Agents IA

Welcome to AI agents. After watching this video, you will be able to explain what AI agents are and
discuss the characteristics and types of AI agents. Additionally, you will be able to explore multi-agent
systems and the various applications of AI agents. Did you know? The global AI market is expected to
increase to $1.8 trillion within the next seven years with a staggering compound annual growth rate, or
CAGR, of 36.6%. According to Gartner, using AI agents in customer service will save companies around
$80 billion by reducing labor costs. Understanding AI agents is crucial as it enables various businesses to
use automation and data-driven decision making, further improving efficiency, innovation, and
competitiveness in the respective fields. Let's define what AI agents are. AI agents are software
programs that interact with their environment, collect and process data, and perform tasks on their own
to achieve the goals set by humans. They can make decisions, solve problems and adapt to new
information without needing constant human help. So how do AI agents work? Let's take the example of
a self-driving car. It includes several key phrases, perception, where AI agents use sensors like cameras
and radar in a car to gather data about the environment, spotting vehicles, pedestrians, and road signs.
By understanding phase algorithms process this data to identify objects, their speed and movement.
Then you have decision-making where the agent uses a knowledge base and logic and determines when
to accelerate, brake, or turn. Further, in action phase, actuators execute these decisions, controlling the
car's steering and brakes to navigate safely. Finally, learning phase, where the agent improves over time
using machine learning, analyzing past experiences to enhance performance. For a self-driving car, this
means continually learning from its driving experiences. AI agents possess several key characteristics
such as social ability, autonomy, reactiveness, and proactiveness. Social ability refers to an agent's
capacity to communicate and collaborate with other agents and entities in its environment. Autonomy is
the agent's ability to operate independently and make decisions without external help. Reactiveness
allows agents to respond immediately to changes in their environment, while proactiveness enables
them to take initiative and make decisions to achieve their objectives. Let's discuss the areas where each
of the characteristics can be used. The social ability feature of AI agents is extensively used in the
healthcare sector. In healthcare, chatbots use social skills to give medical advice, check symptoms, and
schedule appointments and reminders. Self driving cars use the autonomy feature to navigate roads,
make driving decisions, and operate without human assistance. Reactiveness is employed in
thermostats, which adjust home temperatures depending on real-time data. In industries, predictive
maintenance systems use the reactiveness feature to analyze equipment data and schedule
maintenance before breakdowns occur. In today's fast changing tech world, managing and coordinating
many independent systems is becoming more important. The need appears in different areas, like
managing resources spread over a large area, operating smart transportation systems, and creating
complex simulations. This is where multi-agent systems come in. In multi-agent systems, multiple agents
work together to achieve individual and collective goals. Multi-agent systems enable distributed
problem-solving, cooperative decision-making, and emergent behaviors from agent interactions. These
multi-agent systems are used in various fields, including online marketplaces, robotic coordination, and
traffic management. In online marketplaces, buyer and seller agents negotiate prices and manage
transactions for their users, facilitating efficient and automated commerce. In robotic coordination,
teams of robots collaborate on complex tasks such as warehouse logistics or search and rescue missions.
In traffic management, agents representing autonomous cars communicate to optimize traffic flow and
reduce congestion. In today's digital age, tech giants employ AI agents in their services and products.
Google integrates AI agents into YouTube for content recommendation and moderation, analyzing user
behavior to suggest videos and manage content. Additionally, AI agents help in detecting and removing
inappropriate content, enforcing community guidelines, and managing copyright issues. In Gmail, AI
agents power features like smart compose and smart reply, which suggest responses based on the
email's content. The AI agents also help organize emails, filter out spam, and find important messages.
Google Maps uses AI agents for real-time navigation, gathering real time-data from various sources to
provide accurate route recommendations, estimate travel times, and offer alternative routes to avoid
traffic jams. Amazon's Alexa uses AI for voice recognition, understanding natural language, and
controlling smart home devices. The Amazon.com e-commerce platform uses AI to recommend
products, improve search results, and manage inventory. Amazon Web Services, or AWS, offers AI
services for developers such as machine learning models, natural language processing, and computer
vision tools. As we enter a new era, AI agents are becoming more than just tools. They are partners in
making strategic decisions and engaging with customers. Their impact will continue to increase. With
this, you've reached the end of the video. In this video, you learned that agents in AI are software
programs that engage with surroundings, process data, and autonomously carry out tasks to meet
human-defined goals. You also learned about the key traits of AI agents. Social ability means agents can
talk and work with others, autonomy means they can work alone and decide things by themselves.
Reactiveness lets them quickly react to changes, and proactiveness means they can take charge and
make decisions to reach their goals. Finally, you learned that multi-agent systems facilitate distributed
problem-solving and cooperative decision-making, and how the same can be applied in varied areas
such as online marketplaces, robotic coordination, and traffic management.

Qu'est-ce qu'un agent de l'IA ?

2024 will be the year of AI agents. So what are AI agents? And to start explaining that, we have to look
at the various shifts that we're seeing in the field of generative AI. And the first shift I would like to talk
to you about is this move from monolithic models to compound AI systems. So models on their own are
limited by the data they've been trained on. So that impacts what they know about the world and what
sort of tasks they can solve. They are also hard to adapt, so you could tune a model, but it would take an
investment in data and in resources. So let's take a concrete example to illustrate this point. I want to
plan a vacation for this summer, and I want to know how many vacation days are at my disposal.

Lisez la vidéo à partir de :1:13 et suivez la transcription1:13

What I can do is take my query, feed that into a model that can generate a response.

Lisez la vidéo à partir de :1:27 et suivez la transcription1:27

I think we can all expect that this answer will be incorrect because the model doesn't know who I am
and does not have access to this sensitive information about me. So they can help with summarizing
documents, they can help me with creating first drafts for emails and different reports I'm trying to do.
But the magic gets unlocked when I start building systems around the model and actually take the
model and integrate them into to the existing processes I have. So if we were to design a system to solve
this, I would have to give the model access to the database where my vacation data is stored. So that
same query would get fed into the language model. The difference now is the model would be
prompted to create a search query, and that would be a search query that can go into the database that
I have. So that would go and fetch the information from the database, output an answer, and then that
would go back into the model that can generate a sentence to answer. So Maya, you have ten days left
in your vacation database. So the answer that I would get here would be correct. This is an example of a
compound AI system, and it recognizes that certain problems are better solved when you apply the
principles of system design.

Lisez la vidéo à partir de :2:56 et suivez la transcription2:56

So what does that mean by the term system? You can understand there's multiple components, so
systems are inherently modular. I can have a model, I can choose between tuned models, large language
models, image generation models, but also I have programmatic components that can come around it.
So I can have output verifiers, I can have programs that can take a query and then break it down to
increase the chances of the answer being correct. I can combine that with searching databases, I can
combine that with different tools. So when we talking about a system approaches, I can break down
what I desire my program to do and pick the right components to be able to solve that. And this is
inherently easier to solve for than tuning a model. So that makes this much faster and quicker to adapt.
Okay, so the example I used below is an example of a compound AI system. You also might be popular
with retrieval augmented generation, which is one of the most popular and commonly used compound
AI systems out there. Most rag systems and the example I used below are defined in a certain way. So if I
bring a very different query, let's say I ask about the weather. In this example here, it's going to fail,
because the path that this program has to follow is to always search my vacation policy database, and
that has nothing to do with the weather. So when we say the path to answer a query, we are talking
about something called the control logic of a program.

