LIVE | Microsoft CEO Satya Nadella's Keynote Address On The Potential Of Next-generation AI | N18L

Summary

El script aborda la revolución en la tecnología de la información y la gestión, con la introducción de un nuevo motor de razonamiento neural. Se enfatiza la importancia de la adopción de productos como los co-pilotos y la necesidad de integrar tecnologías AI en todos los sectores de la economía para impulsar el crecimiento. Se destaca la misión de Microsoft de empoderar a cada persona y organización en India, así como la iniciativa de capacitación en habilidades de AI para más de 2 millones de personas. Además, se mencionan ejemplos inspiradores de cómo la AI está transformando la vida en India, desde el sector tecnológico hasta la agricultura y la movilidad social.

Takeaways

• 🌐 La nueva capacidad de razonamiento neural permite una mejor comprensión del mundo digital y revoluciona la pila tecnológica.
• 📈 La inversión en tecnologías como la inteligencia artificial tiene un impacto directo en el PIB y el crecimiento económico de un país.
• 🚀 Las empresas Microsoft se centran en empoderar a personas y organizaciones en India para participar plenamente en la era de la inteligencia artificial.
• 🛠️ La adopción de productos como los co-pilotos de IA es fundamental para mejorar la productividad y romper los silos de especialización.
• 🔄 La transformación digital a través de la nube y la IA es la misma, y es la manera en que Microsoft se comunica con sus clientes y optimiza sus procesos empresariales.
• 💡 La implementación de IA en empresas como Axis Bank y HCL Tech muestra cómo se están adaptando sus flujos de trabajo y mejorando la eficiencia.
• 🌟 La infraestructura de Microsoft, desde el nivel de la infraestructura hasta los modelos de aprendizaje automático, está diseñada para apoyar la creación de productos de IA por parte de sus clientes.
• 🔧 Microsoft expone su pila tecnológica a los usuarios para que puedan construir sus propios productos de IA, aprovechando los servicios de Azure y otros recursos.
• 🔐 La seguridad y la confianza son fundamentales en la era de la IA, y Microsoft ofrece soluciones para proteger los datos y garantizar la conformidad con marcos éticos de IA responsables.
• 📚 La capacitación en habilidades de IA es esencial para el éxito en la nueva era tecnológica, y Microsoft tiene la iniciativa de capacitar a más de 2 millones de personas en India.
• 🌍 La capacidad de resolver problemas en India demuestra el potencial global de las soluciones de IA, ya que si se puede hacer en India, se puede hacer en el mundo.

Q & A

¿Qué implica la nueva capacidad de razonamiento neural para el mundo digital actual?

La nueva capacidad de razonamiento neural representa una herramienta avanzada para interpretar y dar sentido al mundo digital, permitiendo una comprensión más profunda y efectiva de los datos digitizados, lo que a su vez puede revolucionar la estructura tecnológica existente y tener un impacto significativo en el PIB de una economía.

¿Cómo se relaciona la inversión en tecnologías generales de propósito con el crecimiento económico de un país?

La inversión intensa y cross-sectoral en nuevas tecnologías generales de propósito, como es el caso de la inteligencia artificial, puede marcar una diferencia significativa en las perspectivas económicas futuras de un país. Un ejemplo histórico es la inversión del 10% del PIB del Reino Unido en el sistema de ferrocarril durante la Revolución Industrial, lo que sentó las bases para su crecimiento futuro.

¿Cuál es la misión de Microsoft en la India en el contexto de la inteligencia artificial?

La misión de Microsoft en la India es empoderar a cada persona y organización para que pueda participar plenamente en la era de la inteligencia artificial, fomentando el crecimiento económico y la innovación en todos los sectores y regiones de la economía india.

¿Qué es un co-piloto en el contexto de la inteligencia artificial y cómo puede afectar la productividad?

Un co-piloto es una herramienta de inteligencia artificial diseñada para complementar las capacidades humanas, facilitando la obtención de experticia a un simple clic y mejorando la productividad en tareas variadas. Su adopción puede cambiar drásticamente la forma en que se realizan las tareas, rompiendo los silos de experticia y acelerando el progreso en áreas como la cadena de suministro, ventas, derecho y más.

¿Qué implicaciones tiene la implementación de co-pilotos en diferentes sectores de la economía india?

La implementación de co-pilotos en diversos sectores de la economía india puede aumentar la velocidad y la eficiencia en la toma de decisiones y en la ejecución de tareas, permitiendo una mayor producción y una mejora en la calidad de la salida de todos los sectores involucrados.

¿Cómo están las empresas en la India adoptando la inteligencia artificial en sus operaciones?

Las empresas en la India están adoptando la inteligencia artificial de manera rápida y amplia, integrándola en sus procesos comerciales, de gestión de proyectos, análisis de datos y más. Empresas como Axis Bank, HCL Tech y M tree están adoptando co-pilotos y adaptándolos a sus flujos de trabajo específicos, lo que demuestra un alto grado de compromiso con la innovación y la mejora continua.

¿Qué es la iniciativa de habilidades en AI que anunció Microsoft para la India y cómo beneficia a la población local?

