the old Alexa.<\/p>\n

While it sometimes struggles with basic commands, it can understand complex ones; saying \u201cI want it dimmer in here and warmer\u201d will adjust the lights and crank up the thermostat. It\u2019s better at managing my calendar, helping me cook, and other home-focused features. Setting up routines with voice is a huge improvement over wrestling with the Alexa app \u2014 even if running them isn\u2019t as reliable.<\/p>\n

\"Google\u2019s<\/a><\/p>\n

Google\u2019s new Gemini for Home AI-powered smart home assistant won\u2019t fully launch until next spring, when its new smart speaker arrives. Photo by Jennifer Pattison Tuohy \/ The Verge<\/p>\n

Google has promised similar capabilities with its Gemini for Home upgrade<\/a> to its smart speakers, although that\u2019s rolling out at a glacial pace<\/a>, and I haven\u2019t been able to try it beyond some on-the-rails demos<\/a>. I was able to test Gemini for Home\u2019s feature that attempts to summarize what\u2019s happened at my home using AI-generated text descriptions<\/a> from Nest camera footage. It was wildly inaccurate<\/a>. As for Apple\u2019s Siri, it\u2019s still firmly stuck in the last decade of voice assistants, and it appears it will stay there for a while longer<\/a>.<\/p>\n

The problem is that the new assistants aren\u2019t as consistent at controlling smart home devices as the old ones. While they were often frustrating to use, the old Alexa and Google Assistant (and the current Siri) would generally always turn on the lights when you asked them to, provided you used precise nomenclature.<\/p>\n

Today, their \u201cupgraded\u201d counterparts struggle with consistency in basic functions like turning on the lights<\/a>, setting timers, reporting on the weather, playing music<\/a>, and running the routines and automations<\/a> on which many of us have built our smart homes.<\/p>\n

I\u2019ve noticed this in my testing, and online forums<\/a> are full of users who have encountered it. Amazon<\/a> and Google<\/a> have acknowledged the struggles they\u2019ve had in making their revamped generative AI-powered assistants reliably perform basic tasks<\/a>. And it\u2019s not limited to smart home assistants; ChatGPT can\u2019t consistently tell time or count.<\/a><\/p>\n

Why is this, and will it ever get better? To understand the problem, I spoke with two professors in the field of human-centric artificial intelligence with experience with agentic AI and smart home systems. My takeaway from those conversations is that, while it\u2019s possible to make these new voice assistants do almost exactly what the old ones did, it will take a lot of work, and that\u2019s possibly work most companies just aren\u2019t interested in doing.<\/p>\n

Basically, we\u2019re all beta testers for the AI.<\/p>\n

Considering there are limited resources in this field and ample opportunity to do something much more exciting (and more profitable) than reliably turn on the lights, that\u2019s the way they\u2019re moving, according to experts I spoke with. Given all these factors, it seems the easiest way to improve the technology is to just deploy it in the real world and let it improve over time. Which is likely why Alexa Plus<\/a> and Gemini for Home <\/a>are in \u201cearly access\u201d phases. Basically, we\u2019re all beta testers for the AI.<\/p>\n

The bad news is it could be a while until it gets better. In his research, Dhruv Jain<\/a>, assistant professor of Computer Science & Engineering at the University of Michigan<\/a> and director of the Soundability Lab<\/a>, has also found that newer models of smart home assistants are less reliable. \u201cIt\u2019s more conversational, people like it, people like to talk to it, but it\u2019s not as good as the previous one,\u201d he says. \u201cI think [tech companies\u2019] model has always been to release it fairly fast, collect data, and improve on it. So, over a few years, we might get a better model, but at the cost of those few years of people wrestling with it.\u201d<\/p>\n

\"The<\/a><\/p>\n

The Alexa that launched in 2014 on the original Echo smart speaker isn\u2019t capable enough for the future Amazon is working toward. Image: Amazon<\/p>\n

The inherent problem appears to be that the old and new technologies don\u2019t mesh. So, to build their new voice assistants, Amazon, Google, and Apple have had to throw out the old<\/a> and build something entirely new<\/a>. However, they quickly discovered that these new LLMs were not designed<\/a> for the predictability and repetitiveness that their predecessors excelled at. \u201cIt was not as trivial an upgrade as everyone originally thought,\u201d says Mark Riedl, a professor at the School of Interactive Computing at Georgia Tech<\/a>. \u201cLLMs understand a lot more and are open to more arbitrary ways to communicate, which then opens them to interpretation and interpretation mistakes.\u201d<\/p>\n

Basically, LLMs just aren\u2019t designed to do what prior command-and-control-style voice assistants did. \u201cThose voice assistants are what we call \u2018template matchers,\u2019\u201d explains Riedl. \u201cThey look for a keyword, when they see it, they know that there are one to three additional words to expect.\u201d For example, you say \u201cPlay radio,\u201d and they know to expect a station call code next.<\/p>\n

\u201cIt was not as trivial an upgrade as everyone originally thought.\u201d<\/p>\n

\u2014 Mark Riedl<\/p>\n

LLMs, on the other hand, \u201cbring in a lot of stochasticity \u2014 randomness,\u201d explains Riedl. Asking ChatGPT the same prompt multiple times may produce multiple responses.<\/a> This is part of their value, but it\u2019s also why when you ask your LLM-powered voice assistant to do the same thing you asked it yesterday, it might not respond the same way. \u201cThis randomness can lead to misunderstanding basic commands because sometimes they try to overthink things too much,\u201d he says.<\/p>\n

