Amazon's Big Dreams for Alexa Fall Short (ft.com) 58
It has been more than a decade since Jeff Bezos excitedly sketched out his vision for Alexa on a whiteboard at Amazon's headquarters. His voice assistant would help do all manner of tasks, such as shop online, control gadgets, or even read kids a bedtime story. But the Amazon founder's grand vision of a new computing platform controlled by voice has fallen short. From a report: As hype in the tech world turns feverishly to generative AI as the "next big thing," the moment has caused many to ask hard questions of the previous "next big thing" -- the much-lauded voice assistants from Amazon, Google, Apple, Microsoft and others. A "grow grow grow" culture described by one former Amazon Alexa marketing executive has now shifted to a more intense focus on how the device can help the ecommerce giant make money. "If you have anything you can do that you might be able to directly monetise, you should do it," was the recent diktat from Amazon leaders, according to one current employee on the Alexa team.
Under new chief executive Andy Jassy's tenure this change of focus has resulted in significant lay-offs in Amazon's Alexa team late last year as executives scrutinise the product's direct contribution to the company's bottom line. The belt-tightening came as part of broader cuts that have seen the ecommerce giant slash 18,000 jobs across the group amid pressure to improve profits during a global tech downturn. At Microsoft, whose chief executive Satya Nadella declared in 2016 that "bots are the new apps," it is now acknowledged that voice assistants, including its own Cortana, did not live up to the hype. "They were all dumb as a rock," Nadella told the Financial Times last month. "Whether it's Cortana or Alexa or Google Assistant or Siri, all these just don't work. We had a product that was supposed to be the new front-end to a lot of [information] that didn't work." Nadella can afford to be blunt: Microsoft's recent introduction of AI chatbot ChatGPT to its Bing search engine means the company is now seen as a leader in the field, having previously been mostly forgotten by the majority of internet users. ChatGPT's ability to understand complex instructions left existing voice assistants looking comparatively stupid, said Adam Cheyer, the co-creator of Siri, the voice assistant acquired by Apple in 2010 and introduced to the iPhone a year later.
Under new chief executive Andy Jassy's tenure this change of focus has resulted in significant lay-offs in Amazon's Alexa team late last year as executives scrutinise the product's direct contribution to the company's bottom line. The belt-tightening came as part of broader cuts that have seen the ecommerce giant slash 18,000 jobs across the group amid pressure to improve profits during a global tech downturn. At Microsoft, whose chief executive Satya Nadella declared in 2016 that "bots are the new apps," it is now acknowledged that voice assistants, including its own Cortana, did not live up to the hype. "They were all dumb as a rock," Nadella told the Financial Times last month. "Whether it's Cortana or Alexa or Google Assistant or Siri, all these just don't work. We had a product that was supposed to be the new front-end to a lot of [information] that didn't work." Nadella can afford to be blunt: Microsoft's recent introduction of AI chatbot ChatGPT to its Bing search engine means the company is now seen as a leader in the field, having previously been mostly forgotten by the majority of internet users. ChatGPT's ability to understand complex instructions left existing voice assistants looking comparatively stupid, said Adam Cheyer, the co-creator of Siri, the voice assistant acquired by Apple in 2010 and introduced to the iPhone a year later.
Bad Alexa ! (Score:2)
Visionary Leadership (Score:5, Interesting)
No $500 Million Dollar Yacht for you! OK maybe just one.
Alexa and related appliances are a market failure, but take note of the leadership involved in that decision.
Lots of people have 1 good idea and make a business out of it: Zuckerberg had an idea for facebook (*), Sergey Brin and Larry Page had an idea for searching the net (and all the other google products were ideas made by employees on their personal time, and most of those were derivative), Bill Gates had an idea for an operating system (which was mostly derivative), and so on.
A very few entrepreneurs are true leaders in the sense that they take their companies in new directions, or make new companies. Elon Musk is the most notable, but also Steve Jobs and Jeff Bezos.
Amazon started out as an online used bookseller, and Bezos used his personal vision to morph it in different directions to where it's now the dominant player in a bunch of categories such as online sales and web/computer services (AWS). He's exploring selling food, he's revamped how warehouses are managed and how books are packaged and sent, and generally made a bunch of improvements to the purchase and supply chain. He's looking into selling discount medicine.
Alexa is not a resounding success, but it is still a product and seems to fulfill the imaginary niche that managers thought would be the next big thing: computers that can control your house ("Alexa - dim the living room lights") and computer simple voice commands ("Alexa - purchase more printer toner"). There just wasn't enough demand for automation in the home, and having the voice command functionality available wasn't enough of a draw for most people. With the singular exception of the Ring doorbell, Alexa and home control was a solution in search of a problem.