Lisez la vidéo à partir de :4:44 et suivez la transcription4:44

So, compound AI systems, we said most of them have programmatic control logic. So that was
something that I defined myself as the human. Now let's talk about where do agents come in? One
other way of controlling the logic of a compound AI system is to put, put a large language model in
charge. And this is only possible because we're seeing tremendous improvements in the capabilities of
reasoning of large language models. So large language models, you can feed them complex problems,
and you can prompt them to break them down and come up with a plan on how to tackle it. Another
way to think about it is on one end of the spectrum, I'm telling my system to think fast, act as
programmed, don't deviate from the instructions I've given you. And on the other end of the spectrum,
you're designing your system to think slow. So create a plan. Attack each part of the plan, see where you
get stuck, see if you need to readjust the plan. So I might give you a complex question, and if you would
just give me the first answer that pops into your head, very likely that answer might be wrong, but you
have higher chances of success if you break it down. Understand, where do you need external help to
solve some parts of the problem, and maybe take an afternoon to solve it? And when we put LLMs in
charge of the logic, this is when we're talking about an agentic approach. So let's break down the
components of LLM agents. The first capability is the ability to reason, which we talked about. So this is
putting the model at the core of how problems are being solved. The model will be prompted to come
up with a plan and to reason about each step of the process along the way. Another capability of agents
is the ability to act, and this is done by external programs that are known in the industry as tools. So
tools are external pieces of the program, and the model can define when to call them and how to call
them in order to best execute the solution to the question they've been asked. So an example of a tool
could be search, searching the web, searching a database at their disposal. Another example could be a
calculator to do some math. This could be a piece of program code that maybe might manipulate a
database. This can also be another language model that maybe you're trying to do a translation task and
you want a model that can be able to do that. And there's so many other possibilities of what can do
there. So these could be APIs, basically any piece of external program you want to give your model
access to. Third capability, that is the ability to access memory. And the term memory can mean a
couple of things. So we talked about the models, thinking through the program, kind of how you think
out loud when you're trying to solve through a problem. So those inner logs can be stored and can be
useful to retrieve at different points in time. But also this could be the history of conversations that you
as a human had when interacting with the agent, and that would allow to make the experience much
more personalized. So the way of configuring AI agents, there's many ways to approach it. One of the
more most popular ways of going about it is through something called react, which as you can tell by the
name, combines the reasoning and act components of LLM agents. So let's make this very concrete.
What happens when I configure a react agent? You have your user query that gets fed into a model, so
an LLM. The LLM is given a prompt, so the instructions that's given is don't give me the first answer that
pops to you. Think slow, plan your work, and then try to execute something, try to act. And when you
want to act, you can define whether if you want to use external tools to help you come up with the
solution. Once you get, you call a tool and you get an answer, maybe it gave you the wrong answer or it
came up with an error. You can observe that. So the LLM would observe the answer, would determine if
it does answer the question at hand, or whether it needs to iterate on the plan and tackle it differently
up until I get to a final answer. So let's go back and make this very concrete again. Let's talk about my
vacation example. And as you can tell I'm really excited to go on one, so I want to take the rest of my
vacation days. I'm planning to go on to Florida next month. I'm planning on being outdoors a lot and I'm
prone to burning. So I want to know what is the number of two ounce sunscreen bottles that I should
bring with me. And this is a complex problem, there's a number of things to plan. One is how many
vacation days am I planning to take? And maybe that is information the system can retrieve from its
memory, because I asked that question before. Two is how many hours do I plan to be in the sun? I said
I plan to be in there a lot. So maybe that would mean looking into the weather forecast for next month
in Florida and seeing what is the average sun hours that are expected. Three is maybe going to a public
health website to understand what is the recommended dosage of sunscreen per hour in the sun. And
then four, doing some math to be able to determine how much of that sunscreen fits into two ounce
bottles. So that's quite complicated. But what's really powerful here is there's so many paths that can be
explored in order to solve a problem. So this makes the system quite modular, and I can hit it with much
more complex problems. So going back to the concept of compound AI systems, compound AI systems
are here to stay. What we're going to observe this year is that they're going to become more agentic.
The way that I like to think about it is you have a sliding scale of LLM autonomy,

Lisez la vidéo à partir de :10:59 et suivez la transcription10:59

and you would, the person defining the system would examine what trade offs they want in terms of
autonomy in the system for certain problems, especially problems that are narrow, well defined, so you
don't expect someone to ask about the weather when they mean to ask about vacations. So a narrow
problem set, you can define a narrow system like this one. It's more efficient to go the programmatic
route because every single query will be answered the same way. If I were to apply the agentic approach
here, there might be unnecessarily looping and iteration. So for narrow problems, programmatic
approach can be more efficient than going the agentic route. But if I expect to have a system accomplish
very complex tasks, like say, trying to solve GitHub issues independently and handle a variety of queries,
a spectrum of queries. This is where an agentic route could be helpful because it would take you too
much effort to configure every single path in the system. And we're still in the early days of agentic
systems. We're seeing rapid progress when you combine the effects of system design with agentic
behavior. And, of course, you will have a human in the loop in most cases, as the accuracy is improving.

Robotique et automatisation
Welcome to robotics and automation. After watching this video, you will be able to define the term
robotics and how robots work. You will further discuss how AI technologies are integrated into robotics.
You will also be able to examine how robotics enable automation in various industries. Lastly, you will
explore the concept of robotic process automation or RPA and its applications. Across the industrial
world, companies are investing heavily in robotics and automation. With the advancement of
technology, robots have become more capable of performing complex tasks that were previously done
by humans. This shift is driven by the need for higher productivity, improved quality, and the ability to
operate continuously. According to a recent global industrial robotic survey conducted by McKenzie,
many industries plan to spend 25% of their budgets on automated systems over the next four years. Let
us understand the term robotics and how it enables automation. Robotics involves designing,
constructing, and operating robots, machines that can perform tasks by themselves or with some help.
Robots are complex systems made up of several key components, sensors, actuators, and controllers.
Sensors act as the robots eyes and ears, gathering information about the environment. For instance, a
camera sensor helps a robot navigate by capturing images and detecting obstacles. While temperature
sensors ensure the robot operates within safe limits. Actuators are the robots muscles, enabling
movement and interaction with the environment. This includes electric motors which move robotic arms
and wheels, and hydraulic and pneumatic actuators, which control various parts of the robot.
Controllers act as the robot's brain and run software that determines the robot's actions. For instance, in
a robotic arm, the controller interprets sensor data and directs the motors to move the arm accurately.
Robots use these components and work around various AI technologies, such as artificial intelligence,
machine learning, and natural language processing or NLP to enable automation. Let's explore a few
such examples. A robot vacuum cleaner uses AI to clean your home efficiently. It has sensors to detect
dirt, walls, and furniture, and an AI controller to map your home and choose the best cleaning path. The
actuators adjust the vacuums wheels and brush to clean surfaces, such as carpets and hardwood floors.
A smart lawn mower uses machine learning to mow your lawn efficiently. It has sensors to detect grass
height, obstacles, and lawn boundaries. The controller learns the lawns layout and grass growth patterns
to optimize mowing routes and schedules. The actuators adjust the mowers wheels and blades to
ensure an even cut. A voice activated smart speaker robot uses NLP to understand and respond to your
spoken commands. It has microphones as sensors to capture your voice, an NLP controller to process
and understand language, and actuators to perform actions like playing music, providing updates, or
controlling smart home devices. Did you know that these devices act as self operating robots that do not
need human intervention? As they can figure things out on their own? Let's discuss more such
examples. Space exploration and deep sea exploration, find use of self operating robots. Space rovers,
like the ones on Mars, explore other planets, collect valuable information, and send it back to Earth all
by themselves. Similarly, underwater drones explore deep parts of the ocean, collect samples, and do
research in places that are hard for people to reach. Robots working directly alongside humans are
known as cobots or collaborative robots. Unlike traditional robots, cobots are designed to interact
directly with people, making automation more accessible and versatile. Cobots are equipped with
advanced sensors and AI technologies, allowing them to communicate and coordinate their actions to
complete tasks that require teamwork. For instance, in automobile manufacturing, one cobot can hold a
part in place while the other welds it, ensuring precision and speed. In logistics, cobots can work
together to sort and move packages efficiently. Robotics uses AI to enable automation in many
industries. In health care, surgical robots perform precise operations, reducing recovery times and
improving patient outcomes. Rehabilitation robots help patients recover mobility and strength, offering
personalized therapy. Agriculture robots plant, harvest, and monitor crops with precision, boosting
yields and reducing waste. They optimize irrigation, apply fertilizers accurately, and monitor crop health
in real time. In retail, self checkout kiosks and robots manage inventory and assist customers,
streamlining operations, and improving customer service. This reduces wait times and ensures stock
availability. Let's now discuss another concept called robotic process automation or RPA. RPA is a type of
computer software that helps create, use, and control virtual robots. These robots act like humans when
they work with digital systems and software. A PEGA system survey among 500 decision makers
employing a range of RPA solutions revealed, 63% view RPA as a major component of overall digital
transformation, 23% said it is the most important component, whereas 14% said it is a minor
component. RPA can automate repetitive and rule-based tasks across various industries. In finance and
accounting, RPA automates invoice processing by capturing, validating, and posting invoices. It also
manages payroll calculations, deductions, and payments, and generates financial reports by extracting
data from multiple sources. In human resources, RPA streamlines employ onboarding by collecting
documents, entering data, and setting up profiles, leave management by handling leave requests and
updating records, recruitment by screening resumes, scheduling interviews, and sending
communication. In this video, you learn that robotics involves designing, constructing, and operating
robots to perform tasks with or without humans. It utilizes sensors to observe processors for decision
making, actuators for movement, integrating technologies like AI, machine learning, and NLP for
automation. Further, you learned about cobots, which collaborate with humans using advanced sensors
and AI, communicating and coordinating tasks that require teamwork. Lastly, you learned that robotic
process automation or RPA automates repetitive tasks and helps create, use, and control virtual robots.

