擒数网（随信APP) | iOS 18.2 发布了，苹果 AI 新功能再升级！第一批用户体验报告来了，iPhone 有了哪些改进？

文章目录[隐藏]

文生图、图生图，还能草图变大作
你的智能「机械眼」
更聪明的 Siri 来了
Image Creation, from sketches to masterpieces
Your intelligent "mechanical eyes"
A smarter Siri is here
Text and visuals, all within reach

擒数网（随信APP) | iOS 18.2 发布了，苹果 AI 新功能再升级！第一批用户体验报告来了，iPhone 有了哪些改进？
【微信/公众号/视频号/抖音/小红书/快手/bilibili/微博/知乎/今日头条同步报道】

iOS 18.2 的首个开发者测试版来了，这也是最接近下周正式推送的版本。

这一 beta 版本继续履行了 WWDC 2024 所推出的 Apple Intelligence「分阶段升级」的承诺，为开发者用户带来了属于苹果 AI 的四大新功能，其中就包括接入了 ChatGPT 的「更智能」的全新 Siri。

▲随 iOS 18.2 beta 版一同推出的 Apple Intelligence 新功能（图源：Perplexity）

此外，新版的 Apple Intelligence 将支持更多的地区和语言，但目前仍不支持国区。对此，苹果公司 CEO 蒂姆·库克表示「背后有非常具体的监管流程，我们正在努力推进」。

既然暂时无法上手体验，我们可以跟随 Youtube 视频网站上的数码博主 Brandon Butch 的视角，看看海外网友对于最新 Apple Intelligence 的体验和评价如何。首先来盘点一下本次苹果 AI 引人注目的四大新功能：

Image Creation：包括 Image Playground 图像游乐场、Genmoji 生成式 AI 表情符号和 Image Wand 图像棒；
Visual Intelligence 视觉智能（仅 iPhone 16 系列支持）；
集成了 ChatGPT 的最新人工智能语音助手 Siri；
Apple Intelligence 加持下的 Writing Tools 写作工具更新。

除上述新功能以外，最令人的惊喜的是 ChatGPT 的深度内置。不仅 Siri，还有很多其他功能也集成了 ChatGPT。并且值得一提的是，用户无需登录 ChatGPT 的账户就可以在支持其内置的功能中使用这个插件，诚意满满。

文生图、图生图，还能草图变大作

Image Creation 是本次 Apple Intelligence 的一个重大更新。相比于 Siri 的实用属性，这个功能则更加偏向娱乐。

其中，在 Image Playground 图像游乐场中，用户可以自由选择所给元素并融合生成几幅图像：

▲ 博主使用 Image Playground 功能生成图像

用该功能生成的图像可以切换「动漫」和「插画」两个风格，都很讨喜：

▲ 博主将生成的图像从「动漫」切换为「插画

此外，用户还可以添加提示文本，同所选元素一起生成想要的画面，自然也可以通过文本直接生成：

▲ 博主添加了提示文本，并重新生成了一幅图像

这是以歌手 Taylor Swift 和篮球运动员 Shaquille O’Neal 为原型，使用 Image Playground 生成的 Q 版头像，可爱又有辨识度。

同属于 Image Creation 的 Genmoji 生成式 AI 表情符号则更强调社交属性。它可以通过文本描述和用户的个人形象生成专属自己的 emoji，还可以用于后续的短信聊天当中：

▲ 博主使用 Genmoji 功能创建了一个属于自己的表情

Image Wand 图像棒功能则允许用户在笔记中绘制草图，并将其加以润色和细节补充，使其变成一幅完整的，色彩结构丰富的图像。同样，用户也可以通过提示文本对生成的结果加以约束：

▲博主使用 Image Wand 功能润色了一幅草图

有意思的是，如果草图或者提示文本中出现了「人」，那么 Image Wand 将会自动忽略，只生成除「人」以外的其他部分。其中原因，也许是出于隐私保护或者伦理的考虑，也许是这个功能还不健全，我们不得而知。

可以注意到，本次更新中 Image Creation 所创造出来的图像，相比于 Apple Intelligence 刚刚发布时所使用的令无数人诟病的示例图片，在质量和观感方面都有不少提升。我们期待着它在正式发布时会变得更好。

你的智能「机械眼」

苹果公司给搭载了「相机控制」按钮的 iPhone 16 系列机型开发了全新的 Visual Intelligence 视觉智能功能。这一功能可以帮助用户进行识物，获取关于物品的更多信息，还可以提供相应的购买渠道：

▲博主使用 Visual Intelligence 识物

用户也可以直接调用 ChatGPT 对取景框中的物品进行描述：

▲博主用 ChatGPT 直接描述索尼储存卡

此外，对于建筑物一类的对象，Visual Intelligence 可以告诉你它的地理位置；对于餐厅等地点，它甚至可以为用户提供营业时间和菜单等相关信息：

▲博主使用 Visual Intelligence 识别图片中的地点

在我们日常出行和生活越来越便利的今天，Visual Intelligence 智能视觉的终极目标，就是要成为用户的智能「机械眼」，让用户通过苹果 AI 观察世界，进一步降低获取相关信息的成本，提高便利性和效率。

更聪明的 Siri 来了

回顾此前 Brandon Butch 对于支持 Apple Intelligence 的初代 Siri 的体验，我们可以得到的结论是其「智商有待提高」。

相比于 Image Playground 的娱乐属性，Siri 的升级更加实用而且「迫切」。

接入了 ChatGPT 的 2.0 版本 Siri 则可以直接基于 ChatGPT 提供的资料，对于用户的提问给出回答：

▲博主在询问 Siri，Siri根据 ChatGPT 回答

这相当于给 Siri 接入了一个更为智能的「第二大脑」，实实在在地提高了 Siri 的「智商」。

当然，用户也可以在询问时选择是否调用 ChatGPT，甚至可以在提问中直接要求 Siri「询问 ChatGPT 并给出回答」：

The first developer beta of iOS 18.2 has arrived, which is also the closest version to the official release next week.