I find the failure of Alexa interesting from a business management point of view, and note that Jeff Bezos, for all the grief we give him about how he treats warehouse employees, is an example of management vision. Just about everyone else is a one-hit-wonder in the business world.
There's no problem with being a one-hit-wonder, most companies are based on that, but it makes the few leaders with vision stand out all the more.
(*) There's some claim that it wasn't his idea
Re: (Score:2)
It's almost as if they hit the jackpot once and everything since is either someone else's work, plain luck or in most other cases and utter and complete failure.
Almost like these people are not the geniuses they're paraded out to be but just pissing in the wind and gotten lucky once.
Re: (Score:2)
"Alexa is not a resounding success, but it is still a product and seems to fulfill the imaginary niche that managers thought would be the next big thing: computers that can control your house ("Alexa - dim the living room lights") and computer simple voice commands ("Alexa - purchase more printer toner")"
And the students programming a robot arm, Alexa, jerk me off.
Re: (Score:2)
And the students programming a robot arm, Alexa, jerk me off.
AlexaaAARRGHFUCKGODNOARRGHHHARGGHHH!!!!!
Re:Visionary Leadership (Score:4, Interesting)
I think the problem is with the niches it fulfils, it's not very good for the job.
It can control smart home devices, but easily can get confused if there are names that are too close (kitchen vs kitchen light) and it can run into problems with ghost devices. Something like a ChatGPT AI would help there with remembering how people like to call things.
The other part which is where Amazon probably hoped to make money with people ordering products off of it, but that's been ruined by Amazon's opening up of their marketplace to less reputable sellers. When I order something I want to see how it compares in prices to others and reviews and such. Even re-ordering faces the same problem with sometimes large swings in prices for the same product
Re: (Score:2)
Re: (Score:2)
Bill Gates had an idea for an operating system (which was mostly derivative)
Not just derivative but actually written by someone else and then licensed by Microsoft to IBM for $50,000. Gates didn't have an idea for the OS. Rather, IBM needed an OS and asked Gates if he could provide one. Gates originally suggested CP/M but IBM's negotiations with Digital didn't go well. That's when Gates licensed what became DOS, after the original creator modified it for IBM.
By all accounts, Gates was smart, academically as well as with computers. However, there are tons of smart people. Gate
Re: Visionary Leadership (Score:1)
Oh, honey
*no*
Re: (Score:2)
Jeff Bezos is an example of management vision... the few leaders with vision stand out all the more.
I wonder what does it mean to be a "visionary" leader? Either your decisions are based on real data that makes a compelling case, or they're not. If you make data-driven decisions then no one calls you visionary. Therefore: a visionary leader is one who makes decisions unsupported by evidence.
My personal story (Score:5, Insightful)
I only bought an Echo around Christmas for like $25, figuring I'd use it for light control in my garage and perhaps music.
Long of the short of it: they're infuriating garbage.
* The audio quality is atrocious. A $5 bluetooth -> 3.5mm jack adapter has better audio quality.
* It's basically just an Amazon ad box. I set up Apple Music on it and within a week it had updated itself to use Amazon Music. It likes "helping" me with recommendations on what Amazon Music can do.
* It constantly disconnects from wifi and chimes at me, even though the connection is strong for everything else.
I never even got around to using it for its intended purpose, it was simply too infuriating. Took it out and threw it out.
Re: (Score:3)
I only bought an Echo around Christmas for like $25, figuring I'd use it for light control in my garage and perhaps music.
Long of the short of it: they're infuriating garbage.
* The audio quality is atrocious. A $5 bluetooth -> 3.5mm jack adapter has better audio quality. * It's basically just an Amazon ad box. I set up Apple Music on it and within a week it had updated itself to use Amazon Music. It likes "helping" me with recommendations on what Amazon Music can do. * It constantly disconnects from wifi and chimes at me, even though the connection is strong for everything else.
I never even got around to using it for its intended purpose, it was simply too infuriating. Took it out and threw it out.
Perhaps even Alexa is trying to tell you that inviting Amazon into your house to eavesdrop on your everything was really a bad, bad idea.
Re: (Score:2)
I've got three of the previous generation puck-shaped Echo Dots. They primarily get used for controlling the lights, and as alarm clocks/timers. I haven't had any connection issues unrelated to Spectrum being a heaping pile of horse excrement. In fact, asking "Alexa, are you connected to the internet?" has been a helpful way to determine if the device (laptop/tablet/etc.) I'm using needs to be rebooted because the WiFi driver glitched or if we're just having yet another of the many brief outages Spectrum
Re: (Score:2)
Re: (Score:1)
I have mine connected to external speakers through the 3.5" jack. The sound quality is worse than anything I've ever heard, tbh.
That's pretty hard to fuck up.
