Will Prompt Engineering Replace Coding?
Summary
TLDRThe Nvidia GTC conference highlighted the emergence of prompt engineering as a new coding paradigm, showcased by the Nvidia inference micros service. This development emphasizes the increasing abstraction in software development, from assembly code to high-level languages, and now AI-driven solutions. The video discusses the evolution of programming and the importance of understanding underlying engineering principles, critical thinking, and problem-solving in the tech industry, predicting a future where life sciences will transition from discovery to engineering.
Takeaways
- 🚀 **Prompt Engineering's Emergence**: Nvidia's CEO, Jensen Hong, introduced the concept of prompt engineering as the new coding at the GTC conference, emphasizing the minimal code required to build AI applications through the Nvidia Inference Micros service.
- 🧬 **Healthcare Application**: An example given in the transcript illustrates how prompt engineering can be used in healthcare to analyze cancer genetics and imaging data, potentially leading to the development of targeted drugs in minutes.
- 📈 **Benchmarking with AI**: The transcript introduces 'Devon', an AI software engineer, which benchmarks the performance of an AI model on different API providers, showcasing the capabilities of prompt engineering in real-time.
- 🌐 **The Evolution of Coding**: A historical overview of programming languages from the 1940s to the present is provided, highlighting the trend of increasing abstraction and simplification in coding practices.
- 📚 **Educational Relevance**: The importance of understanding low-level programming languages in education is emphasized, as it provides a foundation for grasping engineering principles and problem-solving skills.
- 🛠️ **Software Engineering Principles**: The distinction between coding and engineering is made clear, with the latter involving critical thinking, problem-solving, and the application of engineering principles to build complex software systems.
- 🤖 **AI and Engineering**: Jensen Hong suggests that the future will see the application of engineering principles to fields like life sciences and drug discovery, moving from scientific discovery to engineering solutions.
- 📈 **The Role of AI in Various Fields**: The transcript discusses the potential for AI to impact not just technology but also other sectors like healthcare, climate change, and life sciences, through the application of data analytics and engineering principles.
- 💡 **The Importance of Understanding Outputs**: It is highlighted that understanding the desired output and the engineering principles behind achieving it is crucial, even as coding becomes more abstracted and accessible.
- 🔍 **Research and Development**: The transcript touches on the importance of R&D in both technology and life sciences, suggesting that the future may see more integration between these fields through engineering principles.
- 🎯 **Focus on Productivity and Problem-Solving**: The ultimate goal of prompt engineering and AI tools is to enhance productivity and problem-solving capabilities, as exemplified by the use of AI like Devon for benchmarking and analysis tasks.
Q & A
What is the main idea Jensen presented at the Nvidia GTC conference?
-Jensen introduced the concept that prompt engineering is the new coding, emphasizing the efficiency and minimal coding required to build and deploy AI applications through the use of Nvidia's inference micros service or NIM.
What is prompt engineering, and how does it differ from traditional coding?
-Prompt engineering is the process of guiding generative AI to produce desired outputs with minimal coding. It differs from traditional coding, which involves creating detailed instructions for computers to follow, by abstracting much of the complexity and allowing users to achieve goals with fewer lines of code.
How did the field of programming evolve from the 1940s to the 2020s?
-Programming evolved from complex, low-level languages like assembly code to high-level, abstracted languages and frameworks like Python, Ruby, and TensorFlow. This evolution made coding simpler, faster, and more accessible, with a continuous trend towards higher-level abstractions and easier development processes.
What is the significance of Margaret Hamilton's contribution to software engineering?
-Margaret Hamilton was a pioneer in software engineering, playing a crucial role in the Apollo moon missions and coining the term 'software engineering.' Her work emphasized the importance of rigorous engineering principles in creating complex software systems.
What is Jensen's perspective on the future of computer science and life sciences?
-Jensen envisions a future where life sciences will transition from a field of scientific discovery to one of engineering, with digital biology and drug discovery becoming part of life engineering. He encourages learning about human biology and engineering principles to contribute to this emerging field.
How does the script relate the historical development of programming languages to the emergence of prompt engineering?
-The script draws a parallel between the historical simplification of programming languages and the emergence of prompt engineering, suggesting that prompt engineering is a continuation of the trend towards more accessible and efficient computation processes.
What is the role of prompt engineering in the development of AI products and tools?
-Prompt engineering is seen as a critical skill for developing AI products and tools, as it allows for the efficient creation and deployment of AI applications across various domains with minimal coding, thus increasing productivity and innovation.