Transformer les entreprises grâce à l'IA

Welcome to transforming businesses through AI. After watching this video, you will be able to recognize
how AI is transforming business operations and identify how various industries use AI to improve
efficiency. AI is the game changer that is revolutionizing business operations with smart workflow
automation, powerful data analysis, and efficient decision making. This transformation is helping
companies achieve new levels of success and stay competitive in fast paced markets. Accenture
predicted that AI in business operations would double workforce efficiency and boost profitability by an
impressive 38% over the next 10 years. Organizations handle a wide array of tasks on a daily basis. AI
automates repetitive tasks like data entry, scheduling, and report generation, allowing employees to
focus more on creative and strategic work. Depending on which tasks you're looking to automate, a
wide range of AI tools are available. For example, in customer service, AI chat bots understand natural
language and customer sentiment to offer personalized support. Helping businesses resolve issues faster
and making customers happy. AI chat bots are also more budget-friendly than human agents, saving
businesses money. Take AirHelp, for example. This global service uses AI to assist airline passengers with
flight statuses, delivering speedy and cost efficient customer service that boosts satisfaction and
strengthens their brand reputation. Recruitment, AI tools analyze resumes and video interviews to find
the best candidates. This helps businesses to fill positions faster, reduce productivity gaps, and ensure
quality hires. Hilton's AI chat bot accelerates hiring by answering candidates questions and scheduling
interviews, saving HRs time, and potentially lowering costs. In accounting, AI platforms automate payroll
and auditing. This helps businesses in reducing the risk of human error in manual processes, leading to
enhanced accuracy. For instance, EY has transformed auditing using AI to evaluate fraud risks and spot
hidden patterns. So EY can catch financial misstatements early, keeping clients and investors safe and
sound. AI allows businesses to make informed decisions by swiftly analyzing large data volumes. It
identifies patterns and trends that help to predict outcomes based on the history of data. This leads to
fewer errors. AI-driven tools offer real-time analytics and dashboards, helping organizations monitor
performance and identify improvement opportunities. In marketing and sales, AI tools analyze
purchasing patterns and suggest items based on customer purchases and browsing history. This boosts
sales and conversions, leading to higher customer satisfaction. Amazon uses AI for personalized
recommendations and inventory management. This reduces the risk of stock-outs and ensures they
have the right products in stock. Ultimately boosting sales and customer satisfaction. In manufacturing,
AI tools aid in predictive maintenance by using sensors and data analysis to monitor equipment,
preventing costly breakdowns. This allows businesses to schedule maintenance proactively, and avoid
catastrophic equipment failures. For instance, BMW uses AI to inspect parts on the production line,
quickly spotting quality issues. This helps BMW produce higher quality cars more efficiently. Similarly, in
health care and research, AI accelerates drug discovery and development by analyzing medical data,
helping businesses develop successful drugs faster. For example, IBM's healthcare technology uses AI to
analyze large volumes of medical data, providing insights for genomic analysis and disease diagnosis.
This attracts new clients and strengthens IBM's reputation in the healthcare industry. AI is enhancing
creativity in businesses by generating innovative ideas and overcoming creative blocks. This helps
businesses resonate with their target audience, leading to higher engagement and increased sales. For
example, in design and marketing, AI helps designers create personalized and user friendly designs
through exploration, leading to happier customers and increased sales. Canva's magic design feature
suggests layouts, color schemes, and fonts, based on content type and user preferences. This enhances
their user experience and keeps people engaged with Canva. These capabilities are just a glimpse of the
many ways AI is utilized in business. From creating operational efficiency and elevating customer
experience to achieving significant cost savings, the benefits of AI are far reaching. In this video, you
learn that AI automates repetitive tasks for businesses, like data entry, scheduling, and report
generation, and allows employees to focus more on creative and strategic work. AI enables businesses
to swiftly analyze large data volumes by identifying patterns and trends, leading to fewer errors and
better predictions. AI generates innovative ideas and overcomes creative blocks by helping businesses
resonate with the target audience, leading to higher engagement and sales.
L'essor de l'IA générative pour les
entreprises
Welcome to the video on the Rise of Generative AI for Business. After watching this video, you will be
able to recognize the transformative impact of Generative AI on businesses and identify its practical
applications in various business contexts. Businesses are hitting remarkable milestones and transforming
technology at its application by harnessing Generative AI. Let's see what some experts have to say about
Generative AI. Arthur C Clarke famously said that any sufficiently advanced technology is
indistinguishable from magic. Perhaps the first time that you played with Generative AI, it did evoke a
sense of magic. Suddenly, for the first time in our history, we have technology that can speak our
languages, understand our requests, and produce entirely novel output. AI can write poetry and draw
other worldly images. It can write code. It can surprise and delight us with an original joke or musical
composition. It can create, and an act of creation often inspires wonder. AI isn't just about the digital
world. It's also about the physical world. Applied properly. imagine what AI can do for the pace of
discovery and innovation, what it can do for discovering new materials, for medicine, for energy, for
climate, and so many of the pressing challenges that we face as a species. Ultimately, our success
depends on how we approach AI. A survey by JP Morgan indicates that investors predict the highest
generative AI technology adoption in marketing at 28%, followed by legal services and insurance at 21%,
media at 20%, data analytics at 18%, and consumer technology at 13%. This indicates significant
transformations ahead for these industries. In the modern business landscape, companies are constantly
grappling with the need to be more efficient, innovative, and customer-centric. Let's explore some key
areas in businesses where generative AI is making contributions. Businesses often face the constant
need for latest content, be it marketing material, product descriptions, or social media posts. Generating
high quality content can be time consuming and resource-intensive. Here, Generative AI platforms step
in as game changers as they can analyze the target audience data to create unique and engaging
content. This helps organizations overcome creative blocks and maximize their reach and impact among
the audience. For example, Persado, a Generative AI platform, helped Vanguard boost their sales
conversion rates by 15%. Vanguard leveraged Persado to find the perfect phrases that captivate
customers and crafted messages with the right emotion and action. This enabled them to deliver
impactful content and seamlessly integrate data-driven insights into their marketing strategy. Another
challenge is that businesses gather enormous amounts of data, but transforming this information into
actionable insights can be overwhelming. Generative AI analyzes complex datasets to identify hidden
patterns and generate reports translating data into actionable insights. According to a survey by
Deloitte, employees leveraging generative AI save 1.75 hours daily, which equates to a full workday each
week. For example, Salesforce integrated Generative AI capabilities into its CRM platform to analyze
customer data and enhance sales and marketing strategies. This helped their sales team identify leads,
predict customer behavior, and personalize interactions to improve conversion rates. Traditional
customer service can be frustrating for both customers and agents. Repetitive tasks, like answering
questions or basic troubleshooting hinder their ability to address complex issues. Generative AI-
powered chatbots can handle repetitive tasks, providing fast and personalized answers to frequently ask
questions. Additionally, AI can analyze customer data and suggest real-time troubleshooting steps,
empowering agents to resolve issues efficiently. For example, Sephora is leveraging Generative AI
capabilities to deliver personalized beauty recommendations. When a customer shops at Sephora, their
preferences, purchase history, and even skin tone and concerns are logged into a database. Generative
AI helps process this data to craft personalized emails that suggest products specifically recommended
for that customer. Product development teams often struggle with limited design options and the slow
iterative prototyping process. Generative AI models can produce multiple variations, enabling
companies to shortlist the most appealing options. Nike, for instance, uses Generative AI to innovate its
product design process. By feeding data on materials, biomechanics, and fashion trends, it helps
generate numerous design possibilities. Not only big companies, but also startups are leveraging the
potential of Generative AI. Young startups often struggle with limited resources and tight budgets.
Creative content development, and product designs can be a significant expense for them. With the help
of Generative AI technology, the team members need not focus on mundane and repetitive tasks, but
on strategic planning and refinement. Along with cost effectiveness and increased efficiency, the early
adoption of Generative AI also provides a competitive advantage to startups. For example, Stitch Fix, a
full stack startup, uses GenAI to help stylists interpret customer feedback and provide product
recommendations. While these are just a few examples, the possibilities with Generative AI are
constantly evolving. As it matures, the benefits will likely become even more significant. The focus is
now on practical ways to use Generative AI while addressing its challenges, leading to more sustainable,
responsible and creative applications in various industries. In this video, you learned that businesses are
leveraging Generative AI in several key areas. Some of them include content generation, where
Generative AI can create high quality content, such as marketing materials, product descriptions, and
social media posts; data analysis, where generative AI identifies patterns and generates actionable
insights; customer service, where Generative AI-powered chat bots handle repetitive tasks and provide
personalized answers, improving customer satisfaction; and product development where AI models
generate multiple design variations, streamlining the prototyping process. Finally, you learn that
startups are also leveraging Generative AI as it helps with cost effectiveness and increased efficiency and
innovation.