HAL 9000 (Score:3)
Yes, I know HAL was supposed to be a TRUE AI, but there was nothing in the movie, or the book that required real intelligence, and his solution to the programming conflict was quite binary.
Re: (Score:2)
Alexa and ChatGTP
This is prescient because we gotta know Microsoft is already working on Cortana V2 with ChatGTP integrated into it because it actually makes sense giving these systems more ability to work contextually and with fuzzy details.
That doesn't address the privacy concerns or other issues with these systems, it probably makes them all worse, but it's coming, it's inevitable.
Re: HAL 9000 (Score:2)
Re: (Score:2)
"if you combined Alexa and ChatGTP," you'd have a dim bot slinging ads at you. It won't be long before MS decides they must "monetise" ChatGTP in order to satisfy the bean counters.
Re: HAL 9000 (Score:2)
Currently, it's not immediately in the cards.
Generally, machine learning is considered well suited for feeding data and providing output to humans. However it's glitchy enough that you wouldn't want it in the control path for anything important without a human intermediary. It's funny that it mistakes Saturn for Jupiter when feeding it to a human, but it can be very bad if it mistakes unlock for lock in controlling security.
Re: (Score:2)
Re: (Score:2)
A Confusing Mess (Score:5, Insightful)
I tried to write a small app that called the Alexa API to do something simple. It was a painful experience, even for me, a developer with 30+ years experience. I dropped it after that, as I'm sure many other curious developers did.
Re: (Score:3)
Re: (Score:2)
Yeah, tried to build an integration myself to turn a roof heating wire on and off based on weather conditions
Call me crazy, but wouldn't it have been simpler just to use a thermostat?
Re: (Score:2)
Re: (Score:2)
I did the same thing a few years ago. The documentation was crap and the API was awkward just like all other Amazon software, but I was finally able to cause a transcribed voice command to arrive at an external HTTP service I wrote. In principle Alexa could be the voice interface for all kinds of things.
It made zero money for Amazon so I wasn't surprised when it stopped working after a while.
Google assistant doesn't work? (Score:5, Interesting)
Google assistant works fine for me. I don't know what fancy expectations people might have, but it consistently does exactly what I ask and has for years.
Re: (Score:1)
The Echo works fine (Score:3)
I have five of them controlling the house, playing music, handling alarms and timers and calculations, telling me the news and weather, and that my packages have been delivered.
Sure, the voice recognition isn't always perfect. I don't see how it could be helped by ChatGPT, which isn't a voice engine at all. This is just more advertising for ChatGPT, and it's sure getting a lot of advertising dollars for a mindless text mangler.
Of course I've disabled the ability to buy things by voice command. The last thing I need is a funny guest deciding to buy the 500-lb lot of Cheetos.
Re: (Score:2)
ChatGPT might do a better job of turning the voice parse tree into a user intent than the current Alexa intent engine.
Re: (Score:2)
The near future (Score:2)
"Hello Tim, this is Alexa. I notice your footsteps have been sounding heavier lately, would you like to purchase our special deal on workout gear?"
"Would you like to listen to a story?"
(one hour into story) "Thanks for starting your paid Audible trial!"
"Hello Tim, we heard you talking about Covid recently and we have some corrections for you based on what you said."
Re: (Score:1)
Well done!
For my money, that's absolutely the best post I've ever seen from you - not to mention one of the very few that I haven't disagreed with to some extent. Not that there's any reason my approval should matter to you - just sayin'.
Re: (Score:1)
Thanks, appreciate the nice comment!
New tech, same problem (Score:5, Insightful)
It seems to me that the biggest problem with getting money out of assistants that you speak to is that advertising won't work, and that will also be a problem for the ChatGPT's descendants.
So what does that leave for monetization? One scary thought is a sort of paid placement, where the search/assistant company is paid to recommend one result over another, or in a more dystopian vision, paid to verbally manipulate users into buying or doing something. This is way worse than existing paid placement, since people are not going to have multiple choices in front of them on a page with ad markers.
And I don't think users will accept an AI assistant that clearly delineates ads from content and rattles on with multiple choices. The only workable path I can see to monetization right now is a direct charge to the user, I guess similar to what OpenAI is doing now. But that just highlights even more clearly that these services intend to get paid for someone else's content.
Re: (Score:2)
It seems to me that the biggest problem with getting money out of assistants that you speak to is that advertising won't work, and that will also be a problem for the ChatGPT's descendants.
Agreed, though ironically, it's one of the things Amazon might be uniquely in a place to solve: Make it a part of Prime.
