TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Why Apple believes it’s an AI leader–and why it says critics have it all wrong

175 点作者 arunbahl将近 5 年前

25 条评论

jointpdf将近 5 年前
Setting aside all the other work Apple is doing in this area, can someone explain why iMessage autocorrect is unacceptably bad (to put it diplomatically) in the era of weapons-grade language models?<p>Its behavior is totally bizarre. It’s like an underpowered chess engine that makes flagrant blunders: capitalizing random words in the middle of a sentence (rock -&gt; Rock), the context sensitivity of an actual rock, forcing the same correction multiple times (i.e. you go back to fix its error, and it defiantly repeats it), contraction mixups, blindness to off-by-one-keystroke errors (consentuallt -&gt; <i>&lt;no action&gt;</i>), and of course the occasional random word substitution. Only martial artists want to duck people (and it—no joke—just now substituted [duck -&gt; suck]. You had one job autocorrect.)<p>Is this just me? Is it actually 2020? What is going on?
评论 #24089537 未加载
评论 #24090254 未加载
评论 #24089741 未加载
评论 #24089872 未加载
评论 #24089681 未加载
评论 #24090624 未加载
评论 #24089479 未加载
评论 #24089689 未加载
评论 #24089682 未加载
评论 #24089712 未加载
评论 #24090408 未加载
评论 #24089491 未加载
评论 #24092810 未加载
评论 #24091407 未加载
评论 #24089589 未加载
评论 #24090152 未加载
评论 #24094043 未加载
评论 #24091805 未加载
评论 #24089782 未加载
评论 #24090330 未加载
评论 #24092016 未加载
评论 #24091800 未加载
评论 #24092529 未加载
评论 #24089554 未加载
评论 #24089737 未加载
评论 #24091224 未加载
throwaway202020将近 5 年前
You can&#x27;t be AI leader when every AI leader is staying away from you by 10ft pole. When Ian Goodfellow joined Apple, there was literally rain of criticism on him from ending his career as researcher to bowing down to money. I don&#x27;t know of any researcher who does want to have continued research career willing to join Apple. They simply don&#x27;t allow that kind of freedom or publishing results. While Apple has some strong points in imaging (thanks to their 1000+ people team) and wrist rejection, virtually everything else they do that requires AI sucks and lags behind competition, including, Siri, maps, autocorrect, spell check, iCloud, search, calendar, spam detection, recommendations etc. For most of these things, most people don&#x27;t even count them as real competition. Google on the other hand is able to achieve very competitive performance through software and AI without such large team on phone camera and frankly quite pathetic hardware.<p>These kind of reality distortion pieces aren&#x27;t going to help them. They have $100+B, in cash, they can easily start reputable open research lab that can rival FAIR, OpenAI or DeepMind. Even a smaller companies like Intel and Adobe is starting to realize that this is necessary so they can tap into expertise on demand. At minimum that will be totally worth for a talent pipeline that can be motivated to do &quot;rotation&quot; or &quot;sabbaticals&quot; into product groups from an open lab.
评论 #24090481 未加载
评论 #24090161 未加载
评论 #24090390 未加载
评论 #24093551 未加载
评论 #24090112 未加载
评论 #24090028 未加载
评论 #24090985 未加载
lacker将近 5 年前
<i>&quot;Google is an amazing company, and there&#x27;s some really great technologists working there,&quot; he said. &quot;But fundamentally, their business model is different and they&#x27;re not known for shipping consumer experiences that are used by hundreds of millions of people.&quot;</i><p>What does he mean by this? Google search and Android are both used by more than a billion people. YouTube over 2 billion. These are all bigger than any Apple product. If anything, Google <i>is</i> known for shipping consumer experiences that reach a large number of people. Apple by contrast is known for its high-quality, high-price consumer experiences that reach fewer people.
评论 #24090373 未加载
评论 #24093757 未加载
评论 #24090431 未加载
评论 #24090418 未加载
cromwellian将近 5 年前
(Googler here) I feel article presents a false dichotomy&#x2F;view (imho borderline disingenuous impression) that Apple is doing on-device, and everyone else is doing on-cloud.<p>Android also does on-device. A lot of the time, things start out on-cloud, until whatever machine learning model can be shrunk to run on-device at the right performance. So you see for example, text-to-speech and speech-to-text start out as cloud calls, and now they&#x27;re on-device. Google Translate ran in the cloud, but now for some languages, it happens on-device.<p>Things like Google&#x27;s &quot;Live Transcription&#x2F;Caption&quot; on Android wouldn&#x27;t work if it wasn&#x27;t on-device.<p>Apple similarly went to the cloud for Siri speech recognition and TTS until they could run it locally.<p>For other things which need large models, there is Federated Learning to preserve privacy. Google Keyboard has been using Federated Learning for some time now.