Qu'est-ce que la Génération augmentée

de récupération (RAG) ?
Large language models, they are everywhere. They get some things amazingly right and other things
very interestingly wrong. My name is Marina Danilevsky. I am a senior research scientist here at IBM
research, and I want to tell you about a framework to help large language models be more accurate and
more up to date. Retrieval augmented generation, or RAG. Let's just talk about the generation part for a
minute. So forget the retrieval augmented. So the generation, this refers to large language models, or
LLMs, that generate text in response to a user query referred to as a prompt. These models can have
some undesirable behavior. I want to tell you an anecdote to illustrate this. So, my kids, they recently
asked me this question, in our solar system, what planet has the most moons? And my response was,
that's really great that you're asking me this question. I loved space when I was your age. Of course, that
was like 30 years ago. But I know this. I read an article, and the article said that it was Jupiter and 88
moons. So that's the answer. Now, actually, there's a couple of things wrong with my answer. First of all,
I have no source to support what I'm saying. So even though I confidently said I read an article, I know
the answer. I'm not sourcing it. I'm giving the answer off the top of my head. And also, I actually haven't
kept up with this for a while. And my answer is out of date. So we have two problems here. One is no
source. And the second problem is that I am out of date.

Lisez la vidéo à partir de :1:42 et suivez la transcription1:42

And these, in fact, are two behaviors that are often observed as problematic when interacting with large
language models. They are LLM challenges. Now, what would have happened if I'd taken a beat and first
gone and looked up the answer on a reputable source like NASA? Well, then I would have been able to
say, okay, so the answer is Saturn with 146 moons. And in fact, this keeps changing because scientists
keep on discovering more and more moons. So I have now grounded my answer in something more
believable. I have not hallucinated or made up an answer. And by the way, I didn't leak personal
information about how long ago it's been since I was obsessed with space. All right, so what does this
have to do with large language models? Well, how would a large language model have answered this
question? So let's say that I have a user asking this question about moons. A large language model
would confidently say, okay, I have been trained. And from what I know in my parameters during my
training, the answer is Jupiter.

Lisez la vidéo à partir de :2:53 et suivez la transcription2:53

The answer is wrong. But we don't know. The large language model is very confident in what it
answered. Now, what happens when you add this retrieval augmented part here? What does that
mean? That means that now, instead of just relying on what the LLM knows, we are adding a content
store. This could be open like the Internet, this could be closed like some collection of documents,
collection of policies, whatever. The point though now is that the LLM first goes and talks to the content
store and says, hey, can you retrieve from me information that is relevant to what the user's query was?
And now with this retriever augmented answer, it's not Jupiter anymore, we know that it is Saturn.
What does this look like? Well, first, user prompts the LLM with their question. They say, this is what my
question was. And originally, if we're just talking to a generative model, the generative model says, okay,
I know the response. Here it is. Here's my response. But now, in the RAG framework, the generative
model actually has an instruction that says, no, no, no, first go and retrieve relevant content. Combine
that with the user's question, and only then generate the answer. So the prompt now has three parts.
The instruction to pay attention to the retrieved content together with the user's question, now give a
response. And in fact, now you can give evidence for why your response was what it was. So now
hopefully you can see how does RAG help the two LLM challenges that I had mentioned before? So, first
of all, I'll start with the out of date part. Now, instead of having to retrain your model, if new
information comes up, like, hey, we found some more moons now to Jupiter again, maybe sadder again.
In the future, all you have to do is you augment your data store with new information, updated
information. So now, the next time that a user comes and asks the question, we're ready, we just go
ahead and retrieve the most up to date information. The second problem, source. Well, the large
language model is now being instructed to pay attention to primary source data before giving its
response, and in fact, now being able to give evidence. This makes it less likely to hallucinate or to leak
data because it is less likely to rely only on information that it learned during training. It also allows us to
get the model to have a behavior that can be very positive, which is knowing when to say, I don't know.
If the user's question cannot be reliably answered based on your data store, the model should say, I
don't know. Well, instead of making up something that is believable and may mislead the user. This can
have a negative effect as well, though, because if the retriever is not sufficiently good to give the large
language model the best, most highest quality grounding information, then maybe the user's query that
is answerable doesn't get an answer.

