Placeholder Image

字幕列表 影片播放

由 AI 自動生成
  • What's up, Josh here.


  • So today Google wrapped up IO 2024, where they announced several new AI powered features and integrations and generative AI that are really, really exciting and there's a lot going on.

    今天,谷歌在 IO 2024 上宣佈了幾項新的人工智能功能、集成和生成式人工智能,這些都非常非常令人興奮,有很多事情正在發生。

  • So in this video,


  • I'm gonna attempt to break down the two hour long presentation into something more manageable and hopefully help you figure out what's going on and how you can take advantage of these features for yourself.


  • So yeah, go ahead and drop a like, subscribe and let's get into it.


  • Now, if I could summarize this presentation into two sort of main buckets of features, it would be number one, integrations and number two, long context.


  • Now, integration is something that really only the Googles and Apples of the world can do.


  • I'm talking about companies that own different products that seamlessly integrate with each other, which makes it easy for them to sort of integrate a new feature, such as AI into all of these products.


  • So for a big portion of the presentation,


  • Google really showed off a bunch of ways that they're now integrating Gemini into all of these products, which in all of the demos is basically some variation of helping you find and organize information.


  • Now, one thing I'm sure we all deal with is emails.


  • And one of the most impressive integrations with Gmail was being able to go through your entire inbox to organize and track, let's say, receipts.

    與 Gmail 集成的最令人印象深刻的功能之一,就是可以查看整個收件箱,整理和跟蹤收據。

  • So in this demo, it found 37 receipts and also created an entire spreadsheet that really would have taken hours and hours to create if you were to do it on your own.

    是以,在這個演示中,它找到了 37 張收據,還創建了整個電子表格,如果你自己做的話,真的要花費好幾個小時。

  • Then once that spreadsheet was created, we also saw Gemini analyze that data, even creating a graph to visualize everything, which is pretty neat.


  • There was also another example where let's say you missed an entire email thread.


  • Well, now you can just ask Gemini to summarize the entire email thread for you.


  • And then based off of that summary can also help you draft up an email.


  • It can also summarize an entire recording of a video conference up to an hour long, as long as it's on Google Meet.

    只要視頻會議是在 Google Meet 上進行的,它還能總結長達一小時的整個視頻會議錄音。

  • Another example is within Google Photos where they rolled out Ask Photos, which allows you to just search for photos just like you would on Google, except it's your own library.

    另一個例子是在 Google Photos 中,他們推出了 Ask Photos,讓你可以像在 Google 上一樣搜索照片,只不過搜索的是你自己的資料庫。

  • Say you're at a parking station ready to pay, but you can't recall your license plate number.


  • Before, you could search photos for keywords and then scroll through years worth of photos looking for the right one.


  • Now you can simply ask photos.


  • It knows the cars that appear often, it triangulates which one is yours, and just tells you the license plate number.


  • And then within the Google Workspaces suite, they're rolling out side panels, which again is just another bridge to Google Gemini, which is just this floating window where you'll always have access to Gemini to search through your documents or even summarize them for you.

    在 Google Workspaces 套件中,他們還推出了側邊面板,這同樣只是通往 Google 雙子座的另一座橋樑,在這個浮動窗口中,你可以隨時訪問雙子座,搜索你的文檔,甚至為你彙總文檔。

  • So like I said, these are all examples of integrations within Google's existing products that allow you to organize and find information faster.


  • But then something a little bit interesting here is that they're integrating Gemini into Google Search.


  • And so you can actually enable this feature right now, and I think the presenter said it best about what this feature actually does.


  • And Google will do the Googling for you.


  • So yeah, she said Google will do the Googling for you, which is just ridiculous to say.


  • But yeah, the new Google Search powered by Gemini now includes AI overviews, where it gives high-level summary of the results with suggested links for you to check out, as well as multi-step reasoning where you can ask it really long and specific things, like for example, a yoga or Pilates studio in your city that's highly rated and within a half hour walk.

    是的,由 Gemini 提供支持的新版谷歌搜索現在包括人工智能概覽,它能提供高級別的搜索結果摘要,並建議您查看相關鏈接,還能提供多步驟推理,您可以向它提出非常長且具體的問題,例如,您所在城市中評價較高且步行半小時內的瑜伽或普拉提工作室。

  • And then it'll give you a list of studios for you to potentially check out.


  • You can also ask it to do a bunch of things like meal prep, plan a trip, or create a workout routine.


  • And you might be picking up on the fact that this is sort of blurring the lines between Google Search and Gemini, because they sort of have overlapping functionalities.


  • Now, I think the main difference between Search and Gemini is with Search, you're still getting human-generated like articles and blog posts and that sort of thing.


  • So for example, if you look for a meal plan on Google Search, it's still gonna give you human-created recipes.