Veedrac将近 5 年前
Apple&#x27;s AI work is decent for what it is, and their AI hardware is fine too, but so much of this article is just weird.<p>It starts by talking about how until recently Apple wasn&#x27;t doing AI work where it needed to be. Then there&#x27;s the weird excerpt where he claims Google is “not known for shipping consumer experiences that are used by hundreds of millions of people.” The author then raises the legitimate point that AI benefits from having lots of data to train on, but then quotes an answer by Giannandrea to a different question, which includes him stating that bigger models aren&#x27;t more accurate than smaller ones. The point that on-device inference is more responsive is valid but not unique to Apple; the article says “Android phones don&#x27;t do nearly as wide an array of machine learning tasks locally”, but I don&#x27;t think this is true.
评论 #24089498 未加载
评论 #24089726 未加载
评论 #24089520 未加载
Despegar将近 5 年前
This piece is interesting because Apple was saying this all along but no one really believed them because it sounded like excuse making. But here JG is basically saying the same thing:<p>&gt;Yes, I understand this perception of bigger models in data centers somehow are more accurate, but it&#x27;s actually wrong. It&#x27;s actually technically wrong. It&#x27;s better to run the model close to the data, rather than moving the data around. And whether that&#x27;s location data—like what are you doing— [or] exercise data—what&#x27;s the accelerometer doing in your phone—it&#x27;s just better to be close to the source of the data, and so it&#x27;s also privacy preserving.<p>A few years ago was when this narrative was at its peak and I believe it was mostly because Google (and to a lesser extent Facebook) were talking about machine learning and AI in basically every public communication. What came of it? Were all the people who claimed Apple&#x27;s privacy stance would leave them in the dust proven right? For one, being &quot;good at machine learning&quot; is like saying you&#x27;re good at database technology. It&#x27;s a building block, not a product. Maybe Google and Facebook are doing cutting edge research in the field, but so was Xerox PARC.
评论 #24089096 未加载
评论 #24088967 未加载
评论 #24088911 未加载
评论 #24089071 未加载
georgespencer将近 5 年前
I don&#x27;t know anyone who believes that Siri is as good as Alexa. I spent four months self-isolating in an apartment with the small hockey puck Alexa devices in random corners and returned to my big open plan apartment with two paired HomePods on the kitchen counter.<p>The frequency with which Siri shits its pants (can&#x27;t help, asks me to excuse it being slow as it tries to set a timer, mishears me, etc.) is honestly remarkable.<p>(Not to mention the fact that my phone continues to alert me to text messages read on my Mac or iPad whole minutes ago.)<p>Apple is still working to overcome deep problems in both its cloud infrastructure and AI&#x2F;ML. If they cannot be honest about this, they should not be dishonestly trying to present a picture of all being well.
评论 #24093999 未加载
tialaramex将近 5 年前
Show don&#x27;t tell. You don&#x27;t lead in AI or anything else by insisting everybody who noticed you&#x27;re bad at something &quot;has it wrong&quot;.
评论 #24089154 未加载
bambax将近 5 年前
&gt; <i>Machine learning is used to help the iPad&#x27;s software distinguish between a user accidentally pressing their palm against the screen while drawing with the Apple Pencil, and an intentional press meant to provide an input. It&#x27;s used to monitor users&#x27; usage habits to optimize device battery life and charging, both to improve the time users can spend between charges and to protect the battery&#x27;s longterm viability. It&#x27;s used to make app recommendations.</i><p>The problem with machine learning altering the behavior of a device is it shortcuts human learning. The human brain is very good at learning deep insights about things and its environment and alters its behavior accordingly.<p>If things change while we&#x27;re learning about them, it confuses and upsets us. A dumb machine is much easier to use than a &quot;smart&quot; one.
