跟读练习: Can AI Replace UX Designers? Claude Ultracode Wins! - 通过视频学习英语口语

选择练习模式

每种模式都会打开全屏练习室。

151 句

如果句子过短或过长，请点击 Edit 进行调整。

So let's say you have a working app that solves a real problem, but users are struggling with it.

What do you do?

You either hire a UX designer or you try to fix it yourself with AI.

So you decide to give it a shot with AI.

Now where do you start?

You can't just tell AI to fix the UX and let it figure everything out, right?

Well, I just realized that with Claude, you can actually do that.

Just last week, I made a video where I shared my simplified professional UX process into prompts

that will get your AI to follow the correct methods to identify UX problems and to fix them.

And I specifically said that you can't just tell AI to fix your UX

and expect it to create a coherent user -friendly product.

Well, I was wrong.

The problem with my assumption is that I only tested it with Codex, and I assumed that Codex and Claude were somewhat at the same level,

since that's what the benchmarks are saying.

But that's not the case.

Claude just reached a level where I honestly didn't think would be possible this soon.

And after this video, I'm not going to be using Codex for coding and UI UX anymore.

My name is Jad.

I've been UI UX designer for over 10 years, but now everything is changing

if you want to stay updated on how ai is changing my work make sure

you turn on the notification bell and

if you need any help with ui ux

or coding you can submit your project on ai tooltip .com the example

i'm going to show you is a marketing platform

that i built in a recent video it's for my wife to manage her own marketing clients

and she's been adding features to it based on her usage requirements

so of course now it has a lot of usability issues and it's my turn to improve the UX.

I made two copies of the same platform.

I put one through my own AI UX workflow with the prompts

that I shared in my last video and I gave the other to Claude to fix all on its own.

Then I'll show you how Claude compares to Codex just for fun.

I think you know where this is going, but keep watching till the end because it all comes together.

And I'm going to tell you how I'm going to be approaching UX projects from now on.

Now, for my workflow, I gave AI multiple consecutive prompts to identify UX issues.

And it found 26 problems.

And then I fixed them by priority.

Now, this is not manual UX work.

I'm still using AI to identify the problems and to fix them.

But I'm guiding the AI based on my own process to identify user context issues,

to make sure the user journey is optimized and to fix consistency issues

and states issues and to make sure everything is considered.

Now for the other copy of the platform, I simply gave it to Claude and I said, you are a UX expert,

analyze and optimize the whole platform.

I put it on Opus 5 .5 in UltraCode, which is the maximum effort on the latest model.

It worked for a while on its own.

It ran multiple agent tests and many, many commands.

It found 40 problems and it fixed everything just with that one prompt.

Now I went through the two improved platforms and I compared them.

There's noticeable differences and I'm gonna say it up front.

I liked what Claude did on its own more than what my own workflow produced and that's why I'm making this video.

Now that doesn't mean we don't need you ex -designers anymore.

I'm just comparing AI with AI here.

Whether we need to know how to prompt it or if anyone can just say fix the UX.

Let me show you the results side by side and then I'll show you an analysis of how the two results compare.

On the left side is the result of my own workflow

and on the right side is what Claude did all on its own.

The first thing I noticed here is how Claude moved the client selector to the top bar

and made that bar sticky so it doesn't move when I scroll.

That means I always know which client I am working on.

This is a very good improvement that my workflow did not identify.

Here when you select a client and scroll you won't be able to identify which client you are working on.

But here it adds clarity.

Very much needed.

Claude also consolidated all the text in the top section of the page to have less debt space.

That's also very important because this is a high priority real estate.

What you see on your screen before scrolling should be the high priority stuff.

So more space for content is always better.

Another thing I immediately noticed here is how the navigation is organized.

Originally I had no groups but now both results are improving on that.

They grouped the menu items in different ways and based on my wife's usage she prefers what Claude did.

It makes more sense in terms of usability and user journey.

It's also visually easier to scan because it's more compact.

Even here at the bottom Claude reduced the amount of text and made the whole thing clickable.

As opposed to here where you have just a small target to click, it makes sense.

Another great improvement from Claude.

Now the dashboard itself also got improved in both cases, but Claude is more optimized towards what the user actually needs.

Here on the left side, there's a lot of wasted space just to display the status of the client,

then a long list of to -do items, then you get to the progress tracking.

But what Claude did here on the right side is impressive from my UX perspective.

This to -do list is smaller, more compact, and it grouped the important tasks at the top.

Here on the left, you have to look all the way down to see a task due today.

But Claude surfaced it to the top here.

And even the language is much more helpful.

Needs attention is better than tasks and this see all button is much better than all reminders.

All reminders does not tell me what action I am taking if I click it.

It's a little bit confusing because I don't know

if these are all my reminders or if I click it I see all my reminders.

Cloud also introduced something different here where you can see more items without leaving the page.

That way it can make the list smaller.