There are four basic ways to charge for a service like this: Subscriptions, Ads, Personal Data Sales, Bigger Purchase. We agree that ads aren't really viable for a voice assistant. Amazon already has a gold mine of personal data based on shopping habits, so Alexa is going to be a nominal improvement on that. Bigger Purchase is what Apple does; Siri is essentially covered by hardware sales,
Re: (Score:2)
I'm one of those people that uses Alexa for home control, but I also find it useful for quick answers to stupid questions, you know, like settling a argument or confirming some speculation when talking with my wife. I am pretty happy with the home control aspects, but I would pay more for a more capable "oracle". I could even see myself talking to it to research some ideas while resting my eyes in the recliner. If they added that to Prime, that would be perfect, since I already pay for it.
And yes, I kno
Re: New tech, same problem (Score:1)
"You're right" is the correct way to settle any argument with your wife.
Re: (Score:2)
As is regularly pointed out here, I'm not that smart.
Re: (Score:1)
Pretty happy with Google Assistant (Score:3)
I dunno about this article. I'm pretty happy with Google Assistant, but my expectations were not as high, because I knew the limits of the technology. Given those limits, it really does an amazing job and with Android Auto provides a consistent voice operated interface for any car I drive that has it.
As for Microsoft being "seen as a leader in the field," I think Microsoft is this author's darling. Bing with ChatGPT was a half-baked release done too soon that is an embarrassment for the company, IMO. And it also has limitations that are not in line with expectations and marketing promises. So this author will probably see it as a failure after they get over the new hotness of it.
It feels like all these tech giants are looking for the "next big thing," when all they need is to improve the stuff that already works pretty well. Alexa is a failure because it privileges Amazon in its design. Cortana was a failure because it was tied to Windows, which is not a phone. A phone, IMO, is the use case for the thing. Siri and Assistant are success stories to me, and I hope the "next big thing" doesn't hamper their continued development and improvement.
Agile development has become software ADHD. Make what works work better please, guys.
amazons fault it is falling short (Score:2)
It is amazon's fault it is falling short. They have a great system but they wanted to commercialized it and that is where they messed up. People wanted to love it, use it, and make it a part of their lives but the ads, the by the ways, the do you knows and all the other non ai nonsense amazon added to it prevent people from using it as they intended.
Maybe people don't want to be eavesdropped on 24/7 (Score:4, Insightful)
Maybe people don't want a huge corporation listening to every word they say in their own homes 24/7 ?
I'm a techie, but my wife is not, and she is even *more* insistent that we not have any live microphones in the house. It got to the point that she was about to veto our new TV purchase because it had Alexa "built-in", but we compromised by getting the TV but never connecting it to the internet. (OLED TVs that *don't* have Alexa or some other assistant built-in are hard to find)
Re:Maybe people don't want to be eavesdropped on 2 (Score:5, Interesting)
After the second law-enforcement abuses were reported I ditched the Alexas.
Maybe I'm rare but if I'm not they can't rebuild after that
Re: Maybe people don't want to be eavesdropped on (Score:2)
I like my home quiet. I don't want to hear some gizmo gabbering at me and I don't want to be yelling at it until it finally understands what I'm telling it...when the alternative is pushing some buttons with my fingers...*quietly*
Re: (Score:2)
Star Trek gets it right again. (Score:2)
Re: (Score:2)
Unsurprising (Score:2)
Alexa can go sit on the loser step along with Bixby and Cortana.
PHB-to-English translation (Score:1)
> "If you have anything you can do that you might be able to directly monetise, you should do it," was the recent diktat from Amazon leaders
= More snooping & spamming.
> ChatGPT's ability to understand complex instructions left existing voice assistants looking comparatively stupid
ChatGTP doesn't understand squat, it just finds, mashes, and echoes bigger overall patterns in its snoopbase/scraperbase.
Just no interest (Score:2)
I have no interest in talking to machines. That includes audio speakers, my tv, my computer, my car, some "virtual assistant" on my phone, those interactive voice things on the other end of 800 numbers, or anything else.
I'll make an exception for a realistic sex robot.
Self-serving product purchases is the issue. (Score:1)
They wanted everyone to easily get into the habbit of ordering things regularly using their Alexa as a convenient method.
It fails hard the same way most marketing it does. You can't trust the corporation to handle your purchases when they have a vested interest in controlling what you buy for profit.
Reasons I do not trust it to make financial decisions for me like purchases:
1. I can't trust it's recommended product. It will recommend bad products or items that don't match my needs in favor of matching the
Once upon a time (Score:1)
I worked at Amazon Games about 5 years ago. Back then, Amazon was so protective of the Alexa team's first-to-market advantage that their team had the right of first refusal for any new hire we wanted to make; before we could offer on any candidate they were given the opportunity to swoop in and poach them.
Now, I see the investment totally scaling back and the quality declining as the market becomes saturated and the future of AI seems to have moved elsewhere. They have a pretty excellent natural language pa