ฝึกพูดภาษาอังกฤษด้วยเทคนิค Shadowing จากวิดีโอ: How I use ChatGPT as a Data Analyst

เลือกโหมดฝึก

แต่ละโหมดจะเปิดห้องฝึกแบบเต็มจอ

196 ประโยค

หากประโยคสั้นหรือยาวเกินไป กดที่ Edit เพื่อปรับแก้

Data nerds!

I've been using ChatGPT to analyze data for my job.

And it's been saving me a lot of time.

Let me show you what I mean.

So here we have a spreadsheet with some data we want to analyze.

First, I asked it to connect to the data by providing it with a link to the spreadsheet.

It gives me the first few reads of the data so I know it's connected.

So next, I ask it, what's this dataset even about?

And it looks like it's a bunch of job postings specifically for the role of data analyst.

It even provides some basic stats about the columns in the spreadsheet.

Let's do some quick EDA to see what are the 10 most common jobs.

And it provides me this bar graph with data analyst as the most common job title, followed by senior data analysts.

Let's wrap this analysis up and check out what the median salary is for these 10 most common jobs.

With this calculation, let's actually time it.

And in just over a minute, we're able to generate both these visualizations.

You can see that data scientists have some of the highest, while data analysts have some of the lowest.

I don't like this.

So I did this entire analysis before filming this without ChatGPT, and it took me almost an hour.

And here, in less than a few minutes, we've generated all of these different visualizations and insights without even coding one bit.

Now that we've shown ChatGPT's capabilities, let's move on to an even bigger problem that we're going to solve.

All right, so I stayed up a little bit too late last night.

How's my audio?

My audio good?

And I put together this PowerPoint of how we're going to solve this problem

and there's a lot of hype around AI lately

and whether it's going to actually automate and take away jobs and is it going to take over the world

probably but more importantly we want to focus on how is AI impacting data jobs

and we'll look at two main aspects first from the people's aspect from the person

that has the job how satisfied are they with AI impacting

their jobs the next one is businesses is there going to be an actual impact such as financial

or they're going to be actually be looking for those with skills of this AI

and how we're going to accomplish this all well here's the game plan

I have a two -step approach the first is using chat gpt for data research

we're going to be using different plugins

and internet searching methods in order to find research papers data whatever

it may be in order to find insights into those two main areas

that we're trying to look at from there we're going to move into data analysis

and we're going to once again use chat gpt to actually analyze this data

that we found in order to find insights and maybe prove one way

or another whether how bad ai is impacting data science jobs

but before we just dive into this analysis we need to understand the capabilities of this tool chat gbt

so um brb so what the heck is chat gpt well i asked it

and come to find out it's an artificial intelligence that's running on one of the models from OpenAI.

Specifically, this one right here, I'm using model GPT -4, which is their newest and greatest one.

And in a recent survey from Stack Overflow, over 50 ,000 respondents reported that they prefer ChatGPT over any other AI chatbot at the moment.

I've also tried all these other options and done a cold comparison video,

and I've found that ChatGPT is the best at performing data analysis at the moment.

But there is a catch with ChatGPT, and that comes to pricing.

So you can use the free version of ChatGPT and go in and prompt it and get different insights from it,

but you're really going to be limited in what you get.

Now, if you go with the pro version, not only do you have access to an improved model of GPT -4,

but also you have access to plugins.

And plugins are really the money.

Plugins allow these chatbots to connect to different applications.

So recently I've been using it with Wolfram in order to connect

and do analysis on the backend and then feed it back into ChatGPT so I can get my numbers crunched.

All right, enough with this green screen break.

Let's get back to the video.

All right, I made some progress on using ChatGPT to go

and search the internet to find things around AI impacting data science jobs.

And first up is one of the limitations that I found with ChatGPT,

and that's that the model itself was only trained on data up to September of 2021.

This has been a really big problem

because you can't actually ask the model directly about things that have happened right now because it won't know.

That's where plugins come in.

The first main one that I've been using to search the internet is their Bing plugin

that allows it to go in and search and find different articles, research papers, and publications, and I can find out what's most impacting.

Now, ChatGPT isn't really that great.