How does the script suggest we should approach learning and applying AI technologies?
-The script encourages learning prompt engineering as a way to leverage AI effectively. It suggests using resources like HubSpot's chat GPT bundle to practice and incorporate AI into work and life, emphasizing the importance of understanding the principles behind AI and engineering rather than just the coding aspect.
What is the significance of understanding the underlying engineering principles in software development?
-Understanding the underlying engineering principles is crucial because it allows developers to think critically, solve problems effectively, and build complex software systems efficiently. It also helps in quickly grasping new technologies and adapting to the evolving landscape of software development.
How does the script address the concern that prompt engineering might replace traditional coding?
-The script acknowledges that prompt engineering may reduce the need for traditional coding skills but emphasizes that the fundamental principles of data analytics, engineering, and problem-solving will remain essential. It suggests that the ability to understand and apply these principles to achieve desired outcomes is irreplaceable.
What advice does Jensen give for someone starting their education anew?
-Jensen advises focusing on understanding the complexities of human biology and embracing the field of life sciences, as he believes that the future lies in life engineering, which includes digital biology and drug engineering, transforming life sciences into a more engineering-focused discipline.
Outlines
🚀 The Emergence of Prompt Engineering and AI Software Engineers
This paragraph discusses the Nvidia GTC conference where CEO Jensen Hong introduced the concept of prompt engineering as the new coding. He showcased the Nvidia Inference Microservice (NIM) that simplifies AI application development through minimal coding. The example of a healthcare company leveraging AI to analyze cancer data and devise targeted drugs is provided. The narrative also touches on the historical evolution of programming, from the early days of assembly code to the advent of high-level languages like Python and JavaScript, emphasizing the continuous abstraction and simplification of coding processes over the decades.
🤖 Introduction to Chat GPT and Prompt Engineering
The second paragraph introduces Chat GPT as a tool for practicing prompt engineering. It highlights the availability of a comprehensive Chat GPT bundle by HubSpot, which contains numerous prompts for users to experiment with. The paragraph discusses the utility of Chat GPT in information gathering, virtual assistance, and email automation. It emphasizes the free nature of the HubSpot bundle and its potential to enhance productivity. The video then transitions back to the main topic, defining prompt engineering and comparing it with traditional coding, noting that both involve input, processing, and output, but with prompt engineering being more streamlined and efficient.
🌟 The Evolution of Coding and the Role of Engineering Principles
This section delves into the personal experience of the speaker as a student and a software engineer, reflecting on the importance of understanding low-level programming languages like assembly code. It argues that this foundational knowledge is crucial for grasping engineering principles, which are the bedrock of software development, rather than just coding. The paragraph also discusses the broader implications of prompt engineering, suggesting that while it may simplify the coding process, the underlying engineering principles remain essential. These principles include data analytics, iterative problem-solving, and critical thinking, which are vital for defining and achieving desired outcomes in technology and other fields.
🌐 The Future of Life Sciences and Engineering
The final paragraph shifts focus to the future of life sciences, highlighting Jensen Hong's perspective on the field's transition from a scientific to an engineering discipline. The speaker discusses the potential of digital biology and life engineering, drawing parallels with their own background in pharmacology and computer science. The video concludes with a call to action for engineers, data scientists, and technical professionals to apply their expertise in solving complex problems in life sciences, drug discovery, and addressing global challenges like climate change. The speaker expresses hope for a new generation of individuals who will drive innovation in these areas, concluding the video on a forward-looking and optimistic note.
Mindmap
Keywords
💡Prompt Engineering
💡AI Software Engineer
💡API Providers
💡Coding
💡Software Engineering
Highlights
Jensen Huang, Nvidia's CEO, introduced the concept of prompt engineering as the new coding at the GTC conference.
Nvidia's inference micros, service or NIM, simplifies AI application development through prompt engineering, requiring minimal coding.
Healthcare companies can utilize AI applications to analyze cancer genetics and imaging data, potentially discovering targeted drugs in minutes.
The notion that traditional coding may become obsolete, with prompt engineering taking its place, was discussed by Jensen Huang.
Devon, the first AI software engineer, was introduced and demonstrated benchmarking the LLM, LLa3 on different API providers.
The historical development of programming languages, from assembly code to high-level languages like Python and JavaScript, shows a trend of simplification and abstraction.
Margaret Hamilton's contributions to software engineering and the Apollo 11 moon landing were mentioned, emphasizing the importance of engineering principles.
The evolution from low-level programming languages to high-level languages and frameworks has made software development more accessible and efficient.