La iniciativa de habilidades en AI de Microsoft tiene como objetivo capacitar a más de 2 millones de personas en India con habilidades en inteligencia artificial. Esto ayudará a la fuerza laboral local a prepararse y prosperar en la nueva era de la tecnología, creando nuevas oportunidades de empleo y fomentando el crecimiento económico a través de la educación y la capacitación en tecnologías emergentes.

¿Qué es la plataforma kara y cómo está contribuyendo al desarrollo económico en rural India?

Kara es una organización que se basa en la inteligencia artificial para crear soluciones que ayudan a comunidades en el campo de la India. Ofrece trabajos bien remunerados en la etiquetaje de tareas AI a personas en zonas rurales, lo que no solo mejora sus ingresos sino que también contribuye a la creación de oportunidades económicas en áreas que de otro modo podrían estar desatendidas.

¿Qué es la plataforma Jogle Bundi Studio y cómo está siendo utilizada para el empoderamiento social?

Jogle Bundi Studio es una herramienta de codificación de bajo nivel que permite a las personas crear aplicaciones de inteligencia artificial personalizadas. Está siendo utilizada por organizaciones benéficas y sociales para brindar información y servicios a grupos como los trabajadores migrantes, mejorando su acceso a servicios y oportunidades en nuevas localidades y empoderándolos a tomar decisiones informadas.

¿Qué medidas están tomando para garantizar la seguridad y la confianza en la inteligencia artificial?

Para asegurar la seguridad y confianza en la inteligencia artificial, se están implementando medidas como entornos de confianza cero, salvaguardias para proteger los datos y la propiedad intelectual, y un marco de inteligencia artificial responsable que incluye principios, estándares y procesos de auditoría para garantizar que la IA se desarrolle y utilice de manera ética y segura.

¿Cómo están las empresas en la India aprovechando la inteligencia artificial para transformar la experiencia del cliente?

Las empresas en la India están utilizando la inteligencia artificial para crear agentes virtuales que pueden interactuar con los clientes, proporcionándoles asistencia personalizada y procesando transacciones de manera eficiente. Esto mejora significativamente la experiencia del cliente, facilitando la interacción y la planificación de viajes, por ejemplo, en el caso de Air India.

¿Qué es la plataforma de datos y cómo está ayudando a las empresas a mejorar sus operaciones con inteligencia artificial?

La plataforma de datos es una solución que integra el almacenamiento de datos y el cálculo de inteligencia artificial, permitiendo a las empresas acceder a todos sus datos y aplicar el procesamiento de IA en tiempo real. Esto ayuda a las empresas a tomar decisiones más informadas y agilizar sus procesos operativos, mejorando la eficiencia y la toma de decisiones.

Outlines

00:00

🌐 Revolución Tecnológica y su Impacto en el PIB

Este párrafo aborda la evolución hacia la digitalización y cómo la introducción de un nuevo motor de razonamiento neural transforma completamente la pila tecnológica, impactando potencialmente en el PIB. Se destaca la importancia de la inversión intensiva en tecnologías generales y su implementación transversal en la economía para el progreso de un país. Utiliza el ejemplo histórico del Reino Unido invirtiendo en su sistema ferroviario durante la Revolución Industrial para ilustrar cómo la inversión en tecnología impulsa el crecimiento económico. El papel de la inteligencia artificial (IA) en este contexto se resalta como un factor clave para el futuro crecimiento económico, especialmente en mercados de alto crecimiento como India.

05:02

🚀 Transformación Impulsada por la IA y Adopción de Copilotos

El segundo párrafo se enfoca en la transformación empresarial impulsada por la IA, específicamente en la adopción de 'copilotos' de IA para mejorar la productividad en diversas áreas como la cadena de suministro, ventas, y desarrollo legal. Argumenta que la adopción de estas herramientas es crucial para romper silos dentro de las organizaciones, cambiando fundamentalmente la manera en que se procesa el conocimiento y se realizan las tareas. Ejemplos de empresas indias adoptando estas tecnologías muestran cómo la IA está siendo integrada en los flujos de trabajo para optimizar procesos y fomentar la innovación.

10:03

🌍 Infraestructura y Modelos para la Innovación en IA

Este párrafo detalla los esfuerzos de inversión en infraestructura y desarrollo de modelos de IA como fundamentos para la innovación tecnológica. Se menciona la expansión de la infraestructura de computación para soportar la IA a gran escala, incluyendo la formación de modelos y la inferencia en tiempo real. Destaca la importancia de la diversidad en el hardware de silicona y cómo los avances en la IA están redefiniendo la arquitectura de los sistemas centrales, prometiendo una reducción continua en los costos gracias a la ley de Moore. Además, se hace énfasis en el desarrollo de modelos de lenguaje de última generación y modelos de IA más eficientes y accesibles.