I also tested out

Bing and also the Google competitor, Bard, and they're not really search engines, right?

So I still rely on Google itself to go in and search, but I was able to find some recent articles on this,

specifically this one from McKinsey talking about the economic impact of generative AI.

Now this bad boy is almost a hundred pages long.

And if you think I'm going to go through and read it, you're probably wrong.

So what's good about ChatGPT is they have plugins in order to read these type of PDFs.

I just give it the link or the.. pdf itself

and it will go in and actually read it summarize it

and provide me with insights now generative ai has to deal with ais

that produce new content so basically what chat chpt is anyway it went through

and found this pretty interesting stat on generative ai and it showed

that it had really high potential for automating data related tasks specifically around processing and collecting data.

I mean, this is all sort of meta, right?

I'm using ChatGPT to analyze this PDF and provide data insights to me on data insights of generative AI.

So pretty crazy.

Another key source of information was this website right here from Stack Overflow,

where they actually surveyed a bunch of developers around 90 ,000 on what their sentiment was towards AI.

One of the key findings from this was

that 77 % of those surveyed had a favorable opinion towards using AI in their workflow.

That's four out of five.

That's pretty high considering how much negativity you were towards AI recently.

Now, the results of this report are pretty damn long.

And so once again, I can use ChatGBT to go in

and analyze it and actually asked it to summarize it better around data science professionals

and how they felt AI was impacting their job.

And ChatGPT was able to point out a pretty interesting insight.

Compared to developers, data analysts and scientists and engineers had some of the most positive sentiment towards this technology.

Now, the document itself has a lot of numbers around how these different job titles feel around AI.

So I asked it to analyze these numbers, provide a comparison.

But I ran into some issues where it started hallucinating

100

and making up a lot of the values that it saw in this document right here.

101

So I was really confused with this, especially since I had provided it with the source document.

102

I don't know why I was fabricating these numbers.

103

I couldn't figure that out.

104

But anyway, nonetheless, I was able to find the actual table that it was referring to.

105

And with another plugin, I was actually able to go in,

106

pull these numbers out, and then visualize it in order to see how it compared to other developers.

107

All right, we're back to the green screen.

108

So we discovered that data nerds have some of the most favorable opinions about using AI tools in their job.

109

And next slide.

110

We also, this also satisfies what we needed here in understanding what people's thoughts were towards this technology.

111

And overall, what we're seeing, four out of five prefer.

112

I mean, I'd say that that's a pretty good score.

113

And frankly, I think this is good.

114

But what about this side for the businesses?

115

What are their thoughts about it?

116

I mean, they're the ones

117

that are actually going to be driving this type of change within organizations and ensure people are adapting to this new technology.

118

The McKinsey study that I referenced at the beginning of my research

119

basically confirmed that these AI tools are going to increase our economic output.

120

So it's a win -win situation for companies to actually utilize this for their employees.

121

Yes, I do understand that there's different privacy concerns around what can be put into these type of tools,

122

but that's something that we need to start thinking about now.

123

So with the research we did with ChatGPT, we weren't able to find any data that related to how companies are implementing these AI tools.

124

But I have an idea.

125

So I've been collecting job postings since last year, and I have more details about it in this video right here.

126

Anyway, I've collected over a million jobs for data science job postings.

127

This should show indirectly how companies feel about implementing these AI tools, whether they're even asking new employees to have these skills.

128

Now, this data set includes things like job title, salary, and even job description.

129

So what we're going to do is we're going to use ChatGPT to perform some NLP

130

and go in and extract out any data related to AI being used in these jobs.

131

So let's get into it.

132

All right, it's been a few days and I've made a lot of progress using ChatGPT and specifically using the Notable plugin.

133

And not only is it providing visualizations through the front end, but you can also see the code behind how this all worked.

134

And that's pretty cool that you can double check it with this.

135

One quick note is that ChatGPT will be releasing the Code Interpreter plugin in the near future.

136

And this plugin will be as powerful or even more powerful than this Notable plugin.

137

So getting into the analysis, All of the data is located in a database that has a secure connection.

138

So I asked ChatGPT how do I go about giving these credentials so that way it can access this data.

