Generative AI vs. Conventional AI: Introduction For Operational Risk Professionals

Summary

In this informative video, Manos Kulal, co-founder and chief risk officer at Risk Spotlight, introduces generative AI and contrasts it with conventional narrow AI technologies. He explains how generative AI, capable of producing various outputs like text, videos, and images, can enhance risk management in the financial services industry. Kulal highlights the accessibility and affordability of generative AI models, such as OpenAI's chat GPT, which has revolutionized the field by offering advanced AI to businesses of all sizes. The video also emphasizes the importance of prompt engineering to effectively harness AI's expertise and concludes with an offer of a specialized training course on applying generative AI in operational risk management.

Takeaways

• 🧠 Generative AI vs. Narrow AI: The video introduces generative AI as a technology that can produce various formats of output, contrasting it with narrow AI, which is designed for specific use cases.
• 🛡️ Narrow AI Applications: Narrow AI is widely used in financial services for tasks such as credit card fraud detection, cyber risk management, and money laundering detection.
• 🚀 Generative AI Emergence: Generative AI gained prominence with the launch of OpenAI's chat GPT in November 2022, which quickly reached 100 million users.
• 💡 Generative AI Capabilities: This technology can generate outputs in text, videos, images, and audio, offering a wide range of applications in operational risk management and beyond.
• 💼 Business Benefits: Generative AI can enhance productivity and quality of risk management activities in the financial services industry.
• 📈 Adoption and Accessibility: OpenAI's mission to democratize AI has made advanced generative AI models available for free or at a low cost, allowing even small and medium-sized firms to adopt these technologies.
• 📚 Expertise Access: Generative AI models, trained on large internet data, can provide expertise on a wide range of topics, from fitness and nutrition to business strategy and risk management.
• 🔍 Use Cases: The video highlights over 100 operational risk use cases where generative AI models can significantly improve productivity and risk management quality.
• 🛠️ Prompt Engineering: Operational risk professionals are encouraged to learn prompt engineering to effectively utilize AI models and maximize benefits.
• 🎓 Training Opportunities: Risk Spotlight offers a 15-hour training course focused on prompt engineering for operational risk management, available in online or classroom formats.
• 📧 Contact and Resources: Interested parties can reach out to Risk Spotlight for more information on prompt engineering courses and AI services via email or their website.

Q & A

What is the main focus of the video by Manos Kulal?

The video focuses on introducing generative AI, contrasting it with conventional AI technologies, and discussing how generative AI can enhance risk management activities in the financial services industry.

What is the term 'narrow AI' referring to in the context of the video?

Narrow AI refers to AI technologies built for specific use cases and is what is commonly referred to as AI in business discussions, as these technologies have been widely implemented across many financial services firms for around 10 to 15 years.

Can you provide examples of narrow AI implementation in operational risk management?

Examples include credit card fraud detection, detecting malicious transactions on IT networks to manage cyber risks, and detecting money laundering transactions.

What are some first-line use cases of narrow AI technologies mentioned in the video?

First-line use cases include marketing teams predicting customer churn, AI-based algorithmic trading, and credit scoring models based on unstructured data such as social media content.

Why are generative AI technologies labeled as 'generative'?

Generative AI technologies are labeled as such because they can generate outputs in various formats such as text, videos, images, and audio.

What significant event in November 2022 brought the term 'generative AI' into widespread recognition?

The term 'generative AI' became widely known when OpenAI launched ChatGBT, which signed up 100 million users in the first two months, making it the fastest adopted technology of all time.

How does generative AI differ from narrow AI in terms of accessibility and cost?

Generative AI models, such as those provided by OpenAI, are available for free or at a low cost of $20 to $30 per user per month, making them accessible to even small and medium-sized firms, unlike narrow AI technologies which are resource-intensive.

What is the key benefit of generative AI technologies mentioned in the video?

A key benefit of generative AI technologies is that they can provide access to expertise on a wide range of topics due to their training on large amounts of data available on the internet.

Can you provide examples of how generative AI can be utilized in operational risk management?

Examples include identifying risks related to a business process, identifying controls to mitigate operational risks, and building detailed cyber risk scenarios.

What is 'prompt engineering' and why is it important for operational risk professionals?

Prompt engineering is the skill of writing effective prompts to maximize the benefits gained from AI models. It is important for operational risk professionals to learn this skill to utilize AI models for real operational risk management cases effectively.

What training does Risk Spotlight offer to help professionals learn prompt engineering?

Risk Spotlight offers a 15-hour training course focused on generative AI and operational risk management, which includes learning how to apply generative AI in the context of operational risk management and writing effective prompts.

