YT YouTube2Text

YouTube2Text → Transcripts → Sachin Sharma "Building AI assisted education for every learner"

Sachin Sharma "Building AI assisted education for every learner" — Transcript

by Valerii Marenichenko · 8,554 words · 1,560 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:20Okay then we can start on an issue in
0:21that I think everything is fine now.
0:23>> Okay, okay, I will invite everyone from
0:26waiting room and we then we can start.
0:28First I start from introducing you if
0:31you don't mind, okay?
0:33>> Yeah, yeah, got it.
1:04>> Hello colleagues.
1:08>> Hello everyone.
1:10Hello dear participants. Now we can
1:12start.
1:14Today we have a start of our journey.
1:17So
1:19dear colleagues, participants, it's my
1:21great pleasure to welcome you. We will
1:23explore how artificial intelligence in
1:26transforming education, research,
1:28professional practice. And today I am
1:30delighted to introduce our speaker Mr.
1:33Sachin Sharma from New Delhi, India.
1:36Sachin Sharma is an experienced
1:38educator, educational innovator and
1:40founder of several initiatives including
1:44Good Heart Welfare Foundation and Vidya
1:47I hope we will we will explore how it's
1:50working today. So thank you for your
1:53help. His work focuses on making quality
1:56education accessible to everyone,
1:59particularly students from disadvantaged
2:01background.
2:02Uh through innovative educational
2:04approaches and digital technologies. So
2:07today he will share his experience in
2:09his presentation. So please dear Mr.
2:12Sharma, thank you for very much for
2:15joining us today and
2:17uh the floor is yours. Please. Thank
2:19you.
2:22>> Thank you Mr. Valeria. I'm really
2:23grateful that
2:24you have uh
2:26given me the opportunity.
2:28And welcome all.
2:30And uh
2:31today we are going to discuss about
2:34how building AI-assisted education for
2:37every learner.
2:39So before that I would like to tell you
2:40that during the past more than 1 and 1/2
2:43decade I have been into education sector
2:45and I have faced lot of challenges in
2:49developing my educational content,
2:52whether it is web applications and
2:55whether it is content.
3:01Uh could you please just mute
3:03everyone? Yeah, please.
3:09So uh
3:11I have faced lot of challenges in
3:13developing my educational content and
3:17web application.
3:27>> Yeah, now we can mute everyone. Yeah,
3:29thank you. Thank you. We can we can
3:31start.
3:31>> Thank you. Thank you. Thank you. Yes.
3:33Uh so uh the challenges which I have
3:36faced during the past more than 15
3:37years, how AI has helped me a lot in
3:41developing those content, whether it is
3:44for mathematics or English or anything.
3:47And also I have developed lot of web
3:49applications and Android
3:51Android application. Which is helping
3:55uh not only students, but also teachers
3:58and other non-profit organization.
4:01So let's move forward.
4:06So,
4:07I have done all these stuff like 15
4:09years in web, SEO, and content
4:11development. I have developed three web
4:13applications, Android apps, and AI tools
4:16on a daily basis. What I have developed
4:18is vidura.org and I have started my
4:21journey from Udgam Welfare
4:23Foundation.com. That is a non-profit
4:25organization which I have developed,
4:26which I have founded in the year 2012
4:30and where I started educating
4:33underprivileged children and helping
4:35them to fulfill their dreams to become
4:38engineer. And I have faced lot of
4:40challenges during those period, during
4:43those days. After that, during COVID,
4:47specifically during COVID, I was not
4:50able to
4:52join or connect with those
4:54parents as well as teachers from
4:57underprivileged society and I decided to
4:59shift uh towards digital learning. And
5:03there I developed vidura.org.
5:07From then, I started exploring
5:09artificial intelligence and how
5:11artificial intelligence helped me in
5:14developing my web applications and those
5:17tools which are helping now lot of
5:20students across the globe and also to
5:24other non-profit organizations.
5:29So, here we are going to discuss about
5:31uh my AI story, how to prompt, and your
5:34practical and live questions we are
5:36going to discuss later.
5:38So, before moving forward,
5:41can we have uh
5:42students
5:44to give these answers whether you have
5:47used LLM, that is large language model.
5:51Anyone you uh would you like to answer
5:54in the chat box?
5:57What all AI tools you have used?
6:00Is it available chatting or not?
6:06Yeah.
6:11Yeah, we have we have answers. Yes.
6:16I suppose I suppose most people use
6:19ChatGPT, of course, but
6:21many others use
6:23some other LLM
6:25models.
6:28>> Okay.
6:30Yeah, students have given answers that
6:32they have used mostly like ChatGPT and
6:35also
6:37Cloud.
6:38That's good.
6:39So, all of you must have aware about all
6:42these things, right? Gemini and all this
6:44stuff. So,
6:46what I did basically ChatGPT say Mistral
6:49AI Cloud and Gemini as well. So, I have
6:52worked on all these LLM models.
6:56That's good.
6:59>> [clears throat]
7:01>> Now, from paying for everything to
7:03building it myself. So, during the past
7:0615 days, here we are going to discuss
7:07about the practical aspect of artificial
7:09intelligence, how it's going to help you
7:11as well if you are planning to develop
7:14some education content for students or
7:16even for your college or even for your
7:18colleagues, right? So,
7:21in the beginning before AI, I have spent
7:24lakhs to your millions of
7:26rupees INR on SEO, that is search engine
7:29optimization. Also paid a developers for
7:32every changes.
7:34That means I have spent a lot of money.
7:46>> Yes, sorry. We can we can
7:49Yeah, yeah.
7:52>> So,
7:53I have a
7:55invested a lot of money in a web
7:57application development. I have paid a
7:59lot of money to web developers. And
8:03also paid content writers to write in
8:06LaTeX. So, how many of you know about
8:08later l a t e x?
8:11Would you like to give the answer here?
8:13How many students have you used later as
8:16well l a t e x?
8:23No one?
8:27No.
8:28Okay.
8:30So, later is a software you can say
8:33which is
8:34from
8:36it's an open source software.
8:38Uh where we use
8:40it to develop mathematical as well as
8:43scientific contents.
8:46So, I have been using later software for
8:48the past more than a decade now. Also
8:51slow dependent on others because I used
8:54to give task to others and they respond
8:57it very late. So, here we are discussing
9:00how AI has helped me a lot. So, what
9:04changes AI I did?
9:06See. Right now in the current phase, I
9:09do SEO of all the web application by
9:12myself. Website and apps built in days.
9:16So, what I did, I developed vidura.org.
9:20I would like to show this here.
9:27So, this is my web application
9:30which is developed by me by using AI
9:34chatbot. And other AI elements as well.
9:38So, here students are using it as well
9:42as parents and you can see teacher also
9:44using this platform. And I have spent a
9:48lot of time also my energy as well as my
9:51experience of the past 15 years in this
9:54particular
9:55web application.
9:57Also
9:58see. You can see that I have also
10:01introduced spoken English, Italian,
10:03French, Spanish, a lot of
10:05languages are there.
10:07And I have also developed these three
10:10Android application.
10:12This is word of Ved
10:15from where students can learn English
10:18also and this is Angora.
10:21Angora is basically a mathematical
10:24Android application which I have
10:26developed for every learner and it is
10:29free of cost available on Google Play
10:31all this stuff have been developed by me
10:34and you can understand that this type of
10:37complex web application cost you a lot.
10:41Also
10:42the developers used to take years to
10:45develop these type of content, these
10:48type of web applications.
10:51So this is my first web application and
10:54my old one which is mathstudy.in. If you
10:58see this is my another web application.
11:01I'm not
11:03advertising my product here but I would
11:05like to show you how AI
11:08is going to help you practically as
11:10well. If you have any plan to develop
11:13these kind of web application by
11:16yourself and you can use these AI tools
11:20effectively.
11:22So this is my mathematical ebook web
11:25website which is currently having more
11:28than 685
11:29ebooks available from grade 1 to grade
11:3312.
11:35And all these stuff has been developed
11:38by me during the past 15 years and you
11:41won't imagine that uh
11:44AI
11:45will help you a lot in saving not only
11:48money but it is also going to save a lot
11:52of time of yours.
11:55Now let's come
11:57to our
12:00presentation again.
12:02So here
12:03is my toolkit basically Cloud Port,
12:06ChatGPT, Visual AI and Deep Fake.
12:11So let's move and understand
12:15So, this is your question first. What is
12:17a prompt in AI?
12:19So, you must have used ChatGPT and you
12:22know Gemini and all that stuff. So, what
12:24is prompt here?
12:26Students.
12:29What is prompt?
12:32>> This is like a request of something.
12:40>> Okay.
12:43An instruction given to an AI. So, let's
12:46reveal the answer, yeah.
12:48So, this is basically the
12:53This is basically a instruction and
12:56instruction which you give to your
12:58model, whether it is ChatGPT, Gemini, or
13:01Claude, so that it is going to develop
13:04the content
13:06according to that particular prompt or
13:09according to that particular
13:11instruction.
13:15We are going to discuss this as well.
13:18So, what AI help me we have already
13:20discussed like Udgam Welfare Foundation
13:22we have, vidara.org,
13:24mathstudy.in, and free Android apps and
13:27latex study material and non-profit
13:29projects. So, I am also helping other
13:32non-profit organization where I am
13:35providing my services free of cost so
13:37that they can also help other
13:39underprivileged children free of cost in
13:42their domain.
13:43So, this is very effective.
13:51So, SEO. So, what is SEO? SEO helps a
13:53website to what?
13:55Anyone?
13:56What is SEO?
14:18>> So SEO is basically be found on search
14:21engines.
14:23Right?
14:24So it is basically search engine
14:27optimization where you develop your
14:31content as per the guidelines of the
14:34Google, right?
14:36So that the pages or your content will
14:40be easily searchable
14:42by the Google crawler.
14:45Am I clear here?
14:47So this presentation is going to help
14:49you practically as well as I discussed
14:51before.
14:52So
14:54my AI workflow idea to launch. So how
14:56you are going to
14:58launch a product or how to develop
15:03content by using AI.
15:06So this is the workflow basically.
15:08First, you have to define
15:10exactly what you would like to develop.
15:14This is very important. Rather writing
15:17unnecessary prompt or random prompt will
15:20not lead you anywhere. That is more
15:23important.
15:24And you have to define your task. You
15:28have to define what you are going to
15:31tell or what you are going to guide,
15:33what you are going to set
15:35to the LLM, right? After that, you have
15:39to create that prompt, write a clear
15:41instruction, and generate AI drafts,
15:44code, or content.
15:46Now
15:48you are
15:50a person who are developing
15:53code or who are developing a content. So
15:58you should not rely 100% on the AI. This
16:02is very important. So that's why I have
16:05mentioned this step. This is very
16:07crucial, review.
16:09So this is basically the step where
16:13human is involved.
16:16Whenever you are going to develop any
16:19content or any code by using AI, you
16:24must have seen that AI itself say that
16:28this is not This is possible that this
16:30is not 100% correct. So, you have to
16:32review it by yourself. That means AI
16:37require that you should
16:39itself review it so that the chances of
16:43error in that particular content should
16:46be reduced.
16:47So, this is step is very important and
16:50this state that we should not rely 100%
16:55on the AI.
17:00Am I clear?
17:04>> Yes, sure.
17:11Okay.
17:14>> After that, after reviewing, you have to
17:17publish or share with the learners.
17:20So, these all steps I have followed
17:24during the past more than
17:265 years from now
17:29since I have been using AI.
17:34Now, this is a live example. We have
17:37study material in LaTeX from one prompt
17:39we have developed. So, LaTeX typeset
17:42maths and science beautifully. So, LaTeX
17:44is a software which is available free of
17:46cost and this is an open-source software
17:48from Linux where we develop content
17:52from
17:53the Linux based software for mathematics
17:56as well as science. So, a lot of
17:58universities are using LaTeX code to
18:02develop their own content. So, here we
18:04have
18:05It used to be paid specialist skill.
18:08Now, prompt gives a full draft and I
18:10check every answer before publishing.
18:12So, here is the prompt. Write three
18:14practice questions from quadratic
18:16equations for class 10 in LaTeX with
18:20step-by-step solution. So, this is the
18:23format of the later which I have coded
18:25by myself manually.
18:28And I even now I have been using it
18:32through [clears throat] AI to just
18:33rectify the errors if there are some
18:36errors in the code.
18:39So, this is the format basically begin
18:41enumerate and enumerate. This is the
18:43numbering. So, there are three questions
18:45in later code which this prompt is going
18:48to develop. There are chances that other
18:51model may develop some other content or
18:54some other questions. So, this is the
18:55basic format what prompt is going to
18:59develop for you.
19:04So, what is later basically? Later we
19:06mostly use to what? So, what is your
19:08answer here?
19:11So, students, what is your answer here?
19:21Good.
19:23So, people are giving answer here. That
19:24is good.
19:25Yes.
19:27Anyone else?
19:32Good.
19:37So, people are understanding what I'm
19:39teaching actually. This is This is
19:41This is really good
19:43that you are understanding what I'm
19:45teaching here.
19:47Great.
19:48So, later is basically used to typeset
19:51math and science.
19:55Next one. So, what is a prompt
19:58engineering? This is very important.
20:00So, to develop any content or to develop
20:03any web application, you should have a
20:06fair knowledge of prompt engineering.
20:10So, what is prompt? We have already
20:12discussed that it is a set of
20:13instruction which you are going to
20:15provide to the LLM to develop your
20:18content or to develop the web
20:20application. Now, let's understand what
20:23is prompt engineering to understand it
20:26in a better way.
20:27So, first is the skill. The skill of
20:29writing clear instruction
20:32so an AI gives you the result you
20:36actually need. Like briefing a very
20:38fast, very literal assistant.
20:42Uh sorry.
20:43The first point is be clear. Say exactly
20:46what you want. Topic, level, and
20:49quantity.
20:51So, rather writing a random prompt is
20:55not going to give you the proper result
20:58what you are expecting from any LLM. So,
21:02don't expect that you write a small
21:04sentence to any LLM and it is going to
21:08provide you the result what you are
21:10expecting from LLM, right? So, you have
21:13to make things clear to develop
21:16the result or to get the result what you
21:19are expecting.
21:20Next one is give context. Tell the AI
21:24who it is, who the audience is, and why.
21:28So, tell the AI who it is.
21:31Who the audience is. That means, if you
21:33are
21:34planning to develop a content for any
21:37school, for any curriculum, and for any
21:40particular student. So, you have to make
21:42it clear
21:44that you are going to develop the
21:46content for this particular specific
21:49curriculum. Whether it is for India,
21:52whether it is for Ukraine, whether it is
21:53for Sweden, whether it is for US,
21:55anywhere else, right? And also you have
21:57to mention
21:59the class, the standard for which you
22:02are developing
22:04that content. Very simple.
22:07After that, iterate. Read the result,
22:09then improve the prompt, and try again.
22:12So, this is also very important that
22:15there are chances that you have written
22:17a well-elaborated
22:20prompt, but you are not getting the
22:23result what you are expecting.
22:25So, you have to read that result
22:29basically so that
22:32you are
22:34in the next prompt, you are going to
22:36improve your prompt. See where the
22:39prompt is giving you the wrong answer or
22:44where you are wanting the detail
22:49in which step or at what steps. This is
22:52very important. And so, this is not
22:54possible. There are chances that you may
22:56get the results
22:58not as per your expectation. So, in that
23:01case, you have to read the result first
23:04and modify your prompt accordingly.
23:12Prompting, the five-part prompt formula.
23:15Number one,
23:16rule. So, you have to define the prompt
23:20or the LLM, what is its rule?
23:23Means, what responsibility you are going
23:26to provide it. This is very important.
23:29So, act as a math teacher, let us take
23:31an example. Context, what is a context?
23:33For 15-year-old students in Ukraine, let
23:36us say.
23:37Task, what is the task basically? Create
23:39five practice questions on fractions.
23:42Format, what is the format? A table with
23:45question, answer, and hint.
23:50Next one is constraint.
23:52Simple English under 100 words.
23:54So, this is the breakup of a
23:57well-formatted prompt with the help of
24:00which you are going to get a better
24:03result.
24:05Right?
24:15So, weak prompt versus strong prompt,
24:17let us understand this to make it more
24:20clear to you.
24:21Right?
24:23What is weak prompt? Make a website. So,
24:25this is what? This is simply a sentence,
24:28you can say or group of words you can
24:31say. Nothing clear, nothing
24:34>> [clears throat]
24:36>> static or this is vague result.
24:37Generally decide many rounds of fixing.
24:39What order website would you like to
24:41make?
24:42How many pages do you require? What is
24:44the profile for which you are developing
24:47that particular website? Right? So,
24:49every question every question should be
24:52clear.
24:53Act as a web developer here see strong.
24:55Act as a web developer, what is the
24:57responsibility of web developer? Build a
25:00one-page website for a village library
25:03in Ukraine with about books and contact
25:05section, mobile-friendly, simple HTML
25:08and CSS and explain each file. So, this
25:11is basically a clear usable first draft
25:14you can improve.
25:15Okay? So, once you provide this type of
25:19prompt, which is more clearer than this
25:22type of vague prompt. Right? So, after
25:26reading the result of this particular
25:30prompt, you can keep improving whatever
25:34you require, wherever you require more
25:37pages,
25:38more
25:40interesting results or more elaborative
25:42content on that particular website. So,
25:46this part is very important.
25:50Next one.
25:52One topic, three prompts.
25:55Right? So, we are going to understand
25:57how prompt give you different results
26:01when you are defining prompts in a
26:03better way. So, here we have a basic
26:06prompt, then we have improved another
26:09prompt, then we have written an expert
26:12type of prompt.
26:14So, first prompt is basically give me a
26:16question on quadratic equation. Right?
26:19So, it is stated give me a question on
26:21quadratic equation. Next one, we have an
26:24advanced mode. Act as a class 10 maths
26:27teacher.
26:29Create one medium-level practice
26:31question on quadratic equation. This is
26:34more refined prompt.
26:36Another one is
26:37act as a CBSE. This is basically a
26:40central board of secondary education
26:41which we are following in India. Uh you
26:44may use some other curriculum board as
26:46well. Class 10 teacher, one medium
26:49question, hint, step-by-step solution,
26:52use simple English.
26:54So, this prompt is more like an expert.
26:58So, let's see what sort of result it is
27:01going to give you.
27:05Give me a question on quadratic
27:06equation. So, what is missing here? No
27:08class level or board, no difficulty
27:10level, no hint or solution, no format or
27:13language control is there.
27:15So, this type of generic result you are
27:18going to expect from such a prompt.
27:23Solve this. As simple as that. There is
27:26no level, no hint, no solution, nothing.
27:28So, this is a
27:29very vague prompt, you can say.
27:31And
27:32it may be useful to some student, but
27:36not everyone. If you are a mathematics
27:37teacher, let us say for grade 10 or for
27:40grade nine, something like that. And you
27:43are requiring a specific type of
27:45question on quadratic equation, then
27:47this type of questions are of
27:49weak quality.
27:51Now, let's go forward.
27:53This is our second prompt, which is a
27:56more better prompt.
27:58Act as a class 10 math teacher.
28:01This is your responsibility. Create one
28:03practice question on quadratic equation
28:05at medium level.
28:07So, if you are using
28:09if you are currently using a laptop or
28:12something, you can simply use
28:15any LLM and you can also
28:19be practicing
28:21at parallel to this lecture, right? So,
28:23that you will get a more better idea how
28:27to enhance prompt.
28:29Right?
28:32So, this is our second prompt.
28:35What is improved here? Right level for
28:37class 10. Difficulty you asked for.
28:41Exactly one question. So, this is the
28:44result. Solve 2x squared this. Good
28:46level, but no hint and no solution to
28:48check against that.
28:50Now, let's move towards our third
28:52prompt.
28:54So, our third prompt is more elaborative
28:57and written like an expert.
29:00So, here we have
29:02Act as an experienced class 10 math
29:04teacher. My students are 15 and learn
29:07English as a second language.
29:10Create one medium level question on a
29:11quadratic equation solvable by
29:13factorization. Show difficulty,
29:16question, hint, step-by-step solution,
29:17and final answer. Use simple English.
29:20Keep each solution under 60 words.
29:23So, you can see that how prompt has
29:26improved your result here. See, video
29:29factorization like question is this.
29:31Hint, split the middle term. So, first
29:34it has given the hint to the student so
29:36that the student can refer this and try
29:39to solve the solution. Now, 2x squared
29:42this is the method obviously elaborative
29:44method is there.
29:46So, you can simply understand by
29:50improving your prompt
29:52how you are going to enhance the quality
29:55of the output what you are expecting
29:58from prompt.
30:02Am I clear here?
30:08>> Yeah, thank you. Everything clear.
30:10>> Yes.
30:12Oops.
30:16Now, let's compare same AI, same topic,
30:19very different results. So, this I have
30:21prepared so that uh you can refer it and
30:25understand
30:27uh which prompt you are going to use and
30:30which not. Here, right level for class
30:3210, prompt number one. Nothing is there,
30:34right? Prompt number two, we have right
30:36level for class 10 difficulty you asked
30:38for is there and method and language
30:41control is there but very less.
30:44Prompt number three which we have used
30:46recently, it covers all the factors here
30:50which we have discussed hints and full
30:51solution already to use this students
30:54and all those stuff. So,
30:57this is your comparison for all the
31:00three prompts.
31:05So, this is
31:07basically a prompt engineering toolkit
31:09which I have prepared for you. Six ways
31:11to level up any prompt.
31:15Give a role.
31:17First one is give a role
31:20to your prompt to the LLM. Name the
31:23audience.
31:24What is the name? Basically, age level,
31:26language of your student. This is very
31:28important.
31:29Set numbers. How many questions? What is
31:32your requirement?
31:34Next, define the format.
31:37This is very important. When you are
31:39going to provide the format,
31:42right? I'm going you give I'm going to
31:44give you some example here.
31:46Whenever I ask the LLM to give me the
31:50result in the form of let us say in
31:53later code, right? So, it is going to
31:56use a lot of tokens. So, everyone is
32:00aware about the tokens?
32:02Anyone?
32:05Tokens, what is token in LLM?
32:13You have used ChatGPT, Gemini.
32:16Do you know about tokens? Anyone?
32:23No.
32:25Okay.
32:29So, token is a term basically used
32:34Someone Okay. No.
32:37Not an issue.
32:39So, if you
32:41if you have used
32:42any LLM a paid version,
32:45I'm talking about the paid version, or
32:48even in the free version,
32:50when you are using any LLM, whether it
32:53is ChatGPT, means Gemini, or Cloud,
32:56anything, right?
32:58So, in the back end,
33:00it has fixed the
33:02the total amount of tokens you are going
33:06to use but uh
33:09to get your result or to get your
33:10content. For example,
33:13>> [clears throat]
33:14>> if
33:16you would like to generate an essay, let
33:18us say, an English essay of let us say
33:213,000 words. Let us take an example here
33:25of 3,000 words.
33:31Okay.
33:32So, someone had I think uh
33:35answered Try to answer that the
33:37definition of a token.
33:40It is also an AI-generated answer, I
33:42believe.
33:44So,
33:46this is good.
33:48So, uh
33:49token is basically used
33:52uh by the LLM at the back end
33:55uh to calculate
33:57So, what is the usage of the user?
34:00This is basically a simple definition
34:03for your understanding.
34:05Okay? So, uh in developing a 3,000 uh
34:09word of essay, let us say,
34:13in free module, in free module, you can
34:16use only
34:24in free module,
34:26but
34:27that is not going to provide you that
34:29much of tokens
34:32at the time of processing that
34:33particular essay. So, what it is going
34:36to uh give you, it is going to give you
34:38only 1,500 words or 1,800 words of
34:41essay, and after that it is going to
34:43stop.
34:44Right?
34:45So, that's how at the back end
34:48any LLM works.
34:51Similarly, if you are going to get a
34:53paid version of any LLM, it is going to
34:55enhance your token. It is going to
34:57enhance that particular usage of any
35:01user because you have done some payment
35:03for that.
35:04Okay? So, here we have tokens.
35:07Set numbers, how many questions, how
35:09long, define the format. We are talking
35:12about defining the format. So, when I
35:14was using a format, say later, right?
35:17I used to give him
35:19develop 10 questions in the later code.
35:21So, it is going to consume more code.
35:24So, for you who are using the AI LLMs,
35:29you can generate your result in JSON.
35:32JavaScript files, right? Rather asking
35:35him to generate any other code, you can
35:37ask him to generate the result in the
35:40JSON format and after that you can
35:42convert that into your required uh code.
35:47That is going to save a lot of tokens of
35:49yours.
35:50Add constraint. So, what is the
35:52constraint basically? Simple English,
35:53word limits, what to avoid, right? If
35:56you
36:03Yes, uh someone has asked me that uh
36:06whether we uh can use
36:09AI to develop a prompt for ourselves.
36:13That is a good question and that is also
36:15very useful.
36:17Basically, what you can do, you can ask
36:20any LLM first
36:22that you have to create a prompt for
36:26yourself, right?
36:28So, that prompt you can use further to
36:31develop your content. That means first
36:34step what you have used, you can ask any
36:38LLM to develop a
36:40uh best prompt, you can say, or to
36:43develop a good prompt, so that by using
36:47that particular prompt, you are going to
36:49get a result which is more important, or
36:53you can say which is more useful for
36:55you, rather than updating a prompt again
36:59and again. So, this is a good question,
37:02and I appreciate your question here, and
37:05others can also use this type of
37:09technique as well, while updating your
37:11content or while developing your
37:12content. Good.
37:14Now, iterate. So, after analyzing or
37:17after following all these steps, right?
37:20Five steps you have followed, and after
37:22that you can iterate or ask them to
37:25process the
37:28prompt.
37:31>> [clears throat]
37:32>> Next,
37:33which prompt gives the best result here?
37:36So, here you have four options.
37:39You can give me the answer.
37:42Here you have four options. Give me the
37:44answer for this. Which one will be the
37:46best prompt to get your result?
38:06Good. People are giving answers, and uh
38:09that's interesting.
38:11That you are getting uh
38:13the things
38:15what I'm trying to convey you here.
38:17Good.
38:19Good. Option A, which is role plus
38:21context plus task plus format. So, this
38:24is a well-defined prompt, you can say,
38:27to get the desired result of for your
38:30content or for your application,
38:32whatever you would like to have. So,
38:35here we have role plus context plus
38:38text.
38:41Next one is
38:43use AI responsibly. So, this is very
38:46very important.
38:49So, you don't need to use the AI
38:53randomly. There are a lot of loopholes
38:55when AI develops some content or when
38:58when you are developing a web
39:00application by using
39:03AI. So, if you have certain amount of
39:07experience for your domain
39:10like me, I have been developing website
39:13web application for more than a decade
39:14now.
39:16So, I know what are the pros and cons
39:19when you are getting a code, random
39:21codes from an AI, and you have to
39:25implement those or remove those
39:28loopholes before publishing your website
39:30because there are a lot of threats.
39:32There are a lot of security
39:34vulnerabilities
39:36while using such prompts. You must have
39:39seen lot of hacking nowadays where AI is
39:44using those loopholes or security
39:47concern on different different web
39:49application, and people are misusing AI
39:53by exploiting
39:55those loopholes on any web portal. So,
39:58that is very important. Verify the
40:00facts. AI can sound sure and still be
40:02wrong.
40:03Protect privacy. Never paste passwords
40:06or personal data.
40:07You should avoid using
40:10password or
40:12give the full
40:13full access to your AI model. This is
40:16very important. Don't give your admin
40:18password and all those stuff to any AI.
40:22Be honest. See where AI helped and keep
40:25your own thinking. Keep humans in the
40:27loop. So, this part is very important.
40:29Teachers and reviewers, check the
40:31quality. If you are developing content
40:33from the AI, then teachers and reviewers
40:37should be there. Don't just copy and
40:39paste all the content
40:42to any user to any student before
40:46reviewing the content here.
40:49So, I think I'm very clear that this
40:51part is very important. So, keep humans
40:53in the loop.
40:54While developing any content from the
40:56AI, you have to be there. You have to
41:00audit thoroughly any content, any web
41:03application, any code so that any flow,
41:07any vulnerability should not be there.
41:09Any error, any error you should have to
41:11remove well before time, well before
41:13publishing, right?
41:18Next one, AI proposes, you decide. Human
41:21defines the goal.
41:23AI drafts the first first version. So,
41:26here it is going to help you by saving
41:29your time, by saving your money a lot.
41:31This step, that means you're going to
41:33define the task. It is going to give you
41:35a first draft. This is your first draft,
41:37not the final one. After that,
41:41sorry. Human reviews that.
41:45Just a second.
42:05So, after that, human reviews it.
42:07So,
42:08after this particular step, you can come
42:11again here. That means human define
42:13again or human reviews. Then again, you
42:16have to define what errors are there,
42:19what is your requirement further so that
42:22you can enhance this prompt or you can
42:24enhance your web application or you can
42:26enhance your
42:27content here.
42:31Am I clear to everyone?
42:35Everything is clear, right? Thank you.
42:39Okay.
42:40After that, reviews,
42:42you can approve that particular content.
42:47>> Can you send it to us, yeah?
42:48>> Yeah.
42:50Uh could you please just
42:52other users?
42:54Thank you so much.
42:58Okay.
43:00Thank you.
43:01Let's move
43:04further.
43:05Who is responsible for AI? This is a
43:07question for you. Who is responsible for
43:08AI made content that you publish?
43:12Who is responsible for AI made content
43:16that you publish?
43:29Good, you have 2 3 seconds more.
43:32Anyone else?
43:33It's simple, obviously.
43:38Okay, great. Thank you so much who have
43:40given the answer here. Option number D.
43:43Obviously, you can't blame at the AI
43:45company. You can't blame the AI tool
43:47itself, nobody else and you, the human
43:49who publish it.
43:54The platform where we publish that
43:56content.
43:58Right.
44:00Okay.
44:02So, there are two things actually. Uh,
44:04someone has written the platform where
44:06you publish the content.
44:08So,
44:10the platform
44:13who is getting the content from you. Let
44:15us say you are
44:17a blog writer, for example. You are a
44:19content writer for me.
44:22So, you have given
44:24the content
44:26for me, for my web application. So,
44:28after that,
44:30that is my responsibility and not yours.
44:33Right?
44:34So, that is the responsibility for me,
44:37for publisher
44:39who has to audit that particular content
44:42thoroughly before publishing. So, that
44:45is the right answer.
44:47Uh, good one.
44:48So, this is the responsibility of the
44:50publisher
44:51to analyze the content thoroughly.
44:55What
44:56the
44:58uh
44:59content creator has sent it to you.
45:03Simple.
45:06Next one.
45:09So,
45:11avoid dependency. A co-pilot, not an
45:14autopilot.
45:16So, anyone who like to answer this, what
45:18is the difference between a co-pilot and
45:19an autopilot?
45:21Please.
45:22In the comment section, what is the
45:24difference between a co-pilot and not an
45:26autopilot?
45:53Anyone?
45:55Co-pilot and autopilot.
46:17Exactly.
46:19Uh so, few have answered me. That is
46:21great.
46:22So, co-pilot is
46:25something where human is in the loop.
46:29Right? Where human is in the loop. And
46:32if you say autopilot, that means where
46:35human is not in the loop.
46:39Right?
46:41So, this is important. So, you have to
46:44be a co-pilot and not an autopilot,
46:46depending on the AI. Copy and paste
46:48without reading.
46:50Skip learning the basics. Cannot tell
46:52the right from the wrong. Stuck when AI
46:54is unavailable. Using AI wisely,
46:58understand first, then ask AI. Use AI to
47:02explain and practice.
47:04So, this part is very important. Second
47:07and third one, use AI to explain and
47:09practice and check every answer.
47:12So, check every answer yourself. So,
47:16here what we have been discussing, that
47:19human should be there in the loop. That
47:22is called
47:24co-pilot. Your skills grow with every
47:28task. So, this is very important. If you
47:30are using an AI,
47:33right?
47:34But, avoid the full dependency on AI.
47:38Try to involve yourself
47:41at each and every step.
47:44Due to which
47:57Someone is asking about
47:59some specification about the Indian
48:01government that uh
48:02whether they are having any restriction
48:05to use AI.
48:07Uh so, can you just explain means in
48:09which sector would you like to ask
48:11whether it is you know education domain
48:13or whether it is for financial domain or
48:15something like that.
48:16So, this is very important. So, this is
48:19not a generic question, basically. You
48:21have to define.
48:23And in my view, in my opinion, I don't
48:26think that in the education domain or uh
48:29any other domain, and specifically in
48:32the financial domain, it may be there.
48:36But, I'm not confirming that.
48:41So, your skills grow with every task.
48:44So, this is very important. Rather
48:46focusing, rather you know, depending
48:48fully on any AI, you have to get
48:51yourself involved so that
48:53you are going to enhance your skill. You
48:56are going to enhance your knowledge
48:58related to your specific domain. So,
49:01during the past 5 years and I have been
49:04developing, you know, web application. I
49:05have developed a lot of web application,
49:08but those were not fully automated. On
49:11every step, I involved myself.
49:15Whether it is content, whether it is web
49:16application, anything.
49:18Right? After that, I have developed so
49:20much of complex web application as well
49:22as Android application and I learned a
49:24lot from that.
49:27Next one.
49:29Proofreading is a must. So, this part is
49:31very important, right? Proofreading.
49:33Proofreading is what? Once you have
49:36developed a content, once you have
49:38developed any web application,
49:42proofreading of that particular content
49:45web that particular code is very, very
49:48important and very, very crucial here.
49:51Because after proofreading only, you are
49:54going to understand what is wrong here
49:57and where you would like to enhance your
50:00code. Where you are going to and how you
50:03are going to enhance your content as
50:05well. This is very important because on
50:07a random basis means, you know, once you
50:09generated any prompt, and it is going to
50:11give you any sort of the result. Without
50:14proofreading, you are not going to
50:17ask any LLM
50:19that
50:20the result you got is perfect, right?
50:23You can't simply copy-paste the result
50:26of any prompt
50:28without proofreading. It is not going to
50:30help you any help anyone, right? Whether
50:33you are a teacher, whether you are a
50:34student, or whether you are a parent, or
50:36anything, right? So, this is very
50:38important. Proofreading is very, very
50:40crucial everywhere. So, check the mess.
50:42Resolve the question. If you talk about
50:43the mathematics, check the facts and the
50:45sources. Confirm with the textbook or
50:46trusted site. Check the level, right
50:49language and difficulty for your
50:50learners, right? What language it has
50:52used? Let us say you are asking for to
50:55write in English and it has introduced
50:57some, you know, a machine language,
50:59something like that or some other
51:00language it has introduced suddenly.
51:03This is very important. Check spelling
51:05and styles. Check for bias. So,
51:08when you are developing
51:11content
51:12when you are developing a content or
51:14whether you are developing a web page,
51:16so it is very important that you ask
51:20for an audit
51:22from the AI itself, right?
51:25It is going to save a lot of time. Once
51:28you ask any LLM
51:30that audit this particular page
51:32thoroughly, it is going to show you a
51:35lot of bugs as well.
51:36Lot of errors there, right?
51:38I myself have faced lot of challenges. I
51:41myself have faced lot of, you know,
51:43errors while developing any content or
51:45while developing any web application.
51:47So, we have to, you know, give
51:50challenges to our model
51:52that you have to audit it thoroughly so
51:54that no bugs will be there.
51:57This is very, very important.
52:02Check for bias. Check for privacy.
52:07>> [clears throat]
52:11>> An AI solved this. Is the answer
52:13correct? So, anyone
52:16if you are from a mathematical
52:17background or you must have studied
52:19mathematics in grade eight or grade
52:26Could you please just mute yourself?
52:27Thank you so much.
52:30So, here another question for you.
52:33An AI solved this. Is the answer
52:35correct?
52:39Anyone?
52:52Anyone from mathematics background? This
52:54is basically a simple question from
52:56grade eight or grade 10 or something
52:57like that.
52:58Can you see the answer here? Is it
53:01correct?
53:02The AI has developed the answers.
53:05Anyone?
53:14No, it's for what? Nobody is there from
53:17mathematical background? Or you are
53:19asking me that
53:21the answer is
53:22correct or not?
53:25Okay, good.
53:27X equals to 2 and X equal to 3.
53:31No.
53:36Okay, so this is basically a very
53:39simple, very small example where AI has
53:43developed a response for a question.
53:47Right?
53:48That it has
53:49written here X minus 2 and X minus 3 is
53:52equal to 0, so X equal to minus 2 and
53:54minus 3. So, this is basically a
53:56domain-specific
53:58question, right?
54:02So, this is very important. If you are
54:05If you are not from a mathematical
54:07background, you won't be able to
54:10find the error, number one.
54:13So, simply asking any AI
54:17to develop some content which is
54:20not from your domain,
54:22then it is going to give you wrong
54:25answers as well.
54:27So, for that, you have to hire some
54:30auditor, you have to hire some domain
54:33expert.
54:34Right?
54:38>> [clears throat]
55:31>> Now, why AI cannot replace the teachers?
55:35So, this question is a debatable
55:38nowadays
55:40that AI is going to replace the teacher.
55:42Why AI can't replace the teacher?
55:48Anyone would like to answer? Critical
55:50thinking, very good.
55:51Great.
55:55Anyone else?
56:09Number one is a teacher notices when a
56:11student is struggling or upset.
56:15This is very important.
56:17From the learner's point of view as
56:19well. Motivation, teachers inspire
56:21learners to believe in themselves.
56:25Right? So, these are the few points.
56:28Local context, they know the class
56:30culture, curriculum, and community.
56:34Judgment and value, they decide what is
56:36fair, right, and age-appropriate.
56:38Mentoring, guidance for life, career,
56:40not only for answers, right? When
56:43students are dealing with the AI
56:45modules,
56:46it is simply
56:48giving you answers.
56:52Get the point? So, this is very
56:53important. Guidance for life and career,
56:56not only for answers. Accountability, a
56:58teacher is responsible for learning. AI
57:01is not.
57:09Teacher plus AI a stronger learning.
57:16So, when we introduce so even I have
57:20introduced the AI
57:22So, currently I'm helping thousands of
57:25students across not only in India but
57:27maybe globally.
57:29And also helping other non-profit
57:31organization.
57:32So, this is very important by using AI
57:37we can help lot of students.
57:46Right, great.
57:49So, uh
57:50teacher plus AI a stronger learning,
57:52drafting practice questions, right?
57:55Making notes and later quickly. So, here
57:59we can take advantage from the AI like
58:01drafting practice questions. We need
58:03question let us say
58:05uh any teacher require 15 questions for
58:07his or her student. So, uh previously it
58:11was a daunting task for any teacher even
58:14for me. If I would like to you know uh
58:16give some assignment to my students of
58:1930 questions, then I used to type in the
58:23later and uh it has wasted a lot of time
58:26of mine. So, here what a teacher can do
58:29basically uh drafting a practice
58:31question. For this you have to provide a
58:33good prompt to an AI so that it is going
58:36to give you a specific type of result, a
58:38specific type of question for that
58:40particular grade, for that particular
58:42students. Making notes and later quickly
58:45and explaining a topic another way,
58:47translating and simplifying. So, it is
58:49going to give you a good and quick
58:52result.
58:53When you are developing content for your
58:55students, when you are developing
58:56content for other school or colleges,
58:59right? So, here we can take an advantage
59:01from the AI. And what teacher being
59:03understanding of each learner. So, this
59:06is very important. Let us say there are
59:10three types of learner.
59:12Right?
59:13A learner, a student who is very weak in
59:16concept. So, I have also worked on this.
59:19A learner who is very weak, a learner
59:21who is an average, a learner who is
59:23intelligent, right? So, I have developed
59:28a module where questions will be shown
59:32to students as per their level of
59:35understanding.
59:37So, that is also a very interesting
59:39module.
59:40Even if you are a teacher or even if you
59:43are an AI developer or content
59:45developer, you can also analyze these
59:49type of uh
59:51you can analyze these type of you know
59:53understanding mental status you can say
59:55or
59:56uh understanding level of understanding
59:58of students and give them different
1:00:00different tasks.
1:00:02Right?
1:00:04So, that
1:00:06every learner will get questions
1:00:09according to their level of
1:00:11understanding. This is going to help
1:00:13every learner in enhancing their skills.
1:00:17Right? You can't give
1:00:20a student who is a you know uh very
1:00:23intelligent
1:00:24and uh you are giving uh questions to
1:00:27him which are very you know mediocre
1:00:28type or very simple questions. So, he's
1:00:30going to lose the interest in that.
1:00:33Inspiration and encouragement is there
1:00:36whenever
1:00:37uh you are a teacher.
1:00:39You must have taught in schools or
1:00:41colleges, right? Then there are students
1:00:44there must be students who are not in uh
1:00:47education but they are good at other
1:00:49fields. So, you can motivate them.
1:00:52This is very important. Judgment on
1:00:53quality and truth, trust, care, and
1:00:56accountability.
1:00:58So, these are the few points what
1:01:01teacher brings in the classroom, in the
1:01:03students, or in the community and here
1:01:07AI
1:01:08help teachers
1:01:10in
1:01:11getting the results quickly, in getting
1:01:13the solution quickly
1:01:15and uh
1:01:17quickly
1:01:19done few task which obviously when
1:01:22teacher manually is doing that take a
1:01:24lot of time and consume lot of time,
1:01:28right?
1:01:30Now my another question is there AI
1:01:32gives a confident
1:01:34but false answer. This is called what?
1:01:38Anyone?
1:02:13So you people are doing answers, yeah,
1:02:15great.
1:02:20Good.
1:02:26So this is called hallucination.
1:02:29Debugging is something catching is
1:02:31different
1:02:32aspect of a browser. Basically we are
1:02:35using debugging is nothing but finding
1:02:37the error in the code and compiling is
1:02:40also
1:02:41basically compiling the different
1:02:43different code and giving you the
1:02:45output. But hallucination is something
1:02:47that AI gives a confident but false
1:02:49answer. Where you know, you provide a
1:02:52prompt to the AI and it is going to give
1:02:55you a false answer.
1:02:57And if you are not a domain expert then
1:03:00there are chances that you may use that
1:03:03particular answer as your final answer,
1:03:06right? So that is called hallucination.
1:03:14This one.
1:03:22>> I'm sorry.
1:03:25>> Uh
1:03:27Would you please just mute others? Thank
1:03:29you so much.
1:03:30Mr. Aladi, thanks.
1:03:35Uh so, uh this is next one, best
1:03:38practice with AI-made study material.
1:03:41What is your answer?
1:04:00Good. The time is up.
1:04:02Thank you so much who have participated
1:04:03in this particular chat, and answer is
1:04:05B, basically, review and verify. We have
1:04:08been
1:04:09focusing on this particular step when
1:04:12you are using any prompt and getting
1:04:16your result from any LLM.
1:04:20Great.
1:04:21Right?
1:04:22So, review first and verify it.
1:04:25Whatever result you are getting.
1:04:27Okay? So, this step is very, very
1:04:31crucial.
1:04:32Next one.
1:04:37From India for the world. AI-assisted
1:04:39education for every learner. When one
1:04:41person can build with AI, create
1:04:43learning. What we have done, basically,
1:04:45study material for nonprofits, web
1:04:47application at lower cost, learning
1:04:49anywhere on any device.
1:04:52So, what we have prepared, or what even
1:04:55you have, you can prepare with the help
1:04:58of this AI.
1:05:00You can not only help, you know,
1:05:02yourself, but you can also help other
1:05:05students across the globe by developing
1:05:08AI tools, by developing web
1:05:10applications. What I have developed in
1:05:13the past, and also currently developing
1:05:17few of them.
1:05:23>> So, what are the key takeaways? A clear
1:05:25prompt is a skill you can practice.
1:05:27Iterate, draft, review, and improve.
1:05:32Okay?
1:05:34Verify before you trust or publish. Use
1:05:37AI to help every learner.
1:05:40So, these are the key takeaways.
1:05:43Any question about these things?
1:05:46About prompt or
1:05:49what you have learned or any question in
1:05:50this?
1:06:12Thank you so much for appreciating me.
1:06:15I'm really grateful that you are
1:06:17enjoying
1:06:19these
1:06:20slides which I have prepared
1:06:23for specifically for this particular
1:06:25program.
1:06:27And
1:06:29in my opinion, this uh
1:06:33will help you in developing few you know
1:06:36your content or even
1:06:39uh you can develop web application as
1:06:41well.
1:06:47Now,
1:06:49so Valery, what you are asking that
1:06:50about this particular workshop join the
1:06:52code. So, this is quite interesting.
1:06:55What you can do,
1:06:57you can log in on my you know web portal
1:06:59which is free of cost obviously
1:07:01vidura.org. With this particular code,
1:07:03you have to get yourself registered. And
1:07:07once you register, you will see
1:07:10uh
1:07:11your task basically. Workshop join code
1:07:13workshop task is there. And what you
1:07:16have to do, you have to develop
1:07:19some content or you know some website
1:07:22page.
1:07:23There are different different tasks
1:07:24basically. I think 10 to 12 tasks are
1:07:26there. You have to choose any task and
1:07:30submit that particular task with the
1:07:32help of your AI tool. And after
1:07:35reviewing your, you know, answers, I'm
1:07:38going to give you the result
1:07:40and grading after, you know, 24 hours or
1:07:43maybe after 48 hours. And it is going to
1:07:47show you. I'm going to show you on your
1:07:49dashboard itself. Once you submit, you
1:07:52are going to get your result
1:07:53on your dashboard.
1:07:56I think you this code is clear, Mr.
1:07:58Valeri.
1:08:01>> Yes. Yes. Thank you.
1:08:05>> So, these are the steps which you have
1:08:06to follow while registering on your
1:08:09on my web application new students with
1:08:11a lot of energy and after that it is
1:08:12going to ask you this particular code
1:08:15and your assignment appear on your
1:08:18dashboard.
1:08:35Everyone has written this code?
1:08:38Everyone? You can answer in the chat.
1:08:41Everyone got this code?
1:08:46Thank you so much
1:08:47for your appreciation.
1:08:51Uh you can answer me in the chat that uh
1:08:54have you written this code and obviously
1:08:57uh you must be recording this lecture so
1:08:59that you can share this particular code
1:09:01later on with the students.
1:09:13So, thank you so much. And if you have
1:09:15any question, then you can ask me here.
1:09:18I will be more interested in giving your
1:09:20answers.
1:09:21>> Thank you very much, Mr. Sharma. This is
1:09:24really interesting and important
1:09:27content for us Uh because you're
1:09:30you're um
1:09:32you consider it in many many questions
1:09:34in one topic, but it's it was
1:09:37really clearly for us.
1:09:39So, thank you and your combined
1:09:44like developing tools and educational
1:09:47tools. So, it's
1:09:49it's most important for us because we
1:09:51are educator and now we can
1:09:55try to do something new to create
1:09:58new content, new
1:10:00some approaches. And maybe we have some
1:10:03questions in chat, of course.
1:10:06>> Yes.
1:10:08Thank you for your appreciation and
1:10:10people are, you know, clapping for me. I
1:10:13really appreciate that.
1:10:15I really appreciate that you enjoyed the
1:10:17session a lot. And I prepared the
1:10:20session
1:10:21from the practical aspect of all the
1:10:23users so that it may be useful to
1:10:26everyone. And whosoever, you know, in
1:10:28this particular domain, in this
1:10:29education domain, will be getting
1:10:32benefited from this. Right? It is very
1:10:34important how to write a prompt properly
1:10:37to interact with any LLM to get your
1:10:39desired result.
1:10:45>> Yeah, thank you. And I hope, as you
1:10:48said, that AI never replace teachers.
1:10:52Thank you for for, you know, this all
1:10:55this
1:10:56issues. You're
1:10:58you find this concepts. And thank you.
1:11:01Thank you for for everything.
1:11:06>> Okay, thank you so much and I appreciate
1:11:08that you have invited me once again. And
1:11:12I'm really thankful
1:11:14to all the, you know, participants here.
1:11:17You have enjoyed and you have, you know,
1:11:20attained some knowledge from my side.
1:11:24Whatever I have, you know, uh during the
1:11:27past more than 1 and 1/2 decade. It's
1:11:29almost 15 years now.
1:11:31And uh during the past more than 5
1:11:33years, I have been using AI tools. And I
1:11:35think I have achieved a lot of, you
1:11:37know, things from AI. And I have
1:11:40developed more complex, you know, web
1:11:42application and Android application by
1:11:44myself.
1:11:47Which people, you know, are charging
1:11:48millions of rupees or millions of, you
1:11:51know, dollars in developing those, you
1:11:53know, free tools. And these tools are
1:11:55basically free of cost and any student
1:11:57across the globe and even other
1:12:00non-profit organizations are also, you
1:12:02know,
1:12:04uh getting help from me.
1:12:09>> Yeah, thank you. Thank Thank you for all
1:12:11this content. And we will keep in mind
1:12:14all these ideas. And we will try. And
1:12:17then you can
1:12:19you know, you just started this our
1:12:21journey. And thank you for for your
1:12:23efforts. And
1:12:25for your experience, because you really
1:12:28experienced
1:12:29researcher and developer. So, thank you
1:12:32so much for
1:12:34for everything you've done for us.
1:12:38>> You're welcome. You're welcome.
1:12:40If you have any question, any student,
1:12:41you can ask me any question. You can
1:12:43also mail me anytime, right? I am
1:12:45available for your help. And if any
1:12:47non-profit organization is there, I am
1:12:49ready to help to them as well. Thank you
1:12:51so much.
1:13:03Uh my email, you can just explore my,
1:13:05you know,
1:13:06web application. You can write an email
1:13:09to me. It is going to you know,
1:13:12uh
1:13:15take you to my mailbox.
1:13:18Right?
1:13:21>> Yes, thank you. We will explore your
1:13:23platform and and try to to to use your
1:13:27platform
1:13:28of course for one day with this code. Uh
1:13:30so
1:13:31I think everyone will will find
1:13:35really helpful for for them and try this
1:13:38platform. And maybe another platforms
1:13:41you
1:13:42you
1:13:43you created before. So thank you so much
1:13:45and I think if we don't have any
1:13:48questions for now, maybe we will have in
1:13:51the future and we will keep in touch.
1:13:55We will communicate with you in the
1:13:56future and thank you so much for for
1:13:59your time.
1:14:01>> It's my pleasure. Thanks a lot. Thank
1:14:03you so much everyone every participant
1:14:04Mr. Valeriy specifically you. Thanks a
1:14:06lot. Bye. Goodbye. Take care.
1:14:08>> Yes, thank you. Thank you. Goodbye.

About this transcript

This page contains the full transcript of Sachin Sharma "Building AI assisted education for every learner" by Valerii Marenichenko, generated from the public captions YouTube serves with the video. The transcript has 8,554 words across 1,560 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.