15:05

🔬 IA y la Revolución Científica

Se discute cómo la IA puede acelerar significativamente el progreso científico, especialmente en campos como la química, la ciencia de materiales y la biología. La IA se presenta como una herramienta capaz de emular procesos y reducir el espacio de búsqueda para el descubrimiento de nuevos materiales o fármacos, citando el ejemplo de un desarrollo que redujo el contenido de litio en las baterías en un 70%. Este párrafo enfatiza el potencial transformador de la IA para condensar siglos de avance científico en décadas, facilitando así transiciones energéticas críticas y avances en diversos campos de la ciencia.

20:07

🏆 Aplicaciones Prácticas y Empoderamiento mediante la IA

Este segmento explora diversas aplicaciones prácticas de la IA en empresas y en el sector público, mostrando cómo estas herramientas están siendo utilizadas para optimizar procesos y mejorar la toma de decisiones. Ejemplos incluyen el uso de agentes de IA en Air India para asistencia al cliente, el desarrollo de bots para agricultores por ITC, y la adopción de IA en el sector legal para analizar contratos. Se destaca el impacto positivo de estas aplicaciones en la eficiencia operativa y la capacidad para romper barreras de conocimiento, mejorando la calidad del trabajo y la toma de decisiones en diversas industrias.

25:09

🛡️ Seguridad, Privacidad y IA Responsable

El último párrafo aborda la importancia de la seguridad, la privacidad de los datos y el desarrollo responsable de la IA. Se discuten las medidas de seguridad implementadas para proteger la infraestructura de IA y los datos de los usuarios, asegurando que la propiedad y la privacidad sean respetadas. También se menciona el compromiso con la indemnización por derechos de autor y el marco de IA responsable de la empresa, destacando la necesidad de procesos auditables y principios éticos en el desarrollo de tecnologías de IA. La sección subraya el compromiso con la construcción de un entorno de IA seguro y confiable.

30:10

🌟 Empoderamiento Social a través de la IA

Este párrafo se centra en la iniciativa de una empresa social que utiliza soluciones basadas en IA para promover la movilidad social en la India rural. A través de la grabación de datos en lenguas maternas y la creación de conjuntos de datos valiosos, esta empresa está mejorando significativamente los ingresos de las comunidades rurales. Se destaca el uso de servicios de IA y la infraestructura de Azure para validar y enriquecer estos conjuntos de datos, subrayando el potencial de la tecnología para resolver problemas a gran escala y mejorar la calidad de vida de las personas en regiones desatendidas.

Keywords

💡Inteligencia artificial (AI)

La inteligencia artificial es una tecnología avanzada que permite a las máquinas realizar tareas que normalmente requieren inteligencia humana, como el aprendizaje, el razonamiento y la toma de decisiones. En el video, se discute cómo la AI está transformando la economía y la productividad en India, y se menciona su aplicación en diversos sectores como la tecnología de la información, la cadena de suministro y la venta.

💡Transformación digital

La transformación digital se refiere al proceso de convertir y adaptar un negocio o una industria a las tecnologías digitales. En el contexto del video, se destaca la importancia de la transformación digital en la mejora de los procesos empresariales, la optimización de los recursos y la creación de nuevas oportunidades económicas.

💡Economía de India

La economía de India es uno de los mayores mercados de crecimiento en el mundo, con un potencial económico significativo y ambiciones gubernamentales para su desarrollo futuro. En el video, se discute cómo la AI está desempeñándose como un factor clave en el impulso del PIB de India y la transformación de la economía en su conjunto.

💡Co-piloto (Co-pilot)

El co-piloto es una herramienta de inteligencia artificial desarrollada por Microsoft que ayuda a los desarrolladores de software a escribir código de manera más eficiente y precisa. En el video, se destaca la importancia de la adopción de co-pilotos como un paso crítico para mejorar la productividad y la toma de decisiones en las empresas.

💡Infraestructura de tecnología (Tech stack)

La infraestructura de tecnología se refiere a la conjunto de herramientas, servicios y plataformas que una empresa utiliza para sus operaciones tecnológicas. En el video, se discute cómo Microsoft está reformando su infraestructura de tecnología para incluir la inteligencia artificial y cómo esto está impactando positivamente a los usuarios y la economía en India.

💡Gobierno de India

El gobierno de India es el órgano ejecutivo de la República de la India, encargado de la administración pública y la toma de decisiones en el país. En el video, se hace referencia a las ambiciones del gobierno de India en relación con el desarrollo económico y la adopción de tecnologías emergentes como la inteligencia artificial.

💡Nube

La nube se refiere a la infraestructura de tecnología que permite a los usuarios acceder a servicios de computación y almacenamiento a través de Internet. En el video, se discute cómo la nube es un medio esencial para la transformación digital y la adopción de la inteligencia artificial en India.

💡Innovación

La innovación se refiere al proceso de crear o introducir nuevos productos, procesos o ideas en el mercado. En el contexto del video, la innovación es un tema central, ya que se discute cómo la inteligencia artificial está impulsando la creación de soluciones tecnológicas únicas y la mejora de los resultados empresariales en India.

