這次AI工具的升級，連工程師都傻眼‼️ 這已經超出控制範圍，連馬桶都AI化了‼️｜Google｜OpenAI｜Microsoft｜Claude｜Perplexity｜Adobe｜Kohler

Hello everyone, I'm Xiaofa, a senior student.

The changes in AI this year

have been so rapid it's almost breathtaking.

Browsers can now automatically move

videos generate

images from a single sentence, fix things in seconds,

and I'll even tell you

, toilets are now being equipped with AI!

In today's episode,

I'll summarize

some of the most representative recent AI developments.

But I'm not just reporting the news;

I'll also explain

how these things will affect our lives and work, and

what new opportunities they bring.

First, the AI ​​browser era has officially begun.

AI is no longer just for chatting

; it's starting to help you browse the internet.

In the past, we used Google

by entering keywords

, but now we

directly enter commands/tasks.

OpenAI's ChatGPT Atlas

is a new browser.

When you open it, the homepage isn't Google

, but ChatGPT.

You don't enter search terms,

but tasks,

like "Find the cheapest hotel in Tokyo

near the subway for

three nights for two people.

" Atlas doesn't throw a bunch of links at you

; it directly starts searching for you

. There's also a very interesting Agent mode,

which means AI... This

truly allows you to manually operate web pages.

It can filter

click sort and

copy information

, like having a digital assistant

performing your internet browsing actions.

This is a significant shift

because search behavior

is now being replaced by task-oriented methods.

For the average user,

there will be less and less need to compare prices

and find information manually

. However,

the clarity of your instructions

will determine the quality of the results.

In the future, only

those who ask questions will receive truly useful answers.

Another player is Perplexity,

whose Comet browser has

also officially launched for free.

It doesn't replace ChatGPT

but focuses more on information search.

It automatically reads the content of the web pages you are browsing,

provides summaries and

extended information

, and even automatically finds relevant links

for research. For those writing reports or preparing scripts,

this is a magical tool.

For example,

when you're researching a company's news,

it automatically summarizes the key points

and provides sources

to help you verify the authenticity

of the information, greatly increasing its credibility.

Browsers that help you organize information

can save information workers,

students, and creators a lot of time.

It's not just faster

; it makes it easier to understand the world.

Microsoft's approach

is a bit different.

They added a

feature called Copilot mode

to their Edge browser.

It remembers what websites you've visited,

like asking for help finding the blue jacket you saw last week,

and

it can immediately retrieve it for you.

This feature seems small, right?

But it's actually crucial

because it makes memory a part of the search process . In the

past, AI... It helps you answer questions

and now it's starting to remember who you are.

For creators or online shoppers,

this will extend to personalized content recommendations

in the future.

It knows what you like to watch,

what you care about

, and even what kind of subject you

're planning to film.

Next,

let's talk about AI video.

AI video has now entered the narrative era.

From the beginning of the year to now,

the progress of AI video has been frighteningly fast.

But the recent wave of updates

has several directions that

are worth noting

because it's no longer just about visuals,

but has begun to understand stories.

Google's Veo video generation has been upgraded again.

It can automatically generate the process in between

based on the start and end you set .

In other words,

AI is no longer just randomly moving things around

, but will help you fill in the story.

Another feature

is called image fusion.

You give it characters, scenes, and clothing,

and it will generate which person is wearing which clothes

and where.

This will make brand advertisements, product displays,

and even movie clips

more controllable.

Simply put,

Veo is now

an AI that understands storyboarding.

This step is really crucial

because it makes AI... The film

is closer to the logic of realistic shooting.

For short video creators, this means that

in the future,

they can automatically generate complete clips

simply by conceiving the scene . OpenAI's Sora

continues to evolve.

The new version can generate multiple scenes at once

, like storyboards,

and the video length has also increased.

More notably,

it has added character cameos,

allowing the same character

to appear in different videos.

Whether it's your cat, a doll,

or even a virtual IP, they

can maintain a consistent appearance.

Sora is actually

paving a new path. In

the future, short film production

won't necessarily rely on a camera.

As long as you have a story

, characters, and a voice,

AI can help you shoot videos that look real

. I succeeded! Listen carefully:

you can use Sora to generate brand new characters,

new monsters, heroes, or even more

. Or you can upload videos from your camera,

edit videos of your pets, and then turn them into cameos.