To fix this, companies like Amazon and Google have developed ways<\/a> to integrate LLMs with the APIs at the heart of our smart homes (and most of everything we do on the web<\/a>). But this has potentially created a new problem.<\/p>\n

\u201cThe LLMs now have to compose a function call to an API, and it has to work a whole lot harder to correctly create the syntax to get the call exactly right,\u201d Riedl posits. Where the old systems just waited for the keyword, LLM-powered assistants now have to lay out an entire code sequence that the API can recognize. \u201cIt has to keep all that in memory, and it\u2019s another place where it can make mistakes<\/a>.\u201d<\/p>\n

All of this is a scientific way of explaining why my coffee machine sometimes won\u2019t make me a cup of coffee, or why you might run into trouble getting Alexa or Google\u2019s assistant to do something it used to do just fine.<\/p>\n

So, why did these companies abandon a technology that worked for something that doesn\u2019t? Because of its potential. A voice assistant that, rather than being limited to responding to specific inputs, can understand natural language and take action based on that understanding is infinitely more capable.<\/p>\n

\u201cWhat all the companies that make Alexa and Siri and things like that really want to do is chaining of services,\u201d explains Riedl. \u201cThat\u2019s where you want a general language understanding, something that can understand complex relationships through tasks and how they\u2019re conveyed by speech. They can invent the if-else statements that chain everything together, on the fly, and dynamically generate the sequence.\u201d They can become agentic.<\/p>\n

\u201cThe question is whether \u2026 the expanded range of possibilities the new technology offers is worth more than a 100 percent accurate non-probabilistic model.\u201d<\/p>\n

\u2014 Dhruv Jain<\/p>\n

This is why you throw away the old technology, says Riedl, because it had no chance of doing this. \u201cIt\u2019s about the cost-benefit ratio,\u201d says Jain. \u201c[The new technology] is not ever going to be as accurate at this as the non-probabilistic technology before, but the question is whether that sufficiently high accuracy, plus the expanded range of possibilities the new technology offers, is worth more than a 100 percent accurate non-probabilistic model.\u201d<\/p>\n

One solution is to use multiple models to power these assistants. Google\u2019s Gemini for Home consists of two separate systems: Gemini and Gemini Live. Anish Kattukaran, head of product at Google Home and Nest, says the aim is to eventually have the more powerful Gemini Live<\/a> run everything, but today, the more tightly constrained Gemini for Home is in charge. Amazon similarly uses multiple models<\/a> to balance its various capabilities. But it\u2019s an imperfect solution that has led to inconsistency and confusion in our smart homes.<\/p>\n

Riedl says that no one has really figured out how to train LLMs to understand when to be very precise and when to embrace randomness, meaning even the \u201ctame\u201d LLMs can still get things wrong. \u201cIf you wanted to have a machine that just was never random at all, you could tamp it all down,\u201d says Riedl. But that same chatbot would not be more conversational or able to tell your kid fantastical bedtime stories \u2014 both capabilities that Alexa<\/a> and Google<\/a> are touting. \u201cIf you want it all in one, you\u2019re really making some tradeoffs.\u201d<\/p>\n

These struggles in its deployment in the smart home could be a harbinger of broader issues<\/a> for the technology. If AI can\u2019t turn on the lights reliably, why should anyone rely on it to do more complex tasks, asks Riedl. \u201cYou have to walk before you can run.\u201d<\/p>\n

But tech companies are known for their propensity to move fast and break things. \u201cThe story of language models has always been about taming the LLMs,\u201d says Riedl. \u201cOver time, they become more tame, more reliable, more trustworthy. But we keep pushing into the fringe of those spaces where they\u2019re not.\u201d<\/p>\n

Riedl does believe in the path to a purely agentic assistant. \u201cI don\u2019t know if we ever get to AGI<\/a>, but I think over time we do see these things at least being more reliable.\u201d The question for those of us dealing with these unreliable AIs in our homes today, however, is are we willing to wait and at what cost to the smart home in the meantime?<\/p>\n

Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.Jennifer Pattison TuohyClose\"JenniferJennifer Pattison Tuohy<\/p>\n

Senior Reviewer, Smart Home<\/p>\n

Posts from this author will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All by Jennifer Pattison Tuohy<\/a><\/p>\n

AICloseAI<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All AI<\/a><\/p>\n

AmazonCloseAmazon<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Amazon<\/a><\/p>\n

Amazon AlexaCloseAmazon Alexa<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Amazon Alexa<\/a><\/p>\n

GoogleCloseGoogle<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Google<\/a><\/p>\n

Google AssistantCloseGoogle Assistant<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Google Assistant<\/a><\/p>\n

Smart HomeCloseSmart Home<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Smart Home<\/a><\/p>\n

TechCloseTech<\/p>\n

Posts from this topic will be added to your daily email digest and your homepage feed.<\/p>\n

FollowFollow<\/p>\n

See All Tech<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"This morning, I asked my Alexa-enabled Bosch coffee machine to make me a coffee. Instead of running my…\n","protected":false},"author":2,"featured_media":365785,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[31],"tags":[182,960,5252,168,65945,5254,172,74],"class_list":{"0":"post-365784","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-technology","8":"tag-ai","9":"tag-amazon","10":"tag-amazon-alexa","11":"tag-google","12":"tag-google-assistant","13":"tag-smart-home","14":"tag-tech","15":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/365784","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/comments?post=365784"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/posts\/365784\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media\/365785"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/media?parent=365784"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/categories?post=365784"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/us\/wp-json\/wp\/v2\/tags?post=365784"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}