139

And at first, it tried to give me code to perform this, but I re -prompted and told it to use the installed plugin.

140

And with that, it gave me a link that I could input in the credentials of the database.

141

So to make sure that it was connected to the database, I asked it how many entries it had in this table,

142

which it told me it was about a million, so I knew we were connected.

143

Next, I prompted ChatGPT to provide me some descriptive statistics about the data set.

144

And it not only provided information about what columns were in the data set, but it also showed some statistics around the salary information.

145

I'm not gonna really use that right now, but I thought it was interesting that I provided it nonetheless.

146

Next, I wanted to get into analyzing these job postings for keywords around AI.

147

So I prompted ChatGPT for a list of keywords related to this.

148

And although good, it's not really what I wanted, as these are more theoretical terms and not necessarily tools.

149

I've analyzed terms like AI and artificial intelligence before, and I didn't really think the insights from those keywords were really

150

that great instead i want to use tools

151

that relate to af for this we're going to be using those tools

152

that were included in

153

that stack overflow survey not only are we going to focus on search tools

154

but also on developer tools so i gave chat gpt this list of tools

155

and ask it to analyze whether these keywords were included in the job descriptions

156

this is where i ran into a limitation of this notable plugin

157

because this data set is over a million rows

158

and tried to extract those million rows of data and import it into a pandas data frame within the notebook.

159

I did find an option to upgrade my small ass computer, but it looked like it was behind a wait list.

160

So I reached out to Notable on Twitter and asked them if I could access these larger computers.

161

And they told me to wait.

162

So we're only limited around 100 ,000 jobs for this, and so we used a random sampling for this analysis.

163

And double checking all the results that I got at the end of this, it doesn't look like the results really change or fluctuate that much.

164

So I think we're fine with using this sampling.

165

So the first thing I jumped into was telling ChatGPT that each one of these rows was a job posting.

166

And I was curious to find out what percentage of these jobs contain one of those tools.

167

And less than 2 % of jobs is even requesting the skill of AI tools in the job description.

168

So we're not really off to a good start.

169

I thought next, let's actually look into what are the actual tools that are appearing in those job descriptions.

170

And I had to reprompt ChatGPT several times, getting it to sort highest to low, displaying the y -axis as a percentage,

171

and then formatting it to include one decimal place.

172

But eventually, we got this visualization, which shows that

173

Bing and u .com are some of the highest requested tools.

174

I am a little confused by these results, as they don't really correlate really well to that

175

Stack Overflow survey, but they are less than 1%, so maybe this does make sense.

176

So next, I was curious, with this low percentage,

177

are they even paying those with these type of skills a higher salary once again had to reprompt

178

it to correct column names and also format the y -axis as a currency

179

and we finally get this showing data scientists do get paid more

180

but data engineers machine learning engineers and data analysts don't

181

and such a low percentage of these jobs including these search terms

182

i'm not ready to call anything conclusive off of this visualization just think

183

that we maybe should explore this more in the future

184

now i did want to get into one last thing looking

185

at how these different job titles compared to each other for containing these AI tools.

186

And it looks like machine learning engineers are leading the way, followed by data analysts and data scientists.

187

So with these results, I think we've shown pretty conclusively

188

that companies aren't caring as much as people are about implementing AI tools in their jobs.

189

And I think this is a pretty big problem,

190

that there's such a big disparity between those that want to actually implement AI tools in their jobs,

191

and then the companies not really requesting it in these jobs.

192

Anyway, I digress.

193

So, if you got value out of this video and you like AI, smash that like button.

194

If you don't like AI, like those companies, hit that dislike button.

195

I'm curious to know.

196

Anyway, I'll see you in the next one.