💡Seguridad cero (Zero trust)

El concepto de seguridad cero es un enfoque de seguridad informática que假定 que no hay ninguna red confiable y que todos los usuarios, dispositivos y datos deben ser verificados continuamente. En el video, se resalta la importancia de la seguridad cero en la protección de la infraestructura de AI y la necesidad de implementar medidas de seguridad sólidas para proteger los datos y la información.

💡Responsabilidad AI (Responsible AI)

La responsabilidad AI se refiere al conjunto de prácticas y principios éticos que guían el desarrollo y la implementación de sistemas de inteligencia artificial. En el video, se discute cómo Microsoft está adoptando un marco de responsabilidad AI para garantizar que sus tecnologías AI se utilicen de manera segura, justa y transparente.

💡Habilidades en AI

Habilidades en AI se refiere a la competencia y el conocimiento en el campo de la inteligencia artificial, incluyendo el desarrollo de algoritmos, la análisis de datos y la comprensión de cómo se aplican estas tecnologías en diversos contextos. En el video, se anuncia un nuevo initiative de Microsoft para capacitar a más de 2 millones de personas en India con habilidades en AI, lo que demuestra su compromiso con la educación y el desarrollo del talento en esta área.

Highlights

The new capability of a neural reasoning engine is revolutionizing the tech stack by making sense of the digitized world.

AI has the potential to significantly impact GDP growth, as seen in India's high-growth market.

Investment in new general purpose technologies like AI can greatly influence a country's future prospects.

Microsoft's mission is to empower every person and organization in India through the adoption of AI technologies.

AI products from Microsoft are integrated into every aspect of the tech stack, not limited to one specific product.

The AI age is bringing expertise to one's fingertips, breaking down silos of knowledge.

Adopting AI co-pilots is critical for increasing productivity and transforming business processes.

AI transformation is not just about the technology but the tangible business outcomes it enables.

Three imperatives for driving business outcomes with AI include adopting AI products, utilizing Azure services, and focusing on responsible AI practices.

Azure's tech stack, from infrastructure to models, is available for businesses to build their own AI solutions.

Moore's law is very much alive in the context of AI, with costs continuing to decrease.

AI has the potential to greatly accelerate scientific research and development.

Microsoft is building the best toolchain for developers to create applications and democratize technology.

The data platform is crucial for AI applications, linking data with the reasoning engine for real-time insights.

Air India and ITC are examples of organizations leveraging AI to transform their operations and customer experiences.

AI can enhance human agency by providing access to information and enabling action in rural India.

Trust, security, and responsible AI practices are fundamental to the successful adoption of AI technologies.

Microsoft's initiative to skill over 2 million people in India with AI skills aims to empower the workforce for the future.

Organizations like Kara are using AI to create economic opportunities in rural India by bringing AI tasks to local communities.