Prompt engineering and traditional coding share a similar process of input, processing, and output, but with less complexity in prompt engineering.
The importance of understanding the underlying engineering principles, even when using high-level languages and tools, was emphasized for effective problem-solving and critical thinking.
Jensen Huang suggests that the future may see the field of life sciences transform into life engineering, with digital biology becoming a key area of focus.
The potential for AI and engineering principles to revolutionize drug discovery and life sciences was discussed, moving from scientific discovery to engineered solutions.
The importance of asking the right questions and understanding the desired outcomes in engineering and data science was highlighted.
The video encourages viewers to embrace the power of AI and learn prompt engineering to increase productivity and develop AI products and tools.
HubSpot's chat GPT bundle was mentioned as a resource for practicing prompt engineering, offering pre-made prompts for various applications.
The video concludes with a call for technical professionals to use their expertise in engineering solutions for life sciences, drug discovery, and addressing global challenges like climate change.
Transcripts
in the Nvidia GTC conference their CEO
Jensen Hong demonstrated how prompt
engineering is the new coding he
introduced the Nvidia inference micros
service or Nim that basically bundles
together all the software that you need
in order to build and deploy an AI
application in any domain that you want
mostly through prompt engineering you
only need like three lines of code or
less for example if you're a Healthcare
company and you have a lot of data on
cancer genetics and imaging you can
deploy an AI application that's able to
analyze all of the data that you have
and come up with novel targeted drugs
for specific types of cancer in
literally minutes and what's interesting
is that just a few weeks back in a
conference he said it is our job to
create Computing technology such that
nobody has the program basically coding
is dead and the future is impr prompt
engineering now couple this with Devon
and today I'm really excited to
introduce you to Devon the first AI
software engineer and here is Devon
being prompted to Benchmark the llm
llama 3 on three different types of API
providers I'm going to ask Deon to
Benchmark the performance of llama on a
couple different API
providers from now on Devon is in the
driver's seat we'll come back to this
example later because the prompt itself
over here holds the key to answering the
question is prompt engineering truly the
new coding you will leave this video
understanding what Jensen means by
nobody will need to learn coding anymore
and what you should be focusing your
attention on
instead
in 1843 aah Lovel wrote algorithms for a
mechanical General purposed computer
she's often celebrated as the first
computer programmer now mve to the mid
20th century people usually call this
the birth of modern programming looking
forward 20 years I'm quite certain that
the coming of the computer will have a
significant effect on all businesses and
most private lives during World War II
the first electrical computers were
developed such as the Colossus which is
used by British code Breakers and the
eniac in the United States this time
period was also an assembly code
introduced so you no longer have to do
the 0101 01s this development made
coding so much faster and there was
quickly other more higher level
languages that were developed such as
Fortran in 1957 which is developed by
IBM and Cobalt which was developed in
1959 for business data processing as
well as lisp for artificial intelligence
research contrary to what a lot of
people think that AI was something that
is like this new fangle thing actually
AI research started in the 1960s at this
point coding was a lot easier and faster
but compared to modern times it's not
even close for example this is Margaret
Hamilton standing next to the kesir RO
to get a poloo to the Moon she was also
the first one that came up with the term
of software engineering keep that in
mind this is very important and we'll
come back to this later but first let's
finish our history lesson the 1970s saw
the develop velopment of an even higher
level programming language called SE
this is a language that's now able to
support complex data structures and
algorithms which greatly facilitated the
development of very complex large
software systems moving into the end of
the century the 1980s saw the rise of a
lot of software engineering principles
which was quickly adopted with the Avent
of C++ and inherently an objectoriented
programming language and of course
finishing off the century is when the
internet came about but it's very hip to
be on the internet right now this led to
a need for languages that supported web
development starting off with HTML CSS
and JavaScript in order to enable
Dynamic web content finally going into
the 21st century everybody was on the
internet and these things called
smartphones became very popular so did
mobile development this of course led to
a development of lots of different new
languages as well as Frameworks that can
make sure that the software is more
scalable more safe and faster to write
these languages and Frameworks became
more and more abstracted away in high
level like pyth on Ruby react and tensor
flow for machine learning a lot of
people who use these highlevel languages
genuinely don't even know about how
memory management works which was like a
huge thing just a decade ago but yes
finally now going into the 2020s the
decade that we live in now this trend of
things being more and more abstracted
weight does continue with so much data
that's being collected there's now a
huge emphasis artificial intelligence
where machine learning as well as data
science developer tools and environments
such as docker kubernetes and