Even if your pet is a giant duck,

okay, let's go again.

We're excited to see how you use characters.

You've got me hooked!

Another noteworthy feature

is the open-source model LTX2,

which can generate 4K quality

50fps videos,

even with synchronized audio and video.

Most amazingly,

it can run on a regular computer.

What does this mean? It means

AI... With the barrier to entry for video production further lowered,

even companies outside of major corporations

can create professional-grade AI visual content.

This represents a true democratization of the content ecosystem

. YouTube recently

launched a facial recognition feature

, meaning that

if someone uses AI to mimic your face

or voice in a video,

you can directly appeal for its removal.

This might sound like a defensive mechanism

but

it also sends a signal:

AI-generated content is becoming increasingly human-like,

forcing platforms to establish rules

to protect creators like us.

In the future,

this will become a new issue for creators,

with licensing and protection

becoming as important as the creativity itself.

Speaking of AI video,

AI drawing, and image editing,

we've entered an era of integration.

Have you noticed

the trend in recent months ?

AI image editing is no longer confined to a single website or model

but has permeated everything.

For example, Microsoft

launched its own image model, MAI Image 1,

which can draw people and animals.

The generated text

simulates realistic lighting and shadow

effects,

and it's quite stable.

Currently, it can be used on the LM Arena platform.

Interestingly,

Microsoft, a major shareholder of OpenAI

, still launched

its own image model, essentially creating competition within a friendly environment.

This means the AI ​​image market will become increasingly diverse,

which is good for us users

because competition

leads to better image quality

and lower costs.

Google's Nano Banana is

now almost integrated into all Google products;

you can use it

in Photos, Lens, and even Notebook LM.

Just type out what you want to change,

and it will immediately fix it for you.

For example, change the background to a sunset or

remove passersby.

These image editing tasks can be completed with a single sentence.

This allows even those without design knowledge

to produce beautiful images.

I think

this means the technical barrier to image editing is disappearing.

In the future,

the only difference might

be aesthetics and creativity.

Adobe also held its annual MAX conference and

officially announced that Photoshop

Premiere, Illustrator, and Lightroom have all adopted AI.

Now, it's not just about using text commands to edit

images and generate visuals;

you can also freely switch between different models,

including Nano Banana, Flux One, and Topaz. Upscaler and similar

features mean that

professional creators

no longer need to wait for others to generate content.

AI is no longer just a source of inspiration

but has become part of the workflow.

Creating videos, designs, and posters

can be more than twice as fast as before

. And the good news is that

ordinary people can now also use AI to write code. In the past, when

people talked about AI programming,

their first reaction was

that it was an engineer's job

and had nothing to do with them.

But this update's

focus is no longer on increasing engineer efficiency

, but on anyone being able to command AI

to help you create digital tools.

Anthropic's new feature, Claude Code on the Web,

sounds very professional

but

its concept is actually very simple:

you can ask AI to help you modify

or create small tools

through chat.

Imagine having a super-powerful AI in your company. The assistant

not only helps you organize data and reply to emails, but

can also write simple programs

, such as creating a simple

function to automatically organize spreadsheets and send email reminders.

You just need to speak to it and

ask it to organize customer data

into a spreadsheet,

and

delete duplicates .

It will then handle it automatically in the background.

Moreover, it operates on a webpage,

so ordinary people can use it

without downloading software

or worrying about programming syntax.

Programming

is no longer just the domain of engineers.

In the future, anyone who can describe what functions they want

can create tools.

For example, I might think

, "Can I use it to organize video comments

, categorize repetitive questions,

list popular keywords,

and automatically compile a report for my reference?"

This is the direction Claude Code aims to achieve.

Google has also launched Vibe Coding in AI Studio,

which is even more impressive.

You can speak

and let AI create apps for you

—literally speaking!

For example, you can say, "I want a page

that allows people to upload photos

and generate comic-style AI versions

." It will automatically design the interface

and buttons for you.

Previously, you'd need engineers and designers to help

you with the model; now, you

can create it yourself as long as you have an idea.

It also has many built-in sample apps

like music players, chat rooms, and notebooks.

You can directly modify the text, colors

, and functions

, much like the concept of PowerPoint templates.

It even allows you to point to the screen and make changes,

such as circling a button and saying, "Change this to blue,"

and AI can immediately fix it for you.

But actually

, Google is realizing something:

in the future, building websites and apps

won't necessarily require programming skills.