Transcripts

category it's called Information

Management which is you every day you

get up you digitize a little more and

you make sense of the digitized world

now we have a new

capability uh of making sense of that

digitized world a new reasoning engine

uh which is a neural reasoning engine

which we can apply to an increasingly

digitized world and so these two things

a new user experience and a new

reasoning engine pretty much completely

revolutionize the entire Tech stack um

this ultimately is going to have an

impact on GDP right because at the end

of the day you can only talk about tech

uh as a real thing if it is going to

have a real impact in the overall growth

of an economy and in the in India's case

uh as Punit was saying definitely today

this is one of the highest growth

markets uh you see it the buoyancy of it

uh the government and all of you have

high Ambitions of what's going to happen

even by 2025 uh and what percentage of

that GDP growth is going to actually be

driven by AI are all I think going to be

very interesting numbers for us to track

in fact I was recently came across a

fascinating statistic in in the United

Kingdom during the height of the

Industrial Revolution they spent as much

as 10% of their GDP on the railroad

system and then the rest obviously is

history but I think that's another

understanding which is when you have a

new general purpose technology how

intensely you invest and deploy it

cross- sectorally inside an economy I

think makes difference uh to a country's

prospects going forward and that's why

that's exciting that we are at the early

stages of this adoption cycle but

rapidly scaling it um that's what we are

focused on at Microsoft our mission is

to empower every person and every

organization in India to be able to

participate fully uh to me that's what

you know gives me a lot of satisfaction

is to see new technology but applied in

unique intense ways uh across small and

large businesses across public and

private sector across every part of

Indian economy across every region of

India that's what's uh really our

mission that's why we have built this

into the entirety of tech stack people

say what is an AI product from Microsoft

it's called everything from Microsoft

there is not one product quite frankly

uh because we've gone plumbed it all the

way from the core infrastructure it's

fascinating the more I look at it you

know I get up in the morning and I'm

reviewing some data center plan uh the

everything from what we do on the

concrete to the hwx system to the

infrastructure are all being defined by

the power draw of AI systems I mean it's

not about like any one product but the

entirety of what we do has been is being

shaped um and the opportunity of

ultimately though you can even you know

we us to talk about digital

transformation all through Cloud that's

effectively the same thing which is the

way we engage our customers uh the way

we engage you know the experience

experience of our employees the business

processes and how they're optimized uh

and the space of just raw innovation in

any Enterprise all of these are the

ultimate outcomes that we will measure

ourselves by right so it's not about

tech for tech sake but it's a tech

applied for these business outcomes that

I think more so than I would say even in

the cloud era I feel that the ability to

tangibly say what is the change in the

business outcome I think we're able to

go to grock that uh much faster so it's

exciting time for AI

transformation um and AI driven

transformation um so in that context for

us I would say there are three

imperatives if I had to sort of even

look across what's happening around the

world uh what we ourselves quite frankly

are seeing inside the company there are

three imperatives of how you all can get

ahead um on uh driving those business

outcomes the first one is to just adopt

some of these products like co-pilots

Fast uh it reminds me quite frankly of

uh the first paradigm shift I was part

of which is the PCS uh you know even

it's interesting by the way PCS are an

interesting thing because if you even

take some of these critiques of uh the

impact of it in economic growth like

Robert Gordon uh the one thing that he

gives credit to is PCS in the late late

90s when they became standard issue and

part of uh work uh they changed

productivity in fact he will claim that

that was the last time it actually had a

direct relationship uh to productivity

stats um and obviously I take a lot of

pride in it but I feel this AI age is

very similar that is it's diffusing

right if you think about what PCS did

they brought information at your

fingertips this age of AI is really

really expertise at your fingertips

think you know what happened to

forecasting before PCS right before

email and spreadsheets how did we do

forecasting uh and then how did the

business process of forecasting change

same thing is happening now whether it

is supply chain whether it's sales

whether it's legal or what have you when

you can summon the expertise it's the

greatest Silo breaker and so that's why

I think adoption of co-pilots becomes

critical and we seeing all kinds uh of

productivity stats uh here but all of it

stems from this fundamental notion that

knowledge turns inside an organization

are changing right and you can come at

it horizontally whether it's Frontline

work or knowledge work or you can come

at it from business process sales

service software development security

operations so the tangible way you can

in fact my biggest encouragement for you

is to go deploy these tools create a a

cohort measure it yourself that's it uh

because you got to build your own

confidence that you can sense it see it

uh and that way you can drive it uh and

so that's what I think is very much

possible and you see all of these

statistics uh and in fact in in in India

uh that's what been most exciting for me

you know I've been sort of learning

about all of these uh use cases IT

services at some level you could say

it's existential for any one of us right

I was talking to Jensen the other day

he's got all of everybody

uh in Nvidia is deployed copile it makes

sense because he's his thing is like you

know like we will all be found out in

fact the firm level differences will

start showing up uh especially in Tech

uh in short order if you're not an early

adopter and so obviously that you see

even with the IT services in in this

country uh all going uh Fast and Furious

on it but it's just not that and in fact

I think Punit had these uh slides where

there is adoption across the board

across every sector of the Indian

economy and it's exciting to see just to

give you a flavor for it I was just

reading up about what axis Bank did this

is one of the the thing that I sort of

describe as the standard issue right

just like at some point everyone all of

us sort of said hey PCS a standard issue

co-pilot as a standard issue that's what

they're doing so they're basically

saying let's take the most horizontal

approach of having everybody equipped

with a co-pilot so that every piece of

knowledge work can be done fast somebody

described this to me as lean for

knowledge work I think that's a good way

to even think about it right which is

you're increasing the speed and the sort

of alignment across organizations so

that you can get things done with the

least amount of time and effort right

that's essentially the mindset uh so