cloud-based platforms are becoming more
and more popular as we build and run
more and more complex software as you
can see the trend that has been going on
since the 1940s is that you have
something super tedious and somebody
comes along and abstracts some of these
Concepts and processes to make coding
simpler and faster and then somebody
comes along again and does the same
thing to abstract more things away so it
becomes simpler and faster so on and so
forth regardless of how prompt
engineering and traditional coding
intermingle and develop moving into the
future I think we can all agree that
prompt engineering is a must-learn skill
in order to be more productive in life
and in work and eventually be able to
develop AI products and tools the
easiest way to start practicing prompt
engineering is through chat gbt luckily
you don't need to figure things out
yourself from scratch HubSpot has a full
chat GPT bundle that contains hundreds
of proms that you can start using to
practice prompt engineering the
resources are really well laid out and
they give you lots of ideas on how you
can incorporate Chachi BT into your work
and your life to make you much more
productive for example have you ever had
to gather a lot of information and read
through a lot of information to prepare
for a presentation or a report Chach p
is actually really good at this it's
able to consolidate a lot of different
articles tutorials and different
resources it can be your virtual
assistant your personal conci your
writing assistant and so much more my
favorite High Roi section is being able
to automate my emails and to be able to
filter through news because there's just
way too much news out there and not
enough time the best part of this bundle
it is completely free you can download
it at add this link over here also
linked in description thank you so much
HubSpot for providing these free
resources in order to help us Leverage
The Power of AI and for sponsoring this
portion of the video now back to the
video
engineering engineering come in prompt
engineering is defined as the process
where you guide generative artificial
intelligence generative AI solutions to
generate desired outputs like for
example you can ask Devon to come up
with a benchmark comparison for llama
being run on three different API
providers then comes up with these
comparison numbers and even makes a nice
little graph that gets deployed now
let's compare that to a definition of
coding which is the process of creating
instructions that computers then
interpret and follow to do the same
thing of benchmarking llama on three
different API providers you would use a
highlevel modern programming language
such as python or some flavor of
JavaScript you need to read a
documentation to figure out exactly how
to use these apis run the code to run
Lama the large language model record the
time it takes then explicit create that
visualization and deploy it while
probably wasting a lot of time just
debugging things because things rarely
work the first time as you can see to
accomplish the same task coding is a lot
more complex and a lot more timec
consuming but what I want to point out
is that prompt engineering and coding
using python or JavaScript or whatever
language actually still follow the same
process first you have some input or
instruction whether that be through a
prompt or code then you have some sort
of processing that happens whether that
be within in artificial intelligence or
directly to the computer itself and
finally you get this desired output of
that nice little visualization that got
deployed so you see from Modern Day
encoding to prompt engineering it
actually follows that same Trend since
the 1840s you have some sort of
computation process to get the output
that you want but someone comes along to
make this tedious process a little bit
easier by abstracting away some things
there still following that similar
process of having some sort of input
some sort of computation processing that
happens and then your desired output at
the end so it really does seem like
prompt engineering is just the evolution
of coding doesn't
it since grad school when I was doing my
degree in a University of Pennsylvania
one of the first classes that I had to
take which forces you to start from the
lowest level programming language of
assembly code then moving up to C and
C++ Java and then python I still
remember wanting to JMP out of the
window when it took 30 lines of code
just to write a for Loop if you know you
know basically if you don't that's like
the command or function that you have to
use in order to go from one line of code
to some other line of code anyways it
was an awful experience and I have
definitely had times when I have burst
into tears and I had no idea why they
put us through this because who still
programs an assembly code and for most
people now even C or C++ were even Java
but you see after I graduated and got a
job as a software engineer I realized
that that was one of the most important
courses I have ever taken because from
forcing me to program at such a lowlevel
language and then going upwards I was
able to understand what was going on
beneath the hood and the concepts
engineering principles that's been
abstracted away these days as well as
how to think critically build things and
continue coding despite tears streaming
down my face these concepts of
engineering principles is now why I'm
able to learn languages so quickly to
grasp new technologies and learn how to
use them very very quickly you see
there's a difference between coding and
Engineering coding is just the way that
you're getting a computer to do certain
things but engineering is figuring out
what it is and what's the best way to
build the thing that you want to build
going back to Margaret Hamilton from
earlier whose contributions led to the
successful Landing of AP poloo she was
also the one that came up with the term
of software engineering you see before
that we were implementing pretty simple