Adopter l'IA dans votre entreprise

Welcome to Adopting AI in Your Business. After watching this video, you will be able to explain the key
benefits of adopting AI in business operations. You will also be able to explore how businesses can
effectively integrate artificial intelligence into their operations. Finally, you will also be able to describe
the general steps to adopt AI in your business.

Lisez la vidéo à partir de ::27 et suivez la transcription0:27

AI is transforming business operations across industries, be it helping them surpass competitors,

improve customer interactions, or develop new ideas. AI adoption helps create new opportunities for
efficiency and growth. From e-commerce giants leveraging AI for personalized recommendations to
various manufacturing firms using AI-powered robots, the possibilities are endless. Adopting AI in
businesses offers several benefits. AI increases efficiency by automating repetitive tasks, freeing
employees for higher value work. It analyzes vast amounts of data to predict market trends and
consumer behaviors, enhancing decision making. It personalizes customer service through data-driven
recommendations, boosting satisfaction. Finally, AI accelerates innovation, allowing businesses to create
unique offerings and stand out from the competition. A recent global study conducted by Tata
Consultancy Services found that more than eight out of ten, or 86% of senior business leaders
worldwide have adopted AI to boost their current revenue streams or generate new ones. Let's delve
into real world examples of companies benefiting from AI adoption. Amazon utilizes AI across various
aspects of its business operations. Amazon's recommendation engine uses AI algorithms to analyze user
browsing history, purchase patterns, and demographic data to recommend products meeting customer
preferences. This personalized system boosts sales and enhances customer satisfaction by reducing
search times for relevant products. Amazon's AI-powered voice assistant, Alexa, uses natural language
processing, or NLP, and machine learning to handle voice commands. Enabling users to place orders,
control smart home devices, and access information hands-free. Amazon uses AI in its supply chain to
optimize its inventory and warehouses. AI-powered forecasting models analyze historical data and
external factors to predict product demand, optimize inventory levels, minimize stock outs, and
streamline logistics. Thereby reducing costs and improving overall efficiency in product delivery.

Lisez la vidéo à partir de :2:39 et suivez la transcription2:39

Tesla leverages AI in the automotive industry, specifically in autonomous driving technology. Tesla's
autopilot systems leverage AI algorithms, including computer vision and deep learning, to enhance the
performance of advanced driving assistance features. It analyzes data from cameras, radar, and sensors
to navigate, detect obstacles, and make real-time driving decisions, improving safety and efficiency. The
summon feature in Tesla's autopilot system uses AI to detect and react to its surroundings. It enables
drivers to use smartphone apps to remotely park vehicles within a specified range. Additionally, Tesla's
battery management system, or BMS, uses AI to predict energy requirements based on driving
conditions and driver behavior. Adjusting acceleration and climate control to conserve energy. It also
helps plan efficient routes with the navigation system, considering traffic and elevation. Lets now
explore a general roadmap consisting of various steps to adopt AI in business operations. Adopting AI is
a multifaceted process that requires careful planning and execution. From defining business goals and
identifying suitable use cases to data readiness and building AI capabilities, deploying AI solutions, and
monitoring and optimizing AI systems, each step is crucial for success. The journey begins with defining
the business goals. Businesses need to identify the specific problems they aim to solve with AI, engaging
stakeholders to understand their needs and outline clear objectives. It is crucial to ask questions like,
what are the pain points? And how can AI help achieve business goals?

Lisez la vidéo à partir de :4:21 et suivez la transcription4:21

After setting the business goals, the next step is to identify suitable use cases. Perform a thorough
analysis of your business processes to find specific areas where AI can be effectively applied. Focus on
tasks that are repetitive, time consuming, data driven or require complex decision making.

Lisez la vidéo à partir de :4:41 et suivez la transcription4:41

One of the most critical steps in adopting AI is data readiness. It is crucial for businesses to collect, clean
and organize their data to ensure it is ready for AI algorithms. Data readiness involves removing
inconsistencies, filling gaps, and ensuring the data is relevant to the problem. The next step is to build AI
capabilities within the organization. This includes training employees on AI concepts, tools, and
techniques. Investing in upskilling your workforce ensures they can effectively use and manage AI
technologies. The training program helps employees understand how AI can be applied to specific tasks.
Next, AI solutions are deployed, choosing the best AI tools and technologies that match your business
needs. The deployment is done in a phased manner, starting with pilot projects and gradually scaling up.
It is necessary to ensure that AI solutions are integrated seamlessly with the existing systems and
processes. In the final stage, after the AI solutions are deployed into the business environment, it
requires continuous monitoring and optimization to track performance regularly and identify and
resolve issues. This ensures that the AI models are updated and refined, making necessary adjustments
over time.

Lisez la vidéo à partir de :6:2 et suivez la transcription6:02

In this video, you learned that AI can be adopted in businesses to increase efficiency, enhance decision
making, improve customer service, and drive innovation. You also gained insights on a few real world
examples of companies that have adopted AI and obtained benefits from it. Finally, you learned general
steps for adopting AI in business. First, as part of the business requirement analysis, you set clear goals.
Next, you need to identify suitable use cases, identifying areas to apply AI, then prepare data by first
collecting it and then cleaning and organizing it. Further, build AI capabilities to train employees and
develop the necessary setup. Deploy AI solutions, integrate them with existing systems. And finally,
monitor and optimize AI systems to ensure they work well.

Cadres pour l'adoption de l'IA

Welcome to the Frameworks for AI Adoption. After watching this video, you'll be able to explain the
framework developed and used by IBM to adopt AI for business needs and explore the shift from a +AI
to an AI+ approach to adopting AI. Organizations are increasingly adopting artificial intelligence or AI to
drive innovation and gain a competitive edge. However, successfully harnessing the power of AI can be
challenging. To address this, organizations are turning to AI adoption frameworks to guide their AI
implementation efforts. These frameworks provide a structured approach to identifying the right use
cases, aligning AI initiatives with business goals, and creating a culture that supports responsible AI
development and deployment. One comprehensive framework designed to guide enterprises through
this transformative journey is the IBM AI ladder. The IBM AI ladder is a structured framework that
navigates organizations from basic data collection to full-scale AI integration across four key stages:
collect, organize, analyze, and infuse. Let's understand each of these stages in detail. The initial stage,
collect, involves gathering and storing high-quality data from diverse sources. IBM tools like the IBM
Cloud and Watsonx.data facilitate this process. The key actions include identifying data sources such as
databases and IoT devices, implementing collection tools, and ensuring data quality. For instance, a
retail company uses Watsonx.data to gather sales data from various channels. The next stage in the IBM
AI ladder is organize. This stage focuses on cleaning, categorizing, and making data accessible for
analysis. IBM Cloud Pak for data, IBM InfoSphere, IBM DataOps, and IBM watsonx.governance, help with
data management, governance, and security. Actions include cleansing data, categorizing it for easy
retrieval, implementing governance policies, and centralizing storage. In the analyze stage, AI's potential
is harnessed through advanced analytics and machine learning models. IBM tools like SPSS, Cognos
Analytics, Watson Machine Learning, and IBM watsonx.ai enable efficient model building, testing, and
deployment. Activities include applying statistical and machine learning algorithms, developing
predictive models, visualizing insights, and refining models based on performance. For example, data
scientists use watsonx.ai to analyze customer behavior. Finally, the infuse stage integrates AI into daily
business operations to enhance decision-making and automate tasks. IBM watsonx APIs, RPA, and Cloud
Pak for automation are helpful in embedding AI models, automating processes, monitoring models in
real time, and training employees in AI utilization. For instance, AI insights were integrated into
customer relationship management or CRM systems to automate marketing campaigns and optimize
inventory management. The IBM AI ladder framework is versatile. It transforms data into actionable
insights and strategies, supporting industries such as healthcare, finance, manufacturing, and retail. In
the current business landscape, where AI is becoming increasingly prevalent, company leaders need to
shift their mindset and move away from the +AI model, where AI is used as a supplementary tool, and
instead embrace the AI+ approach, where AI is integrated holistically. This means they need to move
from merely incorporating AI into their existing workflows, which only provide limited, incremental
advantages to a more comprehensive approach, where AI is a core part of the business, not just a
supporting tool. This involves designing business processes with AI as a fundamental component from
the outset, rather than as an afterthought or add-on. Let us understand the shift in detail. The AI+
approach works towards a holistic integration process. This means that AI should be seamlessly
integrated across all business functions. It involves embedding AI into product innovation, enhancing
operational efficiency, and fostering a culture that embraces AI-driven decision making. The goal is to
make AI a core part of the organization's DNA, influencing every aspect of the business from strategy to
execution. Then the focus should be on identifying high value AI use cases that can deliver significant
returns on investment. Begin with projects with clear, measurable outcomes, and the potential for
substantial impact. Once these initial projects demonstrate success, scale them across the organization.
This approach ensures that AI investments are strategically aligned with business goals and provide
tangible benefits. Another crucial step is selecting the appropriate AI technologies and architectures for
successful integration. This involves evaluating different AI tools and platforms to ensure they align with
the businesses specific needs and goals. It is essential for the right technology to be scalable and flexible.
Further, a robust data foundation is essential for effective AI implementation. This includes ensuring
that the data is trustworthy, transparent, and well-governed. High quality data is critical for training AI
models accurately and making reliable predictions. Additionally, organizations should comply with data
privacy and security regulations to avoid legal and ethical issues. Finally, it is important for organizations
to continuously innovate by developing new AI applications and modernizing existing ones. Leveraging a
hybrid Cloud platform can provide the necessary flexibility and scalability to support ongoing AI
initiatives. Continuous improvement and adaptation ensure that AI solutions remain relevant and
effective in addressing evolving business challenges. In this video, you learned about the IBM AI ladder,
a framework to adopt AI that involves the four key stages of collect, make data simple and accessible,
organize, create a business ready analytics foundation, analyze, build and scale AI with trust and
transparency, and infuse, operationalize AI throughout the business. You also gained insights into
adopting an AI+ approach that involves holistic integration, identifying use cases, choosing the right AI
technology, robust data foundation, and continuous innovation and modernization.

Transformer votre travail grâce aux

outils de l'IA
Welcome to transforming your work through AI tools. After watching this video, you will be able to list
the key AI tools utilized by various professionals and demonstrate their practical applications. In today's
digital age, AI tools and platforms play vital roles across multiple industries. A report published by
Statista shows a huge rise in global AI tool users. It is expected to reach over 700 million by the end of
the decade. Professionals use AI tools to work smarter and be more creative. By using the tools, they
stay ahead of the curve and unlock new growth opportunities. Popular tools like Microsoft Copilot and
Google Gemini help professionals with tasks such as answering questions and generating text. Let's
explore how professionals across various fields utilize these AI tools in their work. Meet Mark. He is a
research analyst at a finance startup who is responsible for investigating the ongoing market trends, and
for that, he uses Microsoft Edge. Now, to grasp concepts quickly and save his time, he uses Microsoft
Copilot, a powerful AI tool that integrates seamlessly with daily apps like Microsoft Edge, Word, Excel,
PowerPoint, and Outlook. Copilot helps Mark by summarizing the article's key points, defining unfamiliar
terms, and eventually streamlining his tasks. Let's also look at a practical application of Google Gemini.
Vijay, a healthcare professional, manages patient records and contact details. He relies on Gemini to
create a standardized patient intake form template. By having a consistent template, Vijay can ensure
that all necessary information is captured uniformly from each patient. This helps him in maintaining
accuracy and efficiency in work. Tools like Microsoft Copilot and Google Gemini are valuable for people
other than research analysts or healthcare professionals. They are also valuable for content creators,
business operations professionals, customer service agents, and others. Let's look at some more
examples. Marketers, content creators, and editors harness AI tools to streamline processes, generate
innovative content, and achieve marketing goals. Sarah manages the social media team at a tech
startup. To create engaging videos, her team relies on Synthesia. This AI tool helps her team to generate
digital avatars, simplifying video production without requiring extensive technical skills. For crafting
compelling content, her team uses QuillBot, an AI-powered tool that refines and condenses text,
ensuring originality and coherence in their social media posts. Finally, to polish the final drafts,
Grammarly comes into play. This AI-based tool provides grammar and style checking, ensuring the
content is clear, correct, and stylistically consistent. Don't these AI tools simplify her team's workload?
They streamline processes by producing impactful content and effectively engaging the audience online.
Then language learners and translators employ diverse AI tools to personalize their learning experiences
and overcome language barriers. Meet James, an international business consultant who frequently
travels and communicates with clients from diverse linguistic backgrounds. To excel in his role, James
relies on a range of AI tools tailored for language learning and translation. James starts by refining his
language skills using Duolingo's AI-driven platform. The personalized lessons cater to his needs, focusing
on work-specific vocabulary and phrases essential for presentations. During client calls, James relies on
Google Translate to accurately translate messages in real time. This AI tool ensures seamless
communication across languages, allowing him to understand client requirements. By integrating these
AI tools into his professional routine, James enhances his ability to communicate effectively with
international clients. Customer support representatives utilize AI tools to enhance their efficiency and
transform their workflow. Meet Bella, a customer support representative. During her shift, she handles a
high volume of inquiries. AI-powered chat bots instantly generate accurate responses, resolving
common issues like account inquiries or product troubleshooting. This not only saves time, but also
ensures customer satisfaction. Bella also uses Zendesk's AI capabilities which suggests pertinent
knowledge base articles and solutions based on customer inquiries. This streamlines issue resolution,
minimizes escalation needs, and enhances response times effectively. For more personalized
interactions, Bella also uses LivePerson's AI-powered messaging with sentiment analysis. This tool
analyzes customer messages to evaluate emotions and tailor responses accordingly. Data professionals
can also use AI tools for data analysis and visualization. Emily is a data analyst who uses Tableau for
intuitive data visualization. With natural language queries, she quickly creates interactive dashboards,
empowering her team with actionable insights. David, a data scientist, uses Microsoft Power BI to
analyze client data trends and create clear reports and models. These visualizations help him understand
client demographics and resource needs. HR professionals use AI tools to streamline recruitment,
automate tasks, and enhance employee engagement efficiently. Miranda is an HR manager in a tech
company. Her organization is struggling with challenges like low employee engagement and recruitment
pressures. She implements Culture Amp, an AI platform that provides insights into employee
engagement, helping to address issues and improve workplace culture. Such AI platforms often
automate the initial support process and alert management to intervene and provide personal
interactions. For recruitment pressures, Miranda leverages IBM watsonx Orchestrate. Orchestrate
handles interview scheduling and streamlines communications, working tirelessly to ensure that the
recruitment process runs smoothly. These tools help Miranda and her team overcome the challenges
and ease the pressure. In this video, you explored how professionals from different domains leverage AI
tools and platforms to enhance productivity and efficiency. Microsoft Copilot and Gemini provide varied
applications, while Synthesia, Grammarly, and QuillBot aid editors, marketers, and content creators.
Duolingo and Google Translate enable language learning and translation. AI chatbots, Zendesk, and
LivePerson enhance customer service. Tableau and Power BI are used for data analysis and visualization.
AI tools like Culture Amp and IBM watsonx Orchestrate are used by HR professionals to improve
employee engagement and streamline recruitment.

Possibilités de carrière dans l'IA

Welcome to career opportunities with AI. After watching this video, you will be able to identify
key AI-related roles across various industries and recognize the foundational skills necessary to
pursue a career in AI. You will also explore how you can effectively switch your career to AI. Are
you worried that AI might take away your job? While it's true that AI will change the job
landscape by automating multiple tasks, its also important to understand that it will also create
new opportunities. Throughout history, technological advancements have transformed
industries and the job market. Remember the industrial revolution which did replace some
manual labor jobs but also created new industries that didnt exist before? Additionally, in the
information age, people feared that computers would replace human jobs and lead to
widespread unemployment. On the contrary, it created job opportunities like IT support and
web development and opened new avenues. AI is also creating a plethora of career
opportunities that were unimaginable a few years ago. From healthcare to finance,
entertainment to education, the demand for AI professionals is increasing. Let us look at some
of the most sought after AI careers such as AI engineer, data scientist, robotics engineer, NLP
engineer, AI application developer, and AI research scientist. AI engineers specialize in the
design and maintenance of AI systems, leveraging expertise in machine learning, neural
networks, and programming languages like Python and Java. Similarly, data scientists are
essential for analyzing huge datasets using advanced mathematics, statistics, data visualization
tools. And languages such as Python and SQL to develop AI models. Meanwhile, robotics
engineers design and build robots. For this role, you need skills in robotics, machine learning,
mechanical engineering, and programming. Then there are NLP engineers who create AI
systems that process human language. They possess expertise in machine learning, neural
networks, and programming languages like Python and Java. AI application developers are
responsible for developing software applications that incorporate AI. They need expertise in
programming languages, AI frameworks, data analysis, APIs, and software development.
Lisez la vidéo à partir de :2:31 et suivez la transcription2:31
AI research scientists advance AI through research by developing new algorithms and models.
So if you're fascinated by machine learning, neural networks, and NLP, you can consider
becoming an AI researcher. This role demands a strong research background. You can find
these tech savvy experts in industries like technology, finance, healthcare, automobile research,
and many more. Even if you don't have strong programming skills, there are plenty of emerging
AI roles in various domains for you. For example, roles in the areas such as ethics, management,
communication, and application of AI technologies. So there is an array of dynamic and in
demand job roles across many industries that don't require highly technical or programming
experience. Making them accessible for those with varied backgrounds. For example, an AI
ethicist needs to ensure that AI systems are developed and used ethically, considering social
and moral implications. So if you're someone with a background in ethics, philosophy or law
and have strong communication skills, this could be for you. AI product managers oversee the
development and implementation of AI products, ensuring they meet market needs. You can
consider this role if you are good at project management, business analysis and understanding
of AI technologies. AI strategists develop long-term strategies for implementing AI to achieve
business goals. For this role, you need strategic planning, business acumen, and a good
understanding of AI applications. AI marketing specialists utilize AI to analyze consumer data
and optimize marketing strategies. If you're good at data analysis, have marketing knowledge
and know your way around AI tools, this could be your niche. You can find these experts
working in industries like technology, marketing, media, consumer goods, healthcare and
consulting. Now, if you are thinking about switching your career to AI, first identify your existing
skills that can be leveraged. For example, problem solving and analytical thinking or
programming experience if you have any. Then gather a detailed understanding of core AI
concepts such as machine learning, deep learning, and natural language processing. Note that
Python is the most widely used language in AI. You can then contribute to open source AI
projects on platforms like GitHub or develop small AI projects to solidify your learnings. Since
the field of AI is constantly evolving, you need to stay updated. For that, you can get into a
network with AI professionals, attend conferences, or read publications on the latest trends. As
you gain experience, you can delve deeper into specialized areas of AI that align with your
interests like robotics or NLP. Let's explore the examples of inspiring individuals who
successfully transitioned their careers into AI. Emma leveraged her research and analytical skills
from law to switch to AI for the legal field. She now helps develop AI systems that can analyze
legal documents to support lawyers in their work. Meet Sarah, she leveraged her strong
communication and customer service skills from marketing to transition into AI. She also
acquired expertise in NLP, machine learning and python from various courses. And now builds
chatbots powered by AI that can answer customer queries and provide support on company
websites. In this video, you learned about some of the most sought after AI careers, which
include AI engineer, data scientist, robotics engineer, NLP engineer, AI application developer,
and AI research scientist. You also learned that some AI related jobs do not require the
knowledge of programming languages like AI ethicist, AI product manager, AI strategist, and AI
marketing specialist. To switch to a career in an AI related field, you should identify your
existing skills that can be leveraged, learn core AI concepts, apply the skills practically, stay
updated with the latest trends, and finally delve deeper into specialized areas of AI.

Humains contre IA : qui doit prendre la

décision ?
A decision needs to be made. But who should make it. Me a human,
Lisez la vidéo à partir de ::21 et suivez la transcription0:21
or an artificial intelligence, and AI? We've discussed before that humans can outperform AI at
some tasks, but that statistically, AI will make a better job of deciding for other tasks. So for one
single decision, who should decide? Well, the answer is a fascinating combination of holistic
curves and human bias. Let's get into it. Consider a fraud detection system. Fraud detection.
Lisez la vidéo à partir de ::59 et suivez la transcription0:59
The system generates the alerts of potentially fraudulent transactions. Financial analysts review
each alert. Now, there's thousands of events generated each day and the analysts are
overwhelmed with 90% of those alerts being false positives. An AI system could help alleviate
the workload, but which alerts should the AI handle and which should be processed by a skilled
financial analyst? Well, let's draw a graph to answer the question, is this a real alert?
Lisez la vidéo à partir de :1:47 et suivez la transcription1:47
Let's draw a graph of an x and a y axis. The y axis tracks the success rate. An alert comes in, we
make a prediction as to if it is real or not, and we track if that prediction turn out to be right.
Along the X axis is the confidence score.
Lisez la vidéo à partir de :2:16 et suivez la transcription2:16
A confidence score of 0% says a prediction thinks that this is definitely not a real alert. It's a
false positive. Confidence score of 100% means that a prediction is certain that it is a real alert.
Now, a typical AI performance curve will look something like this.
Lisez la vidéo à partir de :2:50 et suivez la transcription2:50
We've got very low confidence scores. This is not a real alert, and very high confidence scores.
This is a real alert. They're correlated to a high success rate. That's these areas up here. When
the AI is not sure about a given prediction, then it's not such a case. Lower success rate when
the AI is not sure. It's effectively the AI algorithm saying, I don't know. Now, human
performance curves are typically a little bit flatter than that. So the human performance curve
might look something like this. Often not quite as accurate as a very confident AI algorithm, but
a little better at making the right decision when the AI is unsure. At a 50% confidence level, a
human is likely to do a better job than an AI. Now, why is that? Well, when an AI is certain of
itself, it's highly performant and beats out humans who can lose consistency and focus on
attention. AIs, they don't get distracted. But on the other hand, when an AI is unsure, often for
cases that are complex or statistically rare, humans can outperform an AI prediction by bringing
in additional information and context. They can look stuff up or ask a colleague, whereas the AI
sticks to its same old decision logic and information. When a new alert comes in, if the AI
assigns a high or a low confidence level, then chances are that statistically speaking, it will do a
better job of deriving if that alert is real or a false positive, than a given financial analyst. But
this is not a zero sum game. It doesn't have to be AI or human. We have one more option.
Augmented. Augmented intelligence combines both a human decision aided by AI. This
performance curve falls somewhere between the two. For somewhat low and for somewhat
high confidence scores which make up a significant number of predictions, it's augmented
intelligence that will have the highest success rate. Except. For augmented intelligence to be
most effective, we need to account for the messy business of human cognitive bias.
Lisez la vidéo à partir de :5:42 et suivez la transcription5:42
We're not always great at doing what we're told. It turns out that how we present information
from an AI algorithm to a human decision maker, has a significant influence on how effectively
that information is used. To illustrate that, let's consider forced display versus optional display.
Lisez la vidéo à partir de :6:16 et suivez la transcription6:16
A forced display simultaneously displays an AI recommendation along with a given decision
case. For every fraud decision alert that I need to make a decision about, I as the analyst, also
see the AI's recommendation. And this can lead to something called automation bias, which is
the propensity for humans to favor suggestions from automated decision making systems and
to ignore contradictory information. Effectively, the human decision maker is saying, the AI
knows best, and going with the AI prediction at the expense of their own judgment. Optional
display means the AI recommendation is only shown to the human decision maker when they
request it. A person sees a decision case and can then ask the AI to reveal its recommendation.
This overcomes automation bias by giving a person time to consider the case for themselves
before consulting an AI recommendation. The human is not overwhelmingly influenced by what
the AI thinks, because they've had a chance to make up their own first impression. Then there's
the whole issue of trust too. When an AI recommendation is accompanied by an accuracy
percentage which indicates how likely this prediction is to be correct, humans are less likely to
incorporate the AI recommendation into their decision regardless of the accuracy percentage
being displayed. Basically, we don't like recommendations that openly tell us that they might be
wrong. We've seen that who should make a decision, a human, an AI or a human assisted by an
AI recommendation is something that we can derive. We can move from subjective decisions to
the quantifiable that for a given decision, who the most effective decision maker is likely to be.
When the most effective decision maker is a combination of AI in human, that's augmented
intelligence, we must consider presentation of that augmentation to minimize human cognitive
bias in the decision making process. Brought together, us humans and AI algorithms make a
pretty powerful team. We can improve decision making outcomes if we just know who to ask.
Resume :
Nous vous félicitons ! Vous avez terminé ce module. À ce stade du module, vous savez ce qui
suit :

Les agents en IA sont des programmes logiciels qui s'engagent dans l'environnement, traitent
des données et exécutent des tâches de manière autonome pour atteindre des objectifs définis
par l'homme.
La robotique implique la conception, la construction et l'utilisation de robots pour effectuer des
tâches avec ou sans l'aide de l'homme.

Les cobots collaborent avec les humains en utilisant des capteurs avancés et l'IA, en
communiquant et en coordonnant les tâches qui nécessitent un travail d'équipe.

L'automatisation des processus robotiques (RPA) automatise les tâches répétitives et permet de
créer, d'utiliser et de contrôler des robots virtuels.

L'IA automatise les tâches répétitives pour les entreprises, comme la saisie de données, la
planification et la génération de rapports, et permet aux employés de se concentrer davantage
sur le travail créatif et stratégique.

L'IA permet aux entreprises d'analyser rapidement de grands volumes de données en

identifiant des modèles et des tendances, ce qui conduit à moins d'erreurs et à de meilleures
prédictions.

L'IA génère des idées innovantes et surmonte les blocages créatifs en aidant les entreprises à
entrer en résonance avec le public cible, ce qui se traduit par un engagement et des ventes plus
élevés.

Les premières IA utilisaient des systèmes basés sur des règles mais manquaient de créativité.
Des décennies plus tard, le Deep learning a permis à l'IA d'apprendre et de s'adapter comme le
cerveau humain.

Les réseaux antagonistes génératifs (GAN) ont marqué une percée, générant des images, de la
musique, du design et de la médecine de haute qualité. Ils ont fait progresser l'IA générative
dans les domaines de la musique, de la conception et de la médecine.

Principaux domaines dans lesquels l'IA générative apporte sa contribution :

Génération de contenu : Analyse du contenu existant et des données relatives au public cible
pour créer un contenu unique

Analyse de données : Analyse d'ensembles de données complexes pour identifier des modèles
et générer des rapports exploitables

Service à la clientèle : Gestion des tâches répétitives, réponses rapides et personnalisées

Développement de produits : Produit de multiples variantes de conception pour un

développement de produits plus rapide et plus innovant

L'IA peut être adoptée dans les entreprises pour accroître l'efficacité, améliorer la prise de
décision, améliorer le service à la clientèle et favoriser l'innovation.

Étapes générales pour adopter l'IA dans les entreprises :

Définir les objectifs de l'entreprise : Identifier les problèmes et fixer des objectifs clairs

Identifier les cas d'utilisation appropriés : Identifier les domaines dans lesquels appliquer l'IA

Préparation des données : Collecter les données, puis les nettoyer et les organiser

Renforcer les capacités en matière d'IA : Former les employés et mettre en place la
configuration nécessaire

Déployer des solutions IA : Intégration aux systèmes existants

Surveiller et optimiser les systèmes d'IA : Assurer le bon fonctionnement

L'IA analyse de grandes quantités de données à des vitesses incroyables, révolutionnant la
façon dont nous fonctionnons, innovons et réussissons.

Les outils d'IA peuvent aider les organisations à améliorer leur efficacité en automatisant
l'engagement des employés, le recrutement, l'intégration et la gestion des performances. Les
chatbots IA ont transformé les interactions avec les clients en automatisant les réponses et en
garantissant la cohérence.

Les professionnels de différents domaines tirent parti des outils et plateformes d'IA pour
améliorer la productivité et l'efficacité.

Outils et plateformes d'IA :

ChatGPT et Gemini : fournit des applications polyvalentes

Copy.ai, Jasper et Synthesia : Aident les spécialistes du marketing et les créateurs de contenu

Grammarly et QuillBot : Utilisation pour les tâches de rédaction et de communication

Duolingo, Google Translate et Babel : Permettent l'apprentissage des langues et la traduction

Chatbots IA, Zendesk et LivePerson : Améliorer les services à la clientèle

Tableau et Power BI : Utilisation pour l'analyse et la visualisation des données

GitHub Copilot : aide les développeurs à compléter le code

Todoist, Microsoft To Do et Evernote : aide à la gestion des tâches

Pour passer à une carrière dans un domaine lié à l'IA, vous devez identifier vos compétences
existantes qui peuvent être exploitées, apprendre les concepts de base de l'IA, appliquer les
compétences de manière pratique, rester à jour avec les dernières tendances et, enfin,
approfondir les domaines spécialisés de l'IA. Voici quelques-uns des profils professionnels :

Éthicien de l'IA : Veiller à ce que les systèmes d'IA soient développés et utilisés de manière
éthique, en tenant compte des implications sociales et morales.

Gestionnaires de produits IA : Supervisent le développement et la mise en œuvre des produits

d'IA, en veillant à ce qu'ils répondent aux besoins du marché.

Stratèges en IA : Élaborent des stratégies à long terme pour la mise en œuvre de l'IA afin
d'atteindre les objectifs de l'entreprise.

Spécialistes du marketing de l'IA : Utilisent l'IA pour analyser les données des consommateurs
et optimiser les stratégies marketing.