that's kind of what axis is trying to do

um both um HCL Tech and um M tree are

also adopting this but they're also

adding their own workflows into it right

so it's not just the horizontal tools

that are baked in but in the case of HCL

a lot of their project management and

Bug triage is integrated into the

co-pilot in the case of mine tree even

the insights into how they drive their

Staffing for various projects is plugged

into it so in some sense it's not just

the one workflow that we have but you

can steer it to the workflows that

matter inside your Enterprise and wire

it all into the co-pilot so copilot's

understanding is not just what was

prefabricated it is what your Enterprise

is and that's exciting and in the case

of infosis the the this is probably one

of the bigger use cases of GitHub

co-pilot uh GitHub Co in fact if I go

back into my own belief in this entire

regime of AI if you will uh all came

about when I first saw gbd3 uh inside or

what became gbd3 inside of GitHub uh

co-pilot um and it's the most proven use

case it's the place where the

productivity stats are most robust I

mean it went from I mean remember

software developers are pretty skeptical

people right it went from this won't

work to oh my God I can't live without

it uh in the shortest amount of time uh

and it's exciting to see now this is now

being broadly deployed uh even across

organizations like infosis now the um

the next imperative of course is adopt

so the first thing is don't reinvent the

wheel get the co-pilots that have been

prefabricated whether it's sort of for

specific business processes like

software development or security or the

horizontal ones add your own data add

your own workflow so that's if that's

number one the number two is we are

exposing all of the tech stack we use to

build our co-pilots to you as first

class Azure services so that means

there's nothing magic about what we did

because you can take that same thing and

apply it to any anything you want to

build duvo as far as your own AI

products uh and so that's the second

imperative this co-pilot stack uh from

the bottom to the top is available to

you uh so for our investments here

obviously start with the raw

infrastructure itself right we have more

regions than any other hyperscaler

around the world in in India itself uh

we have uh three regions already fourth

coming on line we have the two regions

with Geo so we are definitely investing

to make sure that compute with low

latency at scale including all the AI is

available for all of you to be able to

take advantage we're investing a lot as

I said in making sure in fact when I

think back having started uh in our

Cloud group you know what now 15 years

ago um I I sometimes sort of wake up and

sort of can't recognize even the core

system architecture it's so rapidly

changing so we are investing a lot part

in making sure that the AI

infrastructure whether it's for large

model training or small batch training

or inferencing uh how do we make sure we

have the best performance on end uh so

that's something that we're doing across

multip and and having diversity of

silicon uh as well so we are very very

excited about the progress and some of

the Benchmark data uh that we have seen

uh is very very because here's the thing

right people talk about the death of

Mo's law let me tell you Moors law is

very much alive when it comes to to AI

right what used to be uh you know the

cost curves of AI are only going to come

down so that means a set of tokens in

and out today are going to follow the

mors law so that means when you're even

thinking about your AI project uh I I

tell my teams don't worry about cogs we

can know we can optimize cogs um and so

if you want to be the most ambitious

about being able to use this to create

the most value for your customers

partners and for the business and then

know that there is no other raw material

in the world that just drops in price

using Mor's law like this and so that's

if you talk about business leverage this

is business leverage um the next thing

is we have the raw infrastructure but

then we have these models right so these

are like the foundaries of this world

right so therefore you have the lot we

have the the best uh model today even

right with all the sort of hoopla you

know one year after uh still GPD 4 on

nlu take your benchmark and anyone still

is better uh we're waiting for the

competition to arrive it'll arrive I'm

sure but the fact is uh that we have the

most leading uh llm uh out there uh we

have all the open source models so it's

not about like even co-pilot is not

about just uh open AI models uh in fact

we use a combination of models uh in any

of our products which is what all of you

will use so we have in Azure the best of

Open Source models the best of Frontier

models uh we also are innovating in what

is is called the small language models

in fact one of the best if you go to

hugging phase today uh you'll find that

fi is the leading small language model

uh and it's a pretty cool research uh

finding right which is you can take a

slightly different approach uh to

learning as opposed to thinking of all

tokens as equal if you took a cognitive

approach to like for example the

original book or the original paper on

uh Transformers was attention is all you

need uh the paper we wrote uh was

textbooks is all you need uh and it's

intuitive right when you're learning

something uh you would rather go to a

textbook and learn versus saying let me

run from all the text that was ever

written about a subject so that

intuition is what's leading to these

small models actually performing on par

with some of the largest models so we're

very very excited uh about the slms

these are the things that can run on on

your phones and on your PCS as well as

obviously in the cloud at scale and you

can cost stop optimized going back again

to that morw point so you can think

about llms and slms all playing a role

in how you build out your products the

perhaps when I look at the thing that

I'm most excited about uh is uh what I

think all of this can do for Science and

the acceleration of science clearly I

think knowledge work Frontline work um

like whether it's software development

whether it's the everyday uh knowledge

work will be revolutionized

but if I think about one of the pressing

challenges we as a society have is we

have to take whatever the 250 years of

science right they take energy

transition we're not going to make it to

the other side um if we don't compress

maybe 250 years of development of

chemistry into perhaps the next 25 years

uh so we need some new way to even

change the scientific process and its

curve uh and that's where I think AI can

help um AI ENT the way we think about it

is like if you want to simulate in

silico uh that's where high performance

Computing is used right if you want to

even simulate a single molecule or

molecular Dynamic cells and then all way

the complex systems of course the real

breakthrough there will be Quantum but

the interesting thing AI does is AI is

kind of like an emulator that or it

reduces the search space so that you can

for example if you want to build a new

material uh you can discover that new

material or generate in fact generative

diffusion models like matter genen that

we developed are allowing us to build

new materials and in fact we round

tripped one such example recently we

worked with one of the National Labs in