Small things like that improve the usability a lot.

Overall much better visibility and interaction in Claude's UX design.

And if you want a copy of this improved marketing platform let me know in a comment

and I'll be happy to share it.

All right let's go deeper and complete a user task.

Let's assess this content performance.

On the left side I'll click write assessment and here this is a new improvement it's telling me what I'm assessing.

It's actually showing me the result here, which is very important info.

It does not show me what I'm assessing here.

And it does not show me the previous result, which is important info to have.

100

So this one is better in my workflow, because I explicitly told it to analyze user intent and context and goals.

101

So for this feature, I get the win.

102

There's also a lot of improvements

103

that Claude did on its own in the campaign strategies page

104

it decided to move my strategies to the top instead of scrolling down to find them here

105

i can see my strategies then how they are doing my wife also preferred claude here

106

now i won't waste your time too much here analyzing

107

but let me know in the comments

108

if you want a more in -depth analysis for ux for complex platforms like this one i would be happy to share.

109

Here's how the two results compare.

110

This is based on my test result and the report that came out of each AI.

111

Now in terms of UI change, Claude did better by considering these areas that I showed you in my review.

112

But on a deeper level, my workflow actually improved things that Claude did not really catch,

113

like helping the user with relevant context for certain actions and failed save recovery.

114

These are things to avoid user frustration and improve user satisfaction.

115

So if you want max optimization, ask Claude to do its thing, and then follow up by using the prompts from my previous video.

116

And just for fun, let me show you what ChatGPT did with Codex on the latest and strongest model.

117

Here I have sold 6 .1 on UltraEffert, and I gave it the same prompt, and the result is not even close to what Claude did.

118

When I used Codex on the strongest model, I thought it would be close to Claude, but the result here is just not good at all.

119

Here, it did not consider any of the usability changes that Claude did.

120

It kept mostly everything the same as it was.

121

Nothing really changed with the main user flow and the context of the user.

122

Claude is much better.

123

So this tells me that we still need to guide the AI to find deeper usability issues,

124

but AI is only getting more powerful from this point.

125

Who knows, maybe with the next update things can become easy for AI to spot.

126

But what surprised me is that Claude didn't just clean up the interface, it made decisions about what the user needs to see,

127

what needs attention first, and where actions belong.

128

And in many of the flows that I showed you, I preferred Claude's version, And so did my wife, who is actually using the platform.

129

My workflow still helped in specific areas.

130

But what Claude did is real independent UX work from just one prompt.

131

And that changes how I would approach UX next time.

132

I would let Claude do the first pass, then test the important journey with real users who are actually using the product.

133

Where do they hesitate?

134

What information is missing?

135

Real user testing will never be eliminated by AI.

136

A week ago I said you couldn't just ask AI to fix your UX

137

and expect it to do the work we do as designers.

138

This experiment proved me wrong.

139

So I'm changing my opinion.

140

AI can now make good UX improvements without expert guidance.

141

It can still miss some things, but I think it produces good enough UX to launch a product.

142

And considering the rate at which AI is improving, I'm sorry to say this, but soon UI UX jobs won't be needed anymore.

143

We would still need UX research and strategy, yes.

144

We would also need to conduct real user testing and analysis.

145

That still takes human experience.

146

So the most important thing I want to leave you with is that everyone should be using AI.

147

Let me know what you think.

148

Do you agree?

149

Do you disagree?

150

I always answer everything down in the comments.

151

Cheers.

下载应用

流利口语所需的一切

AI 发音评分逐句打分

音标练习掌握每个音

词汇学习建立你的词库

词汇游戏边玩边学

扫码

Download on theApp Store Get it onGoogle Play

同频道

10 Years Of UX/UI Experience In One Claude Skill

Daily ConversationUniversity StudentsWorking Adults

The Simplest UI/UX Workflow With AI

Daily ConversationWorking Adults

FOR YOU

本课的词汇与口语要点

这节口语课以视频“Can AI Replace UX Designers? Claude Ultracode Wins!”为素材。视频中反复出现的词有：claude, user, result, fix, workflow。这段视频共有 151 个句子、1903 个单词可供跟读。讲话部分时长为 10:46。说话人语速自然，每分钟约 177 个词，接近日常对话。 83% 的单词属于英语最常用的 3,000 词，其余的词建议在练习前先学一下。