ดาวน์โหลดแอป

ทุกอย่างที่ต้องใช้เพื่อพูดได้คล่อง

AI ให้คะแนนการออกเสียงให้คะแนนทุกประโยค

ฝึก IPAเชี่ยวชาญทุกเสียง

คำศัพท์สร้างคลังคำของคุณ

เกมคำศัพท์เล่นไปเรียนไป

สแกน QR

Download on theApp Store Get it onGoogle Play

FOR YOU

วิดีโอแนะนำ

73 ประโยค

คำศัพท์และข้อสังเกตด้านการพูดสำหรับบทเรียนนี้

บทเรียนฝึกพูดนี้ใช้วิดีโอ “How I use ChatGPT as a Data Analyst” เป็นสื่อ คำที่ถูกพูดซ้ำบ่อยที่สุด: Chatgpt, tool, plugin วิดีโอนี้มี 196 ประโยค และ 2646 คำ สำหรับฝึกพูดตาม ช่วงที่มีเสียงพูดยาว 13:39 ผู้พูดพูดเร็ว ประมาณ 194 คำต่อนาที จึงมีการเชื่อมเสียงและลดเสียงค่อนข้างมาก 82% ของคำอยู่ใน 3,000 คำที่ใช้บ่อยที่สุดในภาษาอังกฤษ ส่วนที่เหลือควรดูไว้ก่อนเริ่มฝึก

คำศัพท์สำคัญในวิดีโอนี้

คำที่ควรเรียนรู้ 15 คำจากวิดีโอ พร้อมคำอ่านและความหมาย:

คำศัพท์	คำอ่าน	ความหมาย
analyze คำกริยา	/ˈæn.əˌlaɪz/	วิเคราะห์
connect คำกริยา	/kəˈnɛkt/	เชื่อม, ต่อ
percentage คำนาม	/pə(ɹ)ˈsɛntɪd͡ʒ/	อัตราร้อยละ
salary คำนาม	/ˈsæləɹi/	เงินเดือน, ค่าจ้าง
scientist คำนาม	/ˈsaɪəntəst/	นักวิทยาศาสตร์
compare คำกริยา	/kəmˈpɛɚ/	เปรียบเทียบ
collect คำกริยา	/ˈkɑlɪkt/	เก็บ, สะสม
database คำนาม	/ˈdeɪtəˌbeɪs/	ฐานข้อมูล
confused คำคุณศัพท์	/kənˈfjuːzd/	สับสน
solve คำกริยา	/sɒlv/	แก้, แก้ไข
artificial คำคุณศัพท์	/ˌɑː.tɪˈfɪʃ.əl/	เทียม
axis คำนาม	/ˈæksɪs/	แกน
satisfy คำกริยา	/ˈsætɪsfaɪ/	ทำให้พอใจ
necessarily คำกริยาวิเศษณ์	/ˌnɛs.əˈsɛɹ.ə.li/	ต้อง
publication คำนาม	/ˌpʌblɪˈkeɪʃ(ə)n/	การตีพิมพ์

ประโยคที่ควรฝึกพูดซ้ำ

ประโยคสั้น ๆ ที่สมบูรณ์จากวิดีโอ ซึ่งนำไปใช้ในบทสนทนาประจำวันได้:

With this calculation, let's actually time it.
Let's get back to the video.
I don't know why I was fabricating these numbers.
I couldn't figure that out.
All right, we're back to the green screen.

ไวยากรณ์ในวิดีโอนี้

โครงสร้างที่ผู้พูดใช้บ่อยที่สุด พร้อมคำพูดจริงจากวิดีโอ:

โครงสร้าง	ในวิดีโอ
Present perfect continuous have/has been + -ing — การกระทำที่เริ่มก่อนหน้านี้และยังดำเนินอยู่	I've been using · it's been saving · I've been collecting
Present perfect have/has + กริยาช่อง 3 — เหตุการณ์ในอดีตที่ยังเกี่ยวข้องกับปัจจุบัน	we've generated · we've shown · I've also tried
Passive voice be + กริยาช่อง 3 — เน้นสิ่งที่เกิดขึ้น ไม่ใช่ผู้กระทำ	was only trained · can be put · being used

การออกเสียงที่ควรระวัง

ผู้พูดใช้รูปย่อและรูปลดเสียง 57 ครั้ง เช่น we're, I've, I'm ให้พูดแบบสั้นตามที่ได้ยิน