the United States and essentially

generated a new material which reduce

the lithium content in a battery by 70%

right I and I really wanted my team to

show me this that this is a complete

roundtrip so it's not just oh we

generated something in silico but can

you fabricate it and put it back back

into a battery of course we're not a

battery company but think about this in

the hands of somebody who is in the

battery business right so you're not

just building batteries you're able to

kind fundamentally change the frontier

of science uh that helps you build the

next generation of batteries this is

happening in chemistry it's happening in

material science uh it is going to start

happening in biology it's obviously drug

discover you know drug being able to

generate new molecules for drug

Discovery are you know is easy it's the

drug discovery process which is going to

be the more complex one but nevertheless

we're definitely uh very excited about

what AI for science means so we're now

imp thinking of this as first class just

like as you thought about Microsoft 365

for knowledge work you will sort of come

to and we'll do this through partnership

so that's why I think it's very exciting

in a country like India where there is

so much investment and so much Capital

going into the core basic science uh

driven Industries to be able to really

come help India leap froggy even some of

the Scientific Revolution would be very

very exciting um to harness all this

we're building the best tool chain in

some sense one of the things that I take

great satisfaction from is we are a

tools company that's how we got started

in 1975 and here we are in 2024 and we

get a real kick out of building the best

tools for developers uh whether it's

GitHub whether it's vs code whether it's

the low code no code tools and Power

Platform uh all of these things coming

together and the power of all of this

technology has to be democratized so

that people can build applications at

the end of the day uh and that's what

our goal is uh with the entire tool

chain the other very very important part

of the teex stack is the data platform

right so if you think about a reasoning

engine if you want to build an

application it's about taking the data

you have and using this reasoning engine

on top of the data and so we are

bringing essentially AI compute to all

of the data in its all its forms whether

it's operational stores whether it's

analytical stores linking these two

things in real time uh in fact one of

the things that I'm very sort of excited

about is a product called fabric uh and

what we did is I I think of it as the

product built for the AI age because it

separates out uh the core storage from

compute and with even the business model

so that means I can build a application

which is using all the data in one Lake

and I can have different types of comput

sometimes it's SQL sometimes it's spark

sometimes it's Azure open AI there are

three different types of compute jobs

that can all be brought in on the same

data or different data uh joints if you

will and so that to me is a very

breakthrough uh way to think about your

data uh in the context of AI so this is

this is a one of the a lot of people

talk about their AI projects the first

thing is you got to get your data house

in order in order to be able to get uh

the best value out and that's why you

see the adoption Cycles even in India

even though the cloud adoption in India

was a lot slower than what what happened

in the rest of the world but the fact

that it has happened is what's helping

anyone today take AI without any of that

diffusion uh impedance or delay if you

will because you're already in the cloud

your data is already in the cloud and

you're able to rapidly they turn it

around and adopt something like a new

reasoning engine on top of that same

data uh lots of interesting um uh

customer use cases um uh here and I

think Punit had many uh of these

examples and I had a chance this morning

to visit with a few of these folks and

really learn about how they are going

about building things and they're really

inspirational right uh uh take Air India

uh if what Chandra was telling me is how

he's really inst instructed the team

there to accelerate everything that

they're doing right essentially from

lack of um he was telling me you know

they're just putting in all of the

systems uh that had like a lot of

deficit but rapidly getting into uh you

know place where they're leading in fact

going from not having systems to Leading

with some of those experiences so the

thing that they built for example was an

agent an AI agent for the customer it's

one of the leading uh uh products uh out

there so if you go to the Air India home

you know homepage on the web you can

speak to this uh AI agent complete your

entire transaction and travel planning

and they have like high ambition on it

uh and they have on the back end of it

as well they're also creating a bunch of

agents to streamline the business

process not just on the front end to the

consumers I had a chance to meet with

the ITC team that they're building this

um uh essentially a bot for the farmers

uh to be able to then get all the

knowledge they need uh because in some

sense in a vernacular language to be

able to access all of the information

and to be able to even take action uh

right this is the other very interesting

site or rather interesting Dimension to

this which is it reduces the expertise

required to use Computing and the

knowledge that is accessible right so

that is why I think in India the most

exciting thing is for the quality of all

Indian output to go up because it sort

of helps reduce the barrier I think that

that's going to probably be uh an

example in in farming um arnd uh they're

you know this is a fascinating use case

so they obviously uh have been in

business for 100 plus years and here

they are using AI to for example for

their legal department to say let's make

sense of contracts faster let's

understand risk uh and what have you

very obvious use case after all any

lawyer or anyone in the legal department

who spent a a lot of time uh trying to

assess uh risk On Any Given contract the

easiest use case is to just be able to

use gen to think you know make sense of

it but the interesting thing is for

their commercial teams to effectively

summon the legal expertise right right

at the point when the contract or the

RFP is even being issued that's the real

transformation the real transformation

is when the expertise silos are broken

right so you don't have to wait for

finance to at the end of the quarter

tell you all the mistakes you made or

for legal to tell you all the risks you

took without knowing right so it's the

ability to bridge that expertise Gap

just in time that's why I kind of think

of this as lean for knowledge work uh

that's what they're doing um and then

the last example was very inspirational

like last time I was here I saw the the

the demo of jugal bandi and I spoke

about it many many times because it was

just a drop you know it's like drop the

mic moment for me to see how this

digital public good and Bash combined

with jugal bandi was being used to

transform lives of people in rural India

speaking in vernacular languages uh and

being able to not only get access to

information and knowledge but to take

action right that I mean talk about

human agency in the age of AI being

enhanced that was an example and now the