You just need to clearly explain your needs,

and AI can

help you piece together the functions.

AI programming

isn't about replacing engineers

, but about enabling people with ideas but no technical skills

to implement them themselves.

This is similar to learning Photoshop and

other video editing software.

Initially, many people might not know how,

but those who learn

will have a significant advantage.

In the future, it

won't be about who can program,

but who dares to ask AI to realize their ideas.

And we must also mention the current state of AI... Agents

can truly help you with your work.

A clear trend in AI updates over the past few months

is that they're no longer just answering questions and generating images,

but starting to do tasks for you.

OpenAI's AgentKit function

is somewhat like a "self-driving workflow."

You write down the repetitive steps you take daily

—for example, someone fills out a form

, then you categorize it

, then you send a reply email, and

finally you update the form.

Previously, you had to use Zapier or other automation tools to

manually execute

this process;

now ChatGPT

does it for you.

AgentKit

allows you to turn ChatGPT into a true assistant.

It knows how to use external tools,

connect to databases,

and even call APIs for you.

In other words,

it doesn't just answer "what to do

," it does it itself. For

example,

suppose after each video shoot, you

first need to organize the names of the footage

, then upload the subtitles,

and finally send notifications to partner vendors.

All three tasks

can be handled by AgentKit;

it remembers the rules

and executes them in order.

AI is starting to truly understand processes,

not just chat .

For many small teams, work-from-home professionals

, and even students,

you finally have your own automated secretary.

AgentKit can help you get things done.

So, what about the ChatGPT apps? It

helps you connect to external plugins

, turning ChatGPT into an AI super platform.

Now, within ChatGPT, you

can directly use Canva, Spotify, Booking.com, and Figma

without opening new web pages.

You can say,

"Design a thumbnail for me,"

and it will open Canva and generate a template for you.

Or, "Check hotels in Osaka

for tomorrow and plan a three-day itinerary,"

and it will

grab data from Booking.com and

create an itinerary for you.

For the average user,

you won't necessarily need to open a bunch of websites anymore.

AI will handle everything for you in one place,

which is great news

for creators

. You can directly organize materials,

write scripts, schedule,

and design covers

within ChatGPT.

This is no longer just a chatbot

, but a work hub.

Anthropic has also launched a feature called Claude Skills,

which can be said to turn AI into your own clone.

It allows you to store your work habits,

reply styles, and

judgment logic

into a skill pack.

For example,

you can teach it how to write the opening of a video,

how to reply to a vendor,

and how to create a title.

After teaching it once, when

you ask Claude to write copy for you,

it will automatically apply your style. It

's like copying your brain into AI.

In the long run,

this will become a new personal asset

because your way of thinking and your

standard operating procedures for handling things

can be packaged up

and even shared or sold in the future. It's

not just convenient,

but symbolizes AI. It's learning to do your job,

but this time it's not taking your job

; it's giving you more time for creative, strategic, and

communication tasks—things more human-like.

The last one, which I find most outrageous

yet most realistic,

is the development of a smart toilet analyzer from

the American bathroom brand Kohler.

It's not a joke;

it can genuinely analyze your health

using sensors to observe your excrement

, assessing hydration, digestive health

, and even detecting bleeding.

Priced at approximately $599, it

's already available for pre-order.

Sounds absurd, right?

But this actually represents

AI permeating our lives in

the most inconspicuous places.

It's not just saving you time

; it's collecting your body data.

In the future, these everyday behaviors

may become part of your health data.

We often say AI is powerful,

but the real change isn't new models

or new technologies

; it's AI becoming part of our lives. It's like a background

browser, understanding what you're looking for,

automatically adding frames to videos,

fixing pictures with a single click

replying

to emails, recording information

, and even monitoring your health.

In the future, AI

won't be called AI anymore. It will become

an infrastructure like

electricity and the internet

. You won't particularly notice it,

but you'll use it every day.

For creators, office workers, and students,

this means the era of tool dividends is beginning

. Those who learn how to collaborate with AI earlier

will enter the next productivity stage sooner.

If you like this video,

remember to like, subscribe, and turn on notifications. Leave a

comment telling me

which new AI feature you most want to try.

I'll also share some practical tools and

test them

with you. See you in the next video!

Bye-bye,

guys! You should know that advertising agencies are finished.

With AI,

you can create studio-quality ads in seconds.