  • Whereas on Gemini, it's just gonna give you meal plans that are written up by AI.


  • And overall, I think Search is still an extremely important product for Google to hold onto and sort of prolong the life of, because it is how they generate most of their income.


  • So yeah, that is the first majority of sort of like the bulk of what they announced is all of these different integrations into different apps.


  • But then the other thing that Google really leaned into is long context.


  • So in the presentation, we heard this word tokens a lot.


  • And while I still don't really understand what a token is, it seems to indicate that it is a chunk of information.

    雖然我還不太明白 "標記 "是什麼,但它似乎表明它是一大塊資訊。

  • And the big thing that Google kept repeating is their support for up to 1 million tokens in Gemini Pro.

    谷歌一直在重複的一件大事是,他們在雙子座專業版中支持多達 100 萬個代幣。

  • This basically means that Google's latest model is gonna have way more support for storing information, which is gonna be extremely useful for things like research, handling long documents, lines of code, and even analyzing video.


  • And yeah, this is really how Google is able to power a lot of the innovation that they're talking about, which includes looking through tons and tons of emails and documents that might be in your Google Drive.

    是的,這就是谷歌如何實現他們所說的大量創新的原因,其中包括查看 Google Drive 中的大量電子郵件和文檔。

  • And they also announced a couple of experimental apps that you can actually play with right now.


  • So there's this one called Notebook LM, where users can upload tons and tons of documents, charts, diagrams, and then have Gemini generate things like study guides, FAQs, quizzes, and even an AI generated like podcast type thing to help you understand concepts a little bit better.


  • Let's dive into physics.


  • What's on deck for today?


  • Well, we're starting with the basics, force in motion.


  • And that of course means we have to talk about


  • Sir Isaac Newton and his three laws of motion.


  • Ah, yes, the foundation for understanding how objects move and interact.


  • Join.


  • Hold on, we have a question.


  • What's up, Josh?


  • Yeah, can you give my son Jimmy a basketball example?


  • Hey, Jimmy, that's a fantastic idea.


  • Basketball is actually a great way to visualize force and motion.


  • Let's break it down.


  • Another app is called AI Studio, where again, you can upload entire research papers, code repositories, videos, photos.

    另一款名為 AI Studio 的應用程序同樣可以上傳整篇研究論文、代碼庫、視頻和照片。

  • And the way that I almost look at it is you're creating your own database for your own Google search, in a sense.


  • I think it's gonna be especially useful for researchers, students, and analysts.


  • These people who deal with tons and tons of data and documents and spreadsheets, which would otherwise take hours and hours to scrub through can now be done in literally seconds.


  • Now, I actually tested this for myself on a very small scale.


  • So I basically transcribed the entire Google I-O keynote and uploaded it to Gemini and was able to ask it questions about the presentation, which was great for research.

    是以,我基本上抄錄了整個谷歌 I-O 主題演講,並將其上傳到 Gemini,然後就可以向它提出有關演講的問題,這對研究工作非常有幫助。

  • Which by the way, that entire transcript, according to ChatGPT, was only 126 tokens.

    順便說一下,根據 ChatGPT 的數據,整個副本只有 126 個代幣。

  • So it really just goes to show how crazy one million tokens is.


  • It really isn't that necessary of a feature for the everyday user.


  • Now, I think the really weird thing with Gemini is that if you're on the free plan, you actually cannot upload documents, even though with all of these other experimental apps that are free, like Notebook LM and AI Studio, you can.

    現在,我覺得雙子座最奇怪的是,如果你使用的是免費計劃,你實際上無法上傳文檔,儘管其他所有免費的實驗性應用程序,如 Notebook LM 和 AI Studio,都可以上傳文檔。

  • And considering that OpenAI's ChatGPT 4.0, which is now made completely free, does allow you to upload documents on top of photos,

    考慮到 OpenAI 的 ChatGPT 4.0 現在已經完全免費,它確實允許您在照片上上傳文檔、

  • I really think that Google should add that functionality to the free Gemini plan.


  • Now, over on the mobile side of things,


  • Google announced Project Astra and gave us this really early look into this live interaction with Vision, which sort of reminds me of OpenAI's demo with 4.0.

    谷歌宣佈了 "阿斯特拉計劃"(Project Astra),並向我們展示了與 "視覺"(Vision)的實時互動,這讓我想起了 OpenAI 的 4.0 演示。

  • But yeah, in this prerecorded video, the presenter sort of walked around the room, pointing the camera at things and asking questions along the way with real-time responses.


  • What does that part of the code do?


  • This code defines encryption and decryption functions.


  • What neighborhood do you think I'm in?


  • This appears to be the King's Cross area of London.