team has even built a fantastic thing

called jugle Bundy Studio which is a no

code lood Tool uh for people to be able

to build these BS and make them

available to all of the uh social

entrepreneurs the nonprofits who are

doing things um and so I had a chance to

meet uh the folks uh from uh one of the

nonprofits or social Enterprise called

aami and they built a thing for uh

migrant laborers coming in uh and for

them to be able to access uh this

information uh it's called bundu Uh so

therefore somebody who comes in from say

Rural and prades into Bangalore is the

example they showed and sort of in

speaking in telu perhaps getting access

uh to information on how they can get

housing how should they go about getting

the right terms all of the things then

think about the sort of you know

reducing the um i' would say the stress

of being able to get to Services when

you come into a new place across uh the

the you know across India is fantastic

to watch the empowerment it gives to the

end the end citizen here but also the

empowerment to all of the social

Enterprises to be able to use something

like this to make a real difference is

fantastic to watch um so those are some

some of the f i that the ability for you

all to build your own AI applications is

never been easier never be and and I see

it at scale in India so that brings me

to the last imperative which is

ultimately trust uh because in some

sense as we talk about AI we got to talk

about safety and Trust as first class CL

part of it um it it has a couple of

different dimensions the first one is

the dimensional security right because

again just like how data is important

securing your entire AI estate is

important whether it's the end points

whether it is the identity uh whether

it's infrastructure or your applications

this is where the zero trust environment

uh that we have sort of evangelized and

built the product suite for uh becomes

very very important because you can't

talk about security you've got to

operationally operate it right or you

have to have operational security

posture that matches your rhetoric

around security and uh this is everyday

exercise uh as somebody said to me you

can't get fit by watching others go to

the gym you got to go to the gym and

this is uh a place where you got to

practice that um the second uh piece is

uh the safeguards that we have put in

place because one of the questions

people ask is look in this age of AI

who's who what what's happening with my

data is my data my data it's absolutely

your data will always be your data uh

your data is not used to enrich anybody

else's models um uh your AI model so if

you build let's say a fine-tune model

and you deploy it it's protected by

using the same security boundaries that

we talked about uh and also our

commitment to copyright uh

indemnification because at the end of

the day uh this is all going to be you

know a societal decision effectively

what is fair use what's the

transformative use of data

uh very important by the way it's

fascinating to watch countries like

Japan and others take very enlightened

stances around copyright uh because they

have sort of seen what has happened um

you know in other Paradigm shifts where

perhaps they over regulated uh so I

think this is something that I think

each diff each country will choose its

regime the global Norms will emerge but

copyright is something that we are are

very much on the Forefront of and making

sure that there's indemnification

directly from us um and then then the

last thing is of course the responsible

AI framework and even here the you have

what we have done is we have the

principles we have taken those

principles and translated them into a

set of standards and those standards are

implemented uh as a set of processes

starting at the engineering so these are

not just things that we talk about but

they're core part of the Engineering

Process and we have an audit function

right because it's not just about saying

things that you literally have to have

an audit function where you go in and

say are you managing the process the way

you said you are and so this is how we

approaching it all of this is available

as effectively a set of services for you

so whenever you are thinking about

building your own AI applications you

can go ahead and Implement effectively a

robust responsible AI framework uh

around your own AI development so uh so

that's really uh the three imperatives

that I at least wanted to share with you

um I want to wrap up where I started our

mission at the end of the day is to

Empower every person and every

organization in India uh to be able to

achieve more in this age of AI and in

that context I'm very excited today to

even announce a new initiative around

Skilling uh we are going to C 2 million

plus uh people in India with AI skills

uh we think that obviously at the end of

the day uh really taking uh the the

workforce and making sure that they have

the skills in order to be able to thrive

in this new ages the most important

thing that any of us can be doing and we

are you know happy to play our role in

it but it's not just the skills but it's

even the jobs that get created and in

this context I was very inspired when I

had a chance to meet uh with the folks

behind Kara in fact it was uh the

founder of Kara was telling me about how

a lot of the research that you know was

the the foundation of it came from

Microsoft India Microsoft research in

India it's fantastic that it's now

translated into this thriving uh

organization that is taking essentially

what is AI tasks of labeling uh and

bringing those jobs very you know wellp

paying jobs to rural India um and being

able to really create Economic

Opportunity uh that is pretty unique and

it's fantastic to see not only people in

you know who are software developers

participating which we know GitHub for

example in India has more AI Engineers

or the AI engineering community is

second to the United States in India uh

but you even see that R in rural India

people are participating uh in the AI

economy uh thanks to the work uh of

organizations like Kara is fantastic uh

to see and so let me I think we'll roll

the video and you'll get to chance to

meet some of the folks uh who are part

of

this India has 22 official languages and

morati is actually a language that is

spoken by millions and millions of

people but unfortunately many Indian

languages are under resource we cannot

let language be the thing that does not

allow you to use technology AI is a

transformative technology people deserve

to have access to these amazing

Technologies in their language Cara

exists to accelerate social mobility in

rural India by building AI based

solutions to help communities our

workers record sentences in their mother

tongue on their smartphones and for the

simple task we pay them nearly 20 times

the Indian minimum w and that leads to a

data set so they keep on making

royalties every single time those data

sets are resold I think it is imperative

that if we are changing someone's income

structure in this way that we also need

to help them understand financial

literacy

K's platform is hosted on Azure we use

Azure cognitive services for our own AI

based validation work and we also using

a your open

AI if you can solve it for India you can

solve it for the

world

uh anyone may have we have mic's flowing

around just raise your hand and a

question yeah go

ahead

all