Outlines

00:00

🧠 Introduction to Generative AI and Its Impact on Risk Management

Manos Kulal, co-founder and chief risk officer at Risk Spotlight, introduces the concept of generative AI and distinguishes it from conventional AI technologies. He explains that generative AI can produce outputs in various formats like text, videos, images, and audio, and has been rapidly adopted since the launch of OpenAI's chatbot in November 2022. Kulal contrasts this with narrow AI, which is limited to specific use cases and requires significant resources, making it accessible mainly to larger firms. Generative AI, on the other hand, has the potential to enhance productivity and quality in risk management across the financial services industry, including operational risk management and first-line use cases such as marketing and trading.

05:00

📚 Harnessing Generative AI for Expertise and Risk Management

This paragraph delves into the wide range of expertise generative AI can provide, from fitness and nutrition to business strategy and operational risk management. Kulal highlights that Risk Spotlight's AI practice has analyzed over 100 operational risk use cases where generative AI can significantly improve productivity and quality. He also emphasizes the importance of 'prompt engineering' – the skill of writing effective prompts to maximize the benefits from AI models. Kulal demonstrates the difference between a simple prompt and an effective one, showing how the latter can yield more specific and useful results for operational risk identification in a retail banking context.

10:02

🎓 Training in Prompt Engineering for Operational Risk Professionals

The final paragraph focuses on the need for operational risk professionals to learn prompt engineering to effectively utilize generative AI models. Kulal introduces a 15-hour training course developed by Risk Spotlight, which is the world's first course on generative AI for operational risk management. The course aims to teach key concepts of generative AI and its application in operational risk management, with exercises that can be customized to the specific needs of an organization. Kulal invites interested parties to explore the course and other AI services offered by Risk Spotlight by reaching out via email or visiting their website.

Keywords

💡Generative AI

Generative AI refers to artificial intelligence models that can create new content, such as text, images, videos, and audio, in various formats. In the context of the video, generative AI is contrasted with narrow AI and is highlighted as a transformative technology that can significantly enhance productivity and quality in risk management activities. The video mentions generative AI's ability to generate outputs for a wide range of topics, making it a versatile tool for various applications in the financial services industry.

💡Narrow AI

Narrow AI, also known as weak AI, is designed for specific tasks and use cases. The video script uses the term 'narrow' to describe AI technologies that have been prevalent for the last 10 to 15 years and are widely implemented across financial services firms. Examples given in the script include credit card fraud detection, cyber risk management, and money laundering detection, which are all applications of narrow AI to manage operational risks.

💡Operational Risk Management

Operational risk management is the process of identifying, assessing, and controlling risks that arise from an organization's operations. The video emphasizes how both narrow AI and generative AI technologies can be utilized in operational risk management. For instance, narrow AI is used for detecting fraud and malicious transactions, while generative AI can help in identifying risks related to business processes and building detailed cyber risk scenarios.

💡Financial Services

Financial services encompass a broad range of businesses that manage money, including banks, credit unions, credit card companies, insurance companies, and stock brokerages. The video discusses how AI technologies, both narrow and generative, have been implemented in financial services firms to manage various types of risks and improve business operations.

💡OpenAI

OpenAI is a research organization that aims to develop and promote friendly artificial intelligence. The video script highlights OpenAI's launch of ChatGBT in November 2022, which rapidly gained 100 million users, making it the fastest adopted technology ever created. OpenAI's mission to democratize access to advanced AI technologies is underscored by its decision to offer its generative AI models for free or at a low cost, enabling even small and medium-sized firms to adopt these technologies.

💡Prompt Engineering

Prompt engineering is a skill that involves crafting effective prompts or questions to guide AI models to produce desired outputs. The video demonstrates the importance of prompt engineering in maximizing the benefits of generative AI models. It shows how a simple prompt can yield generic results, whereas a more structured and detailed prompt can lead to specific and useful information, as illustrated in the examples provided for identifying operational risks.

💡Risk Spotlight

Risk Spotlight is the company co-founded by Manos Kulal, who is also the Chief Risk Officer. The company specializes in AI practice, focusing on generative AI technologies and their applications in enhancing productivity and quality in risk management activities within the financial services industry. The video script mentions Risk Spotlight's offerings, including a training course on prompt engineering and other services to help organizations leverage the benefits of generative AI.

💡ChatGBT

ChatGBT, launched by OpenAI, is an example of a generative AI model that can generate human-like text based on the prompts given to it. The video script uses ChatGBT to demonstrate the capabilities of generative AI, showing how it can be used to identify operational risks and provide business benefits when given effective prompts.

💡AI Adoption

AI adoption refers to the process of implementing and integrating AI technologies into business operations. The video discusses the rapid adoption of generative AI, particularly ChatGBT, and how it has been made accessible to everyone, even small and medium-sized firms, due to OpenAI's mission and pricing strategy. This has driven the widespread use of AI technologies in various applications, including operational risk management.