things like just counting things doing
some addition processing some things
like summing up some numbers doing some
simple calculations but as the demand
for computation increased and more and
more high level coding languages came
about we were able to write more complex
pieces of software that call for more
engineering principles on how it is that
you should be building software so that
it's fast accurate scalable allows you
to work in parallel with other Engineers
so it's not just the language itself
that's being more abstracted away and
easier to use these highlevel
engineering principles were also being
developed Concepts like data structures
algorithms objectoriented programming
containerization these are all
fundamental developments and huge
Paradigm shifts for how we think about
and develop software coding is simply
the language we use in order to get the
computer to implement these engineering
principles to build the things that we
want to build now going back to this
prompt itself to Deon I'm going to ask
Devon to Benchmark the performance of
llama and a couple different API
providers if you break it into pieces
the first part is hey Devin I'll like
for you to Benchmark llama 2 on three
different providers second part is
replicate together in perplexity third
part is figure out their API formats and
the fourth part is write a script that
sends the same prompt slpms to all of
them okay so if you have absolutely no
understanding of engineering or how any
of these computery things work first of
all how would you even have known that
it's an important thing to Benchmark
llama on different providers why are
there even different providers why is
there differences between them why do I
care if there's any differences then of
course you need to know what are these
different providers for figuring out
their API formats if you like don't know
what an API is like the concept of
software interacting with other software
even if it doesn't matter how it is that
you can pull from an API still how would
you have like come up with this prompt
then script what is script why does it
matter how it is that I'm sending the
same parameters or the prompts uh to
these providers assuming again that I
even know to care about the speeds of
these different providers before y'all
come after me and go like oh like these
things eventually can be simplified as
well well so you don't have to
explicitly write things no no no I'm not
saying that it won't I personally
actually don't doubt that promt
engineering will become better and
better and replace coding in the
traditional sense I agree that everybody
can become a programmer at some point
that you don't need to learn specific
languages like C Java python but what
won't get replaced is the principles
that govern data analytics data science
and engineering being able to
iteratively figure out what needs to be
changed in order to get a better output
figure out what even is the output that
you want you'll be surprised by how many
non-technical people genuinely don't
know what they want like two-thirds of
my job at meta is going like hm is that
really what you want if you're a
software engineer data analyst data
scientist ml engineer AI engineer
whatever some sort of technical person
what you do isn't just coding it's
figuring out the right question
understanding the appropriate solution
and breaking it down into things that
are actually actionable problem solving
critical thinking let's go back now to
Jensen's statement nobody has to program
he says nobody has the program anymore
not nobody has to be an engineer a
scientist an
analyst but what to study now finally
going back to the rest of Jensen was
saying when asked about what people
should learn these days what would you
give me as an advice for something to
pursue if I were starting all over again
I would realize uh one thing one of of
the most complex fields of science is
the understanding of biology human
biology what he follows up with is also
very interesting nobody in computer
science nobody says car Discovery we
don't say computer Discovery we call it
engineering and every single year our
computer science our software becomes
better and better than the year before
however Life Sciences is sporadic life
science to life engineering is upon us
and that digital biology will be a field
of engineering not a field of science
Life Sciences to life engineering drug
Discovery to drug engineering as someone
who has a degree in both pharmacology
and computer science I can definitively
say that as it currently stands drug
Discovery life sciences and Engineering
are very different like seriously most
of the drugs we discover are actually
just wandering around the universe and
on a whim just discovering that a petri
dish of bacteria is being killed by some
sort of thingy that we now know as
penicillin and some dude wandering
around the Eastern Islands laying around
with dirt found this microorganism that
just happens to produce an antibiotic
which we now know as rapamycin I hope
that this is going to start a whole
generation of people who enjoy working
with proteins and chemicals and
Engineering these amazing things that
are more energy efficient all of these
inventions are going to be part of
engineering not scientific discovery I
hope so too that those of us who are
Engineers data scientist data analysts
technical people use our expertise to
engineer Solutions in the life science
Sciences drug Discovery climate change
thank you all so much for watching this
video please let me know what your
thoughts are about this topic and I'll
see you in the next video or live
stream
浏览更多相关视频
Books every software engineer should read in 2024.
Is Prompt Engineering the NEW Software Engineering?
How I Learned to Code in 4 MONTHS & Got a Job Offer (no CS Degree)
AI Engineering with Scrimba CEO Per Borgen – freeCodeCamp.org Podcast Interview
If U Use GitHub CO-PILOT I Wont Hire You!?
Is Coding Still Worth Learning in 2024?
5.0 / 5 (0 votes)