เสียง “th”: nonetheless /ˌnʌnðəˈlɛs/, theoretical /ˌθi.əˈɹɛt.ɪ.kəl/
เสียง “sh” และ “zh”: artificial /ˌɑː.tɪˈfɪʃ.əl/, limitation /lɪmɪˈteɪʃən/, publication /ˌpʌblɪˈkeɪʃ(ə)n/, smash /smæʃ/, accomplish /əˈkɑm.plɪʃ/
คำยาว — ลงเสียงหนักให้ถูกพยางค์: developer /dɪˈvɛləpɚ/, artificial /ˌɑː.tɪˈfɪʃ.əl/, capability /ˌkeɪ.pəˈbɪl.ɪ.ti/, limitation /lɪmɪˈteɪʃən/, favourable /ˈfeɪv(ə)ɹəbəl/

เสียงที่คนไทยมักออกเสียงยาก:

เสียงท้ายคำ — ออกเสียงให้ครบ ไม่เปลี่ยนเป็นตัวสะกดแบบไทย: analyze /ˈæn.əˌlaɪz/, percentage /pə(ɹ)ˈsɛntɪd͡ʒ/, skill /skɪl/, curious /ˈkjʊɹ.i.əs/, database /ˈdeɪtəˌbeɪs/
/v/ — ไม่ใช่ /w/ ให้ฟันบนแตะริมฝีปากล่าง: developer /dɪˈvɛləpɚ/, solve /sɒlv/, favourable /ˈfeɪv(ə)ɹəbəl/, privacy /ˈpɹaɪ.və.si/, dive /daɪv/
/z/ — เสียงก้อง ไม่ใช่ /s/: analyze /ˈæn.əˌlaɪz/, confused /kənˈfjuːzd/, columns /ˈkɑləmz/, utilize /ˈjuː.tɪ.laɪz/, visualization /ˌvɪʒ.ʊ.ə.laɪˈzeɪ.ʃən/

วิธีฝึกกับวิดีโอนี้

ฟังวิดีโอให้จบหนึ่งรอบโดยยังไม่ต้องพูด แล้วจดคำที่ยังไม่รู้จัก
เริ่มที่ความเร็ว 0.75× พูดตามทีละประโยค แล้วกลับไปใช้ความเร็วปกติเมื่อเริ่มคล่อง
อัดเสียงตัวเองแล้วเทียบกับต้นฉบับ โดยสังเกตคำอย่าง analyze, connect, percentage เป็นพิเศษ

เทคนิค Shadowing คืออะไร?

Shadowing เป็นเทคนิคการเรียนรู้ภาษาที่ได้รับการรับรองทางวิทยาศาสตร์ พัฒนาขึ้นสำหรับการฝึกนักแปลมืออาชีพ วิธีการนี้เรียบง่ายแต่ทรงพลัง: คุณฟังเสียงภาษาอังกฤษจากเจ้าของภาษาและพูดตามทันที — เหมือนเงาที่ตามผู้พูดด้วยช่วงเวลาห่าง 1-2 วินาที การวิจัยแสดงว่าเทคนิคนี้ปรับปรุงความแม่นยำในการออกเสียง ทำนองเสียง จังหวะ การเชื่อมเสียง การฟังเข้าใจ และความคล่องแคล่วในการพูดได้อย่างมีนัยสำคัญ

เทคนิค shadowing: อ่านคู่มือฉบับเต็มทีละขั้นตอน →

ฝึกพูดภาษาอังกฤษด้วยเทคนิค Shadowing จากวิดีโอ: How I use ChatGPT as a Data Analyst

เลือกโหมดฝึก

ดาวน์โหลดแอป

ChatGPT Tutorial | Chat GPT Explained | What is Chat GPT ? | Edureka

day in the life working as a business analyst | my business analysis workflow

How AI Is Changing the Working World

She Fell in Love With ChatGPT. Like, Actual Love. With Sex.

AI Made the Most Passive Side Hustle Even Easier

Try Not to Cringe: AI Psychosis

How Anthropic uses Claude in Product Management

Introducing GPT-6 Astra for developers

Can you learn English with AI?

Fastest way to become an AI Engineer in 2026

THIS ChatGPT tool can be your English coach

I Looked Into How Devs Are Actually Using AI