评论 #24091608 未加载
jml78将近 5 年前
Did the interviewer not know enough about ML to even challenge the bogus statements.<p>Yes, apple’s strategy is more privacy protecting.<p>But holy hell, yes larger data sets are going to be more accurate and the resultant model doesn’t have to run in the cloud, it can run locally.<p>AI training in a data center with large data sets, ship the model to local devices to execute.<p>That is ALWAYS going to be better and more accurate than what Apple is doing.
Spooky23将近 5 年前
I agree that Apple is underrated here, and I think their rigid user interfaces for things like Photos hides the power of their platform.<p>It seems absurd to let Google Photos slurp up your data server side when your iPhone can do 80+% of the photo categorization automatically. It’s equally absurd that Apple has a glacial pace of change for the user side.
评论 #24088960 未加载
评论 #24088867 未加载
评论 #24088868 未加载
评论 #24088805 未加载
LeicaLatte将近 5 年前
Fluff Piece. Apple PR machine learning that techies read ars Technica.
评论 #24089434 未加载
hepinhei将近 5 年前
An interview full of nothing.... When they say Google has no experience shipping user experience used by millions of people
majestik将近 5 年前
&gt; After a brief pause, he added: &quot;I guess the biggest problem I have is that many of our most ambitious products are the ones we can&#x27;t talk about and so it&#x27;s a bit of a sales challenge to tell somebody, &#x27;Come and work on the most ambitious thing ever but I can&#x27;t tell you what it is.&#x27;&quot;<p>JG, I don’t think that’s your “biggest problem” - Siri is. Your privacy centric on-device strategy limits your view of user feedback, Google gets a lot of shit wrong but they know how to transmit user data and understand their feedback.
pcr910303将近 5 年前
I like this article. I like Apple’s approach to ML because it blends in. When applied, the feature should not expose that it’s based on ML — that’s a failure. And so Siri, Alexa, Google Assistant are failures. But Face ID, plam rejection are successes.<p>If you have to explain the customers that it’s ML based, that’s the same as asking for the customers to understand it’s unreliability. And unreliable features are worse than no features, and that’s why nobody uses Siri, Alexa, Google Assistant except for a few reliably-working requests.
segmondy将近 5 年前
Apple is a fashion house. They will be around for quite a while, and still have time to turn things around, but if they keep at their current pace, they will be back to irrelevance in 2 decades. Apple builds beautiful looking hardware. Software wise? Complete garbage.
Syeposxr将近 5 年前
It&#x27;s the decline of Intel and the advancements Apple has made in designing its own processors that I think are really interesting. Having powerful and efficient processors on mobile devices - both laptops and phones - allows Apple to do edge computing in a way other companies haven&#x27;t been able to, and integrate ML in much more privacy-focused way.
m3kw9将近 5 年前
Aside from what they say, they still can’t get their iOS AI based spelling correction up to par.
NicoJuicy将近 5 年前
Don&#x27;t have iOS. But I do wish that Android with detect the current language you are writing in.<p>I have 3 languages on my keyboard and sometimes it suggests an autocomplete in the wrong language.
评论 #24090602 未加载
lowdose将近 5 年前
It&#x27;s like saying you are the smartest girl of the city.<p>When you have to say it yourself you probably aren&#x27;t.
amelius将近 5 年前
NVidia thinks they&#x27;re an AI leader too, and they&#x27;re probably closer to the truth.
innagadadavida将近 5 年前
Funnily, the one data that seems to matter for Siri is the web and wiki data for question answering. Siri still uses wolfram for many trivial questions. None of this is about privacy or user data and Siri is behind the state of the art here.
notsureaboutpg将近 5 年前
Apple is behind in AI. Google has Google Lens, real-time transcription of audio recordings, live captions. And Google Assistant is leaps and bounds ahead of Siri in nearly every way.<p>What is there in AI that Google doesn&#x27;t beat Apple at?
评论 #24091207 未加载
评论 #24089228 未加载
评论 #24089260 未加载
29athrowaway将近 5 年前
What is AI? Your washer machine and some rice cookers use AI (fuzzy logic)<p>Everything uses AI. A program with an if statement is AI.
f2322323ffff3将近 5 年前
One thing where apple is shining is brainwashing. It looks very funny when caravans of people are waiting store opening to hysterically buy an overpriced smartphone, even if it&#x27;s technologically 2 years behind cheaper Samsung S20 (except chip performance, but it doesn&#x27;t matter anymore, because there is no tasks that needs so much computing power)
评论 #24088872 未加载
评论 #24088966 未加载
评论 #24088853 未加载