Hide-and-Seek: Uncovering AI Assets
Full Transcript
# Hide-and-Seek: Uncovering AI Assets **Source:** [https://www.youtube.com/watch?v=vI_2LSc6NCg](https://www.youtube.com/watch?v=vI_2LSc6NCg) **Duration:** 00:08:51 ## Sections - [00:00:00](https://www.youtube.com/watch?v=vI_2LSc6NCg&t=0s) **Finding AI Data in IT Silos** - Hillary Hunter likens locating data and models in tangled IT and cloud environments to a game of hide‑and‑seek, emphasizing that unified, consistently designed infrastructure is essential to eliminate silos and accelerate AI deployment. ## Full Transcript
I think all of you will know the Beloved
childhood game called hide-and seek I'm
sure it goes by different names around
different parts of the world but in
today's era of generative AI your data
and your models are going to be two of
your most important assets however
finding those assets in a complex it
environment sometimes can feel a lot
like playing a game of hide-and seek and
it really doesn't need to be that way
you don't want to spend months trying to
make sense of your different it and
Cloud environments and wrestle through
different data governance systems before
you can actually access and gather the
data that you need to become productive
with AI because there's no AI without
data after all but a disjointed chaotic
it environment creates silos that make
it difficult to seek out and collect the
data that you need to train tune and
leverage your AI welcome to AI Academy
my name is Hillary Hunter and I'm the
chief technology officer of the IBM
infrastructure unit I oversee the
technical strategy and the development
and deployment of platform capabilities
for our clients in our hardware systems
and Cloud families and these days my
mission is to identify how we can use
infrastructure and platform capabilities
to enable Ai and maximize its value for
our clients that work starts with making
intentional consistent architectural
decisions that eliminate silos and
strategically bring data and AI together
each component of your it infrastructure
should fit together like a puzzle but
gaining access to your data or finding
your models shouldn't feel like solving
one a decade ago we had the first round
of ai's emergence in the broader
commercial space it was an ERA marked by
extensive experimentation lots of proofs
of concept but relatively limited
adoption early AI couldn't do more than
identify objects in a picture and add a
caption perhaps or deliver responses to
predetermined questions built with
rule-based systems these use cases
weren't pervasive enough to justify
adoption for many Enterprise
organizations s those who did adopt
could find the latest AI capabilities
usually in the cloud and only in the
cloud in order to use it adopters had to
ship all their relevant data to the
specific Cloud environment that housed
that Ai and with so few architectural
options available organizations had to
put a lot of trust in their Cloud
providers and in many cases invest in
the high cost of moving large amounts of
data to the cloud Ai and Cloud
architecture look very different today
gen AI is powered by a class of deep
learning algorithms including Foundation
models large language models which
enable rapid solutions to be built in
the areas of things like natural
processing simplifying Enterprise
operations and enabling Enterprises to
take application modernization and
workflow automation to the next level
gen can perform important tasks on the
Enterprise level like creating and
understanding code authoring marketing
materials detecting fraudulent activity
and extracting insights from
unstructured Text data
with commercial use cases multiplying by
the day and the growing potential to
maximize business value organizations
are eager to start adopting and
deploying AI rightfully so however Cloud
technology has also expanded
significantly we're now in the era of
what we refer to as hybrid cloud in
which Cloud Technologies are available
across the multicloud landscape and
spanning into un promises environment so
traditional it Estates private Cloud
public cloud and even the edge become a
part of the hybrid Cloud conversation
critical business data often resides on
premises in the data center it's siloed
but it's kept secure which is why many
organizations store their most sensitive
data on servers on premises the problem
is that traditional management of
servers without a unifying platform
approach lacks the flexibility needed to
integrate and create AI at scale a 2023
study IBM conducted with the Harris
organization recently found that 65% of
organizations are using a combination of
on-premises Hardware with multiple
private and public Cloud capabilities
creating a hybrid environment and
complicating their it landscape many
organizations have fallen victim to what
we can call a Franken Cloud it's a
monster of each organization's own
creation the Franken Cloud essentially
is a concept that says you have a very
muddled incidental hybrid Cloud
architecture it has all the right
components to meet the definition of
hybrid it has some cloud CL on premises
and public Cloud usage but it lacks the
structure and intentionality to be a
hybrid Cloud a consistent environment
for example a large Enterprise may have
several business units working with the
same data with no centralized place to
access and analyze this data each
business unit creates its own cloud
environment and moves a copy of the data
to that environment the organization is
then left with umen isolated Cloud
deployments that increase data silos
increase storage costs increase security
risk and effectively add architectural
complexity to the overall landscape
trying to run and build AI models in so
many different locations only worsens
these problems so what's the solution
well it's not to choose between on
premises infrastructure where you may
have limited access to AI capabilities
or choose between Cloud environments
that put your data privacy and Security
in a totally different comp context
instead businesses need to adopt an
infrastructure strategy that's hybrid by
Design and work with an AI vendor that
offers choice in AI deployment location
a hybrid by Design approach eliminates
the challenges born from random acts of
cloud usage by establishing consistency
across the various functions tools data
sources platforms and systems that make
up your whole it environment you're not
simply adopting hybrid technology you're
making intentional architectural
decisions that unify every part of your
business you're sharing the same data
applications and policies between on
premises infrastructure private Cloud
public clouds and even out to the edge
you're bringing critical data and
applications into a consistent
environment which means different
business units can access the data they
need without causing fragmentation or
adding
complexity essentially hybrid by Design
means less Frank clouds more
enterprise-wide alignment less time
trying to access your data and more time
actually using it to successfully build
fine-tune integrate and deploy AI into
your business now that we've defined the
power of a hybrid by design it
infrastructure let's answer some of the