This beta version continues to fulfill the promise of Apple Intelligence's "phased upgrades" introduced at WWDC 2024, bringing four new features belonging to Apple AI to developer users, including the "smarter" new Siri powered by ChatGPT.

▲ Apple Intelligence new features released with iOS 18.2 beta version (Source: Perplexity)

Furthermore, the new version of Apple Intelligence will support more regions and languages, but currently does not support the Chinese region. In response to this, Apple CEO Tim Cook stated, "There are very specific regulatory processes behind it, and we are working hard to move forward."

Since it is temporarily unavailable to experience firsthand, we can follow the perspective of digital content creator Brandon Butch on the Youtube video platform to see how overseas users experience and review the latest Apple Intelligence. Let's first take a look at the four major new features of Apple AI:

Image Creation: including Image Playground, Genmoji, and Image Wand;
Visual Intelligence (only supported by iPhone 16 series);
The latest artificial intelligence voice assistant Siri integrated with ChatGPT;
Writing Tools updated under the blessing of Apple Intelligence.

In addition to the above new features, the most surprising aspect is the deep integration of ChatGPT. Not only Siri, but many other features also integrate ChatGPT. It is worth noting that users can use this plugin in supported functions without logging into ChatGPT account, showing great sincerity.

Image Creation, from sketches to masterpieces

Image Creation is a major update of Apple Intelligence this time. Compared to the practical attributes of Siri, this function leans more towards entertainment.

In the Image Playground, users can freely choose the elements provided and merge them to create several images:

▲ Blogger uses Image Playground to generate images

The images generated with this function can switch between "anime" and "illustration" styles, both of which are charming:

▲ Blogger switches the generated image from "anime" to "illustration"

Additionally, users can add text prompts, generate desired scenes along with selected elements, and also generate directly through the text:

▲ Blogger adds text prompts and regenerates an image

Using Taylor Swift and basketball player Shaquille O'Neal as prototypes, the Q version avatars generated using Image Playground are cute and recognizable.

As part of Image Creation, Genmoji emphasizes social attributes. It can generate a personalized emoji based on text descriptions and the user's personal image, which can be used in subsequent text conversations:

▲ Blogger creates a personalized emoji using Genmoji

The Image Wand feature allows users to sketch in notes, refine and add details to turn it into a complete, colorful image. Users can also constrain the generated results through text prompts:

▲ Blogger refines a sketch using Image Wand

Interestingly, if a person appears in the sketch or text prompts, Image Wand will automatically ignore it and generate only parts other than "people." The reason for this, perhaps for privacy protection or ethical considerations, or maybe due to incomplete functionality, remains unknown.

It can be noted that the images created by Image Creation in this update have shown significant improvement in quality and visual appeal compared to the example images that garnered criticism when Apple Intelligence was initially released. We look forward to seeing it improve even further upon official release.

Your intelligent "mechanical eyes"

Apple has developed a new feature called Visual Intelligence for the iPhone 16 series models that include the "camera control" button. This feature can help users identify items, get more information about them, and even provide channels for purchase:

▲ Blogger uses Visual Intelligence for object recognition

Users can also directly describe items in the frame using ChatGPT:

▲ Blogger describes a Sony memory card using ChatGPT

Furthermore, for objects like buildings, Visual Intelligence can provide their geographical locations; for places like restaurants, it can even provide information such as opening hours and menus to users:

▲ Blogger uses Visual Intelligence to identify locations in images

In our increasingly convenient daily travel and life, the ultimate goal of Visual Intelligence is to become the user's intelligent "mechanical eyes," allowing users to observe the world through Apple AI, further reducing the cost of obtaining relevant information, and improving convenience and efficiency.

A smarter Siri is here

Reviewing Brandon Butch's experience with the initial Siri supporting Apple Intelligence, we can conclude that its "intelligence needs improvement."

Compared to the entertainment attributes of Image Playground, the upgrade of Siri is more practical and "urgent."

The 2.0 version of Siri integrated with ChatGPT can directly provide answers to user queries based on the information provided by ChatGPT:

▲ Blogger asks Siri a question, and Siri answers based on ChatGPT

This essentially gives Siri access to a more intelligent "second brain" through ChatGPT, significantly enhancing Siri's "intelligence."

Of course, users can choose to call upon ChatGPT when asking questions, and even request Siri to "query ChatGPT and provide an answer" directly in the query:

▲ Blogger asks Siri to directly query the current time using ChatGPT

Now, even if a user's query is not smooth or the speech is stuttering, Siri can still understand and respond.

After much anticipation, the "handshake" between ChatGPT and Siri has finally happened. The future collaboration of these two major intelligent entities in the industry will bring even more surprises to users, so stay tuned.

Text and visuals, all within reach

The new Writing Tools, integrated with ChatGPT, may to some extent alleviate content creators' "creative anxiety."

In addition to using "Compose" to generate the required text with a single click, users can also modify and adjust the AI-generated text in a more flexible and free way. The most impressive feature of this writing tool is the ability to categorize a user's notes into several small tags, which are then transformed into an image.

The method of implementation is quite simple and elegant—simply draw a circle below the text where needed: