Transcripts
1. Deepseek AI: Has SHPT and EI
changed your life, if you have some key plesting? Let me tell you
something. You may be using AGBT and AI tools, and you may be loving them. But the fact is,
you're just using 10% of the AI's capabilities. The Markin is saying generated AI is gonna change the world, and so are some of the
huge media publications. But is it true? Well,
to some extent, it is. Generated AI is gonna change
world in a lot of ways. Do you want to create wonderful graphics and
images made from EI? Do you want EI to do all of your hard work you just to the smart workor
and sit and relax. These tools actually help
you and they save you money, thousands and tens and
thousands of dollars, $100,000. They save you time like
hours and s of your time, and they also make you work
smarter and not harder. Do you want to just give
one line prompts and that to easy ones and that
from scratch or even advanced proms where hundreds and thousands of
lines and create realistic unrealistic images and graphics and text from scratch? If yes, on this course,
for you, my friend. Then join in right now, and I'll see you in
the first lecture. Thank you for joining, and
I'll see you on the island.
2. DeepSeek AI: Welcome, everybody. This
was our Deep Sk dashboard looks pretty much like a
version of the famous chat GPD, but only this time, this is from the eastern part of the world and not to
western part of the world. Everything's pretty much
the same with the sidebar. You can bring it
up, close it down. This is my chat that
I'm going to show you, which is going to be important. This is you can get the app if you scan here on the phone. This is my profile. Same thing. You need an email,
password, that's all. Little backstory here. If you look at what I'm
zooming in right now, you can see this is deep Tink. Now what is deep Tink? This is the LLM model that deepsk uses for Chad CPD if
you know, for reference, they use the 3.5 model, and now probably they're using the Omni model and the 4.0 model and several other models which are used for some
are good for programming, some are good for
critical thinking, some are good for logical, some good for image generation,
so on and so forth. You can enable that and we can enable that,
we can disable that. This is very important. This is a search the web when necessary. If we need something very
urgent from the web, which is also in GPD as well, we can enable the web. Basically, all these AI models are connected directly
to the Internet. Every split second
of information, not every split second
of information, but probably pretty much
the latest information, not second by second obviously
can be found out here. This is a pin this is
actually a paper clipper, by the way, and I don't
know if you've used it, but these are really cool. Always sound them confusing
when I was in school. And here, you can attach
text extraction only, upload documents or images, maximum 50 to hundred MB each. Guess what? DPC is 100% free. They don't have any
planning options just like GBD in the early days. Remember, November
2022, early 2023. It's pretty cool. Pretty cool. Let's see how it works. We will go to our very
first chat that I did, and I want to show you
something very special which GPD didn't have up until
the last week of my. This is what I take. How are you doing? This
was my first chat. Look at this one. I thought
for 2 seconds as it says. Then this is how it's
thinking process went on. The user greeted me, Hey, how are you? I need to respond. Appropriately since I'm an EI, I don't have feelings, but I should acknowledge the
greetings and offer help. Maybe start with a friendly
reply, so on and so forth. Let me check for
typos and errors. I'm just a program so
I don't have feelings. I'm here and that's exactly
what it typed me here. Hey, I'm just a program,
I don't have feelings, but I'm here and ready to help
you. How can I assist you? This is an interesting
part because the EI models that
we've been used to, they don't show us
the thought process. What are they going
through? What is the critical thinking happening in terms of different
spaces and fields. But in this model, Deep Seek, it shows what it is thinking
right now, it's pretty good. Then I encountered
a bunch of problems like the server is
busy multiple times, so on and so forth, but
pretty much it did well. That was very much our
very first lecture. I'll see the next one with the very first application
with Deep Seek. I'll see you there.
3. Let's work on DeepSeek AI: Okay. All right, folks. We will start with
something very easy and simple like what can you do? Don't worry about
the spelling errors or typos, it would be fixed. Is smart enough to do that. What can you do for me? I think the easiest
question you can ask I model is what
can I do for you? In the following lectures, I'm going to even tell
you more about briefly, slightly where EI model Deep Seek and the models related to Deep
Seek is being used. Let's read its questions. Folks, this one is
really important. This one it's a bit
small in text and a bit Light in terms
of typography. That's how the EI is thinking. Not going to read that. You can read that for yourself
if you want to. Okay, I can assist you
with answer questions, history, tech, whatever,
problem solving, troubleshooting, creative work, learning, practical
writing, resistance, tech guidance, entertainment. Just let me know what you need. An I'll tailor my
help to your request. Cool. That's nice. Let's start with well, what date it is today? What date it is. So folks for your information, I'm recording this on
the fourth of Feb 2025. Let's see if it can
figure that out. I don't have access
to information, so I can't tell you
the current date. However, if you share general
location and today's date, I can help you calculate
past differences and explain, fine, no problem. We will start with well, let's start with
something very simple. Let's start with t I don't know. Let's start with
how do you define quantum computing quantum
computing to 8-year-old. In four lines of a poem. Let's make it a bit tough folks, starting with A
and ending with Z. I think it's a bit tough, but let's see if it
can somehow give us some relevant information
which we are looking after. Quantum computing is a
type. This one's cool. A tiny realm where magic spins, Cupid stands like curious twins. Here there,
everywhere they peek, Zooming answers,
strong yet meek. Each line starts with
the next letter A, QH N C. Let me know if
you'd like to sub tweets. Yeah, sure. Silver was a good attempt. Pretty good attempt. A
tiny realm of magic spins, cubit stands like curious twins. Here there everywhere they peek, Zooming answers,
strong yet meek. It's going to give me
a reiteration of that. I am waiting for
that and it's pretty cool because the first time
we used that a few days ago, it was not this quick. Now it is very quick. Wow, look at that. That is quick, guys. I don't know about you, but this is pretty cool. There you go. Adventures in quantum Zoo, cubits pin and shades of blue. Here, the magic everywhere, Zappy answers, fill the air. That's pretty good. That's
pretty good. It's pretty good. That was the very first of our prompts that we
did with Deep Sk. Next one is going to be
really important because we are going to focus on
something from the net. We are going to
give it a bunch of stuffs and we're going
to see how quickly it does and how it
reacts and responds. Till then, keep learning and I'll see you
in the next one. Thank you.
4. DeepSeek AI in details: Hey, everyone. For this lecture, we decided we're going to
focus on an article from WSJ Wall Street Journal and
the article is titled as Everyone's Title by Rise of DC except Nvidia, which enabled it. That is the EEO of Nvidia
Jensen Wang, great guy. I have copied the
link to the article, and I'm going to
paste it here and I'm going to ask you to
summarize all of it for us. Summarize this, please. I paste it and let's
see if it works or not. Because it's going
to be tough but I hope it will be
able to do that. Dude, this thing is, Wow. I can summarize that rule, but I can help you
share the key points or NDA's role in innovation, you specifics? Yeah, sure. It couldn't summarize all of it, but it did a pretty good job. Yeah, it can see
pretty good results. Whoa, whoa, whoa, whoa, whoa, whoa. All right. It says, India Jensen
Wang and EI in China. IVD CEO Jensen Wang emphasizes MVDA's critical
role in empowering global AI advancements through its high performance
chips, GPUs, US China tech engines,
Chinese innovations, global Air R. Wang acknowledges Chinese China's aggressive ambitions but argues innovation thrives under Innovation
thrives under constraints while in VDA navigates geopolitical hurdles
to retain market share. Good. Let me know if you
want to deep dive in any of these points or clarify
concepts. All right. It's putting on emoges like GPD. They've done such a great
job, deep sink guys. All right. So this
was a good one. Let's try one more time.
Let's search with the web. And let's disable
their own model. Let's search with the
web. What should I do? Well, market crashes today. So let's see if we can
somehow find the web, scrape what I'm asking for. Due to technical service
will be unavailable. I don't have real time data. You can check out da
data. No problem. Let me search for something
like I don't know. Equity markets US
This week this week. So this is Monday right now
it's not going to do that. It's somehow not
going to do that, this actually doesn't
work that well. Maybe I should write
one more time. I should copy this prompt. I should paste it again, and I should say
research carefully. Let's see if it can somehow
do the job it needs too. Wow, I think a good
prompt is going to come up because it's thinking too much
of the interesting. Wow. What do you watch in US equity markets this week,
earnings report. Companies releasing
quarterly reports. They will be releasing the
fourth quarter reports. Tech stocks, NASA tech react to EI register revenue upstates, inflation, job data,
consumer sentiment, Fed policy, geopolitical events, sector, it's interesting guys. It's interesting. I can do a pretty good job. I remember we were talking about Jensenghuang
a few minutes ago, Jenseng Huang was
a guy who said, everyone should have
an EI tutor now, which I humbly agree with. Everyone should have
an EI tutor. Why not? It's pretty interesting. Why not to learn
from these tutors. Yeah, it did a pretty okay job, I would say, or
maybe a good job. Let's see how it works
in the next one.
5. DeepSeek AI and other AI's: Hey, everyone. So I
was wondering if we should try to make some images with deep seek. Let's do that. Create an image of a bakery. I'll put the dimensions, which will be 1,800
by 1,800 pixels. I don't know if it'll
be able to create an image right now,
but I'm trying. I don't think so. I don't think so it's been
trained to do that. Yeah, it wouldn't do that. Rather it will write
me a good prompt. It's also giving me
ways to do that. Yeah, I cannot create
images as of now. No problem. We will try
to upload something. Let's see if we can
get something here. What is that? Here's an image that
I have upload failed. Let's write one more time. Upload, failed again. Just try one more
time. Upload failed. One more time. No problem. Is repeat. I'll try to upload this one. Upload, failed. So upload field after
field after fail. Couldn't do much
good here one more time before we call it the day. Okay. Let's try it. Give me a moment. Million upload failed. Folks, as I said, I tried a lot, but somehow it just didn't
work. It just didn't work. And so, you know, Deep Sik
is fine, the buzz is fine. All of those things are fine, but it's it's not close to what most people are
thinking or seeing. Yes, it's good. No doubt, prompts are good.
Language is good. The way it does its
critical thinking, analytical thinking is good, but not extraordinary to force Nvidia to shed 17%
of the market capital. You know what I'm saying? If you have no idea
what I'm saying, let me search it for
you. Let me search it. Nvidia crashes. And you will see, there you go. There you have it so many times. And D stock crash. There you go. How a Brooklyn based
blogger fueled the EI goint $600,000,000,000
market crash. That's a lot. It
crashed like 17, 18% that day. I remember that. And Yeah, that was
pretty much it. Well, this is a
blog. Well, I don't know. I had no idea. Okay. All right. Oh, it was also a kind of a Oh, so deep seqs bought into the
main frame by a journalist. Wow. That's interesting. I had no idea about that. Hmm. Here we go. The stock tanked, 12.5% actually opposite
was more than that. But probably it was introduced,
so it wouldn't count. Yeah, good, good, good, good. Yeah, so that was it. Wow. Pretty good, man. Pretty good. That is nice. Pretty good. Hm. Nice. So that was it, and I'll see you
in the next one.
Manas Roy, Professor