  • Now, all of that is very impressive.


  • However, pointing your phone around is not the most ideal, so they showed this prototype off, which I thought was pretty interesting.


  • Do you remember where you saw my glasses?


  • Yes, I do.


  • Your glasses were on the desk near a red apple.


  • What can I add here to make this system faster?


  • Adding a cache between the server and database could improve speed.


  • So yeah, there may be a Google Glass resurrection maybe a couple of years down the line.


  • Now, in a really confusing way,


  • Google also teased Gemini Live, which seems to be the actual thing that consumers will be getting in a couple of months.

    谷歌還預告了 "雙子座直播"(Gemini Live),這似乎就是消費者幾個月後將獲得的實際產品。

  • So this is, again, a live conversational feature built right into Gemini that you can interrupt with your voice and it learns your speech patterns and can turn the camera on and point it at things.

    同樣,這是 Gemini 內置的實時對話功能,你可以用語音打斷它,它會學習你的說話模式,並打開攝像頭對準事物。

  • They didn't really demo that at all, but they did say that they are taking all of the learnings from Project Astra and putting it into Gemini Live.

    他們並沒有真正演示這一點,但他們確實說過,他們正在把從 Project Astra 學到的所有知識運用到 Gemini Live 中。

  • Now, within the Gemini Assistant, they also rolled out Gems, which is a feature that allows users to create customizable AI assistants for very specific tasks.


  • They also showed like a similar thing off in the workspace environment where you can sort of like create your own employee.


  • Now, the workspace example was a little bit interesting because that AI employee had access to every worker's chat history.


  • So it could sort of verify if a project was on track or how things are moving along, which I thought was a little bit strange.


  • But yeah, other than that, they didn't really go in depth in the whole like voice assistant aspect of things.


  • So we don't really know if the latency is coming down or if the voices are changing, but there are some other smaller features that they showed off on the mobile side, such as offering help with a video you're watching or being able to search within a PDF.

    是以,我們並不清楚延遲是否會降低,聲音是否會改變,但他們在移動端展示了其他一些較小的功能,例如為您正在觀看的視頻提供幫助,或者能夠在 PDF 文件中進行搜索。

  • And then specifically for Pixel devices, by leveraging Gemini Nano, which all of the processing happens entirely on device, it can actually look at your screen, read your conversations and suggest things to say in that conversation.

    然後,特別針對 Pixel 設備,通過利用 Gemini Nano(所有處理都完全在設備上進行),它實際上可以查看你的螢幕,讀取你的對話,並在對話中提出建議。

  • It can even listen into your phone calls and sort of detect whether if it's a scam or not, which is really wild.


  • I'm going to help you transfer your money to a secure account we've set up for you.


  • Now, finally, the last thing that I'll touch on is all of the generative AI that Google is working on, which is all sort of housed under Google Test Kitchen.

    最後,我要談的是谷歌正在研究的生成式人工智能,它被歸類為谷歌測試廚房(Google Test Kitchen)。

  • So music effects and video effects are completely new.


  • This is their music and video generative AI features.


  • With music effects, you can literally type out any instrument and it'll create a completely new beat and you can even layer on multiple instruments, which is crazy.


  • It's pretty tough actually.


  • Video effects is also completely new and showcases some really impressive work in terms of the physics behind it and the detail.


  • Photo effects is getting even more realistic now with AI generated imagery.


  • And Google also briefly touched on Synth ID, which is basically this tool that embeds these invisible watermarks on anything that is generated by AI so that we as humans are able to identify what works of arts has AI's fingerprints on it.

    谷歌還簡要介紹了Synth ID,它基本上是一種在人工智能生成的任何東西上嵌入隱形水印的工具,這樣我們人類就能識別出哪些藝術作品上有人工智能的指紋。

  • But yeah, that was Google I-O 2024.

    沒錯,那就是谷歌 I-O 2024。

  • Google really is going all in on AI and it's a little bit overwhelming as the consumer.


  • There's just so much going on.


  • I think as they're slowly rolling these features out, it's gonna take consumers a bit of time to sort of get used to the new workflow of things.


  • But once they do get used to it, it's gonna be a game changer.


  • Unfortunately for a lot of these things, we will have to wait a couple of weeks or even months to see these things getting rolled out.


  • So definitely get subscribed to see my coverage of those things when they do come out.


  • I will be leaving links down below where you can start to enable a lot of these experimental features as well as explore some of these experimental apps.


  • That's gonna do it for this video.


  • If you enjoyed it, leave a like on the way out and I'll catch you guys next time.

    如果您喜歡,請在離開時留下一個 "贊",我下次再來找你們。

  • Peace.


What's up, Josh here.


由 AI 自動生成

單字即點即查 點擊單字可以查詢單字解釋