视频中的重点词汇

视频中 15 个值得学习的单词，附发音和释义：

单词	发音	释义
compare 动词	/kəmˈpɛɚ/	比, 比較 /比较
designer 名词	/dɪˈzaɪnɚ/	設計師 /设计师, 設計家 /设计家
reminder 名词	/ɹɪˈmaɪndə(ɹ)/	提醒物, 提醒
scroll 名词	/skɹoʊl/	條幅 /条幅, 卷軸 /卷轴
priority 名词	/pɹaɪˈɒɹɪti/	優先權 /优先权, 更重要的事
analyze 动词	/ˈæn.əˌlaɪz/	分析
usage 名词	/ˈjuːsɪd͡ʒ/	用法
compact 形容词	/kəmˈpækt/	緊湊的 /紧凑的
impressive 形容词	/ɪmˈpɹɛsɪv/	令人印象深刻
assessment 名词	/əˈsɛsmənt/	評價 /评价, 評估 /评估
cloud 名词	/ˈklaʊ̯d/	йүн, 雲 /云
select 动词	/sɪˈlɛkt/	選 /选, 挑選 /挑选
helpful 形容词	/ˈhɛlp.fl̩/	有幫助 /有帮助
solve 动词	/sɒlv/	解決 /解决
hire 动词	/ˈhaɪɹ/	租借, 出租

视频中出现的短语动词

单词	发音	释义
come together 动词	/ˈkʌm tʊˈɡɛðɚ/	集合, 联合
follow up 动词		跟進 /跟进
go through 动词		通過 /通过

视频中的语法

说话人最常用的结构，并附上视频中的原话：

结构	视频中的用法
被动语态 be + 过去分词 — 强调发生了什么，而不是谁做的	is optimized · is considered · is organized
定语从句 who / which + 从句 — 补充说明人或事物	UltraCode, which is · here, which is · result, which is
现在完成时 have/has + 过去分词 — 过去发生但与现在仍有关联的事	I've been · have sold

需要注意的发音

说话人用了 31 次缩略和弱读形式，例如 I'm, I'll, won't。请按听到的简短形式来说。

“sh” 和 “zh” 音: interaction /ˌɪn.təˈɹæk.ʃən/, satisfaction /sætɪsˈfækʃən/, assumption /əˈsʌm(p).ʃ(ə)n/, navigation /nævɪˈɡeɪʃən/, frustration /fɹʌsˈtɹeɪ.ʃən/
长单词——注意重音位置: priority /pɹaɪˈɒɹɪti/, experiment /ɪkˈspɛɹ.ə.mənt/, interaction /ˌɪn.təˈɹæk.ʃən/, consecutive /kənˈsɛkjʊtɪv/, satisfaction /sætɪsˈfækʃən/

中文母语者容易读错的音：

/v/ — 不要读成 /w/，上齿轻触下唇: impressive /ɪmˈpɹɛsɪv/, solve /sɒlv/, consecutive /kənˈsɛkjʊtɪv/, navigation /nævɪˈɡeɪʃən/, visibility /ˌvɪz.əˈbɪl.ɪ.ti/
词尾辅音 — 要读清楚，后面不要加元音: prompt /pɹɑmpt/, assess /əˈsɛs/, analyze /ˈæn.əˌlaɪz/, usage /ˈjuːsɪd͡ʒ/, compact /kəmˈpækt/
/r/ 和 /l/ 的区别: scroll /skɹoʊl/, clarity /ˈklæɹ.ɪ.ti/, selector /sɪˈlɛktə(ɹ)/

如何用这段视频练习

先完整听一遍视频，不要开口，记下不认识的单词。
先用 0.75 倍速逐句跟读，熟练之后再回到正常速度。
录下自己的声音并与原声对比，特别注意 compare, designer, reminder 这类单词。

什么是跟读法？

跟读法 (Shadowing) 是一种有科学依据的语言学习技巧，最初开发用于专业口译员的培训，并由多语言者Alexander Arguelles博士普及。这个方法简单而强大：您在听英语母语原声的同时立即大声重复——就像是一个延迟1-2秒紧跟说话者的影子。与被动听力或语法练习不同，跟读法强迫您的大脑和口腔肌肉同时处理并模仿真实的讲话模式。研究表明它能显着提高发音准确性，语调，节奏，连读，听力理解和口语流利度——使其成为雅思口语备考和真实英语交流最有效的方法之一。

影子跟读法: 阅读完整分步指南 →

跟读练习: Can AI Replace UX Designers? Claude Ultracode Wins! - 通过视频学习英语口语

选择练习模式

下载应用

同频道

10 Years Of UX/UI Experience In One Claude Skill

The Simplest UI/UX Workflow With AI

推荐视频

AI Destroyed My Profession

Is AI Killing Human Fashion Design? Who Will Shape the Future of Design? Fashion Tea #6

How to Build an AI Agent for Every Part of Your Life

How Anthropic uses Claude in Product Management

How AI Empowers Tech Leaders Siwei Shen

The UX Psychology Behind Apps People Can’t Stop Using

How to Edit VIRAL Talking Head Videos Using ONLY Claude! (New Minimal Style)

We Let $200 GPT-Pro Think for 7 Days on Unsolved Math

I Tested The Viral Reverse Engineering Repo

9 Insane Claude Mods I Already Can't live Without

Jack Dorsey’s New Dessa AI Voice Sounds Scary Human