💡Expertise

In the context of the video, expertise refers to the knowledge and understanding that generative AI models can provide on a wide range of topics due to their training on large amounts of data available on the internet. The video highlights how generative AI can offer expertise in areas such as fitness, nutrition, business strategy, and operational risk management, thereby providing valuable insights and support to users across different domains.

Highlights

Introduction to generative AI and its differentiation from conventional AI technologies.

Generative AI enhances productivity and quality of risk management activities in the financial services industry.

Narrow AI is built for specific use cases and has been widely implemented in financial services firms for over a decade.

Examples of narrow AI implementation in operational risk management include credit card fraud detection and cyber risk management.

Narrow AI requires extensive data, significant infrastructure, and a team of AI professionals, making it resource-intensive.

Generative AI can generate outputs in various formats such as text, videos, images, and audio.

Generative AI became widely known with the launch of OpenAI's chat GPT, which gained 100 million users in its first two months.

OpenAI's mission is to provide advanced AI technologies to counter the influence of large technology firms.

Generative AI models are available for free or a low monthly cost, making them accessible to small and medium-sized firms.

Generative AI provides access to expertise on a wide range of topics due to training on large amounts of internet data.

Risk Spotlight's AI practice has analyzed over 100 operational risk use cases where generative AI can provide significant benefits.

Generative AI can be utilized in first-line use cases such as marketing and wealth management advice.

The use of generative AI models can be extended beyond operational risk to other risk management and compliance topics.

Operational risk professionals need to learn prompt engineering to write effective prompts for AI models.

Demonstration of the difference between a simple prompt and an effective prompt for identifying operational risks.

Risk Spotlight offers a 15-hour training course on prompt engineering for operational risk management.

The course covers key topics on generative AI and its application in operational risk management.

Risk Spotlight can deliver the course in online or classroom format and customize it for specific organizational needs.

Transcripts

in this video I will cover a brief

introduction of generative AI

Technologies and how these are different

to Conventional AI

Technologies clearly differentiating

between these two technologies can

facilitate operational risk

professionals to have more thoughtful

and insightful conversations with the

business

stakeholders I'm Manos kulal and I'm

co-founder and chief risk officer at

risk Spotlight at risk Spotlight are AI

practice specializes in generative AI

Technologies and how these can enhance

the productivity and quality of risk

management activities in financial

services

industry for this video I'm going to

contrast generative AI with narrow AI

first let me cover narrow AI

Technologies narrow AI Technologies are

built for specific use cases and hence

I'm referring to them with the label

narrow

in our common business discussions when

someone refers to AI this is typically

what they are referring to this is

mainly because these AI technologies

have been with us for the last 10 to 15

years and have been widely implemented

across many Financial Services

firms example of implementation of

narrow AI Technologies in operational

risk management include credit card

fraud detection detecting malicious

transactions on it Network to manage

cyber risks and detecting money

laundering

transactions all of these are examples

of narrow AI Technologies being used as

part of controls to manage operational

risks examples of implementation of

narrow AI Technologies in firstline use

cases include marketing team predicting

customer Chun AI based algorithmic

trading and credit scoring models based

on unru structured data such as social

media content of a

customer developing and implementing

these Technologies is a major

undertaking they need extensive data for

training significant it infrastructure

for execution and a team of seasoned Ai

and data professionals this makes them

resource intensive often limiting their

use to larger firms small and

medium-sized firms typically do not have

the resources required to take benefits

of these

Technologies now let me cover the key

aspects of the generative AI

Technologies these Technologies are

based on AI models that can generate

output in various formats and hence the

label

generative they can generate outputs in

various formats such as text videos

images and

audio the term generative AI became

widely known in November 2022 when open

AI launched chat gbt and it signed up

100 million users in the first two

months this made it the fastest adopted

technology of all time ever created by

humans the adoption was faster than

mobile phones social media and even the

internet examples of implementation of

gen Technologies in operational risk

management include identifying risks

related to a business process

identifying controls to mitigate an

operational risk and build detailed

cyber risk

scenarios examples of implementation of

gen AI Technologies in firstline use

cases include marketing team

brainstorming ideas for developing

advertising content chatbot to provide

wealth management advice to clients and

it team utilizing gen AI to write

code the purpose of founding openi was

to develop and provide Advanced AI

Technologies to everyone to counter the

influence of large technology firms

holding the benefits of AI

Technologies due to this unique Mission

openai has made its Advanced generative

AI models available for free and the

premium version of the model is

available only for $ 20 to $30 per user

per

month when was the last time you saw a

transformative technology made available

to everyone in the world for free or for

$ 20 to $30 per month due to this now

even small and medium-sized firms can

adopt these AI Technologies and drive

tremendous productivity

benefits a key benefit of generative AI

Technologies is that it can provide

access to expertise on wide range of

topics the Gen models are trained on

large amounts of data available on the

inter

internet due to this they can provide

access to expertise on wide range of

topics such as Fitness Nutrition

business strategy they can be Math's

tutor for your kids and be your travel

guide when you go on

holiday for risk management these models

also provide access to expertise on

credit Market strategy reputational risk

and of course they also provide

expertise on operational risk management

our AI practice has analyzed over 100

operational risk use cases where gen AI

models can provide significant

productivity benefits and enhance the

quality of risk management a sample of

use cases where gen AI can be utilized

for the first line are highlighted

[Music]

here

[Music]

a sample of use cases where generative

AI models can be utilized by the second

line are highlighted

[Music]

here

[Music]

[Music]

since generative AI models are defined

to work with text information the use of

these models can also be extended Beyond

operational risk to many other risk

management and compliance related topics

highlighted

here while generative AI models provide

access to wide range of expertise you

can only get access to these benefits if

you utilize effective prompts a prompt

is a question you ask the AI model to

gain access to the model's expertise all

operational risk professional

now need to learn a new skill called

prompt engineering which involves

writing effective prompts to maximize

the benefits you can gain from the AI

models in the example here I have asked

a question to chat gbt to provide three

key business benefits of managing

operational risks in a financial

services firm you can see the responses

provided by chat G to this

question this is an example of an easy

prone however if you want to utilize the

AI models for real operational risk

management cases then you need to write

a more structured

prompt let me demonstrate an example of

identifying operational risks for a

business process using a simple prompt

versus an effective prompt this will

allow you to see the benefits of

learning prompt engineering I'm logged

into chat gbt now and let me show you a

simple prompt for risk identification

I will paste The Prompt text

here in this prompt I'm asking for list

of five operational risks for a business

process called account opening and

onboarding I'm providing details of the

process such as the process title the

process description and list of

activities performed as part of this

business

process let me submit this PR so I can

show you the output of this

prompt so here you can see chat gbt has

given list of five operational risks but

these are not very useful these risks

are very high level very generic like

data security and privacy compliance and

Regulatory so now let me demonstrate the

same example but this time with prompt

Engineering in this prompt I'm

requesting the risks to be identified

for the same business process however

I'm being very specific with my prompt

and providing detailed instructions to

the AI model to ensure that it provides

me with a high quality output I'm giving

the model A Persona of an expert for

identifying operational risks related to

business processes in the retail bank

banking industry I'm explicitly

specifying the risk categories for which

I want it to identify the operational

risks I'm requesting the output to be in

the table format so I can easily copy

and paste it in Excel and I'm requesting

four columns in the table the risk title

column the risk description column and

some banking examples in this case I'm

asking for two banking examples Les for

each risk and the risk category column

for the risk title column I'm specifying

that the title should be 8 to 12 words

long and it should not provide any

generic risks like geopolitical risk or

vendor dependency risks in its output

I'm also specifying that every risk

title should contain a verb which should

highlight the main event that would

occur as part of that particular

risk so so let me submit this prompt so

you can see the output based on this

prompt so here you can see the output is

now in a table format with the four

columns I

requested and you can see the title of

the risk is a lot more specific now than

what we were getting in our previous

prompt which was without prompt

engineering so I have a very specific

title which is then related to the

business process I have a description

and then I have two banking examples of

how this risk could occur in a retail

bank and because I've asked for risks

for three risk categories it's also then

specifying which risk category each risk

relates

to hopefully this demonstrates the

benefit of prompt engineering to you you

can see that the quality of output that

has been produced by prom prompt

engineering is significantly better than

the output we saw from a generic

prompt to help operational risk

professionals learn the new skill of

prompt engineering we have developed a

15-hour training course you can find the

details of this course on the training

page on our website risks spotlight.com

let me show you the details of this

course

this is world's first generative AI

course focusing on operational risk

management the course covers these key

topics you and your team need to learn

on key generative AI

conent the course will focus on these

operational riskmanagement use cases to

learn examples of how to apply

generative AI in context of operational

risk

management

we can deliver this course in online

format or classroom format and also

customize it to make the exercises

relevant for the operational risk

content of your

organization that's all I wanted to

cover in this video If you interested in

exploring more about our prompt

engineering course and other services we

can provide to facilitate the benefits

of generative AI in your organization

then please email us at AI at risks

spotlight.com you can also find more

details about our AI offerings at

www. risks spotlight.com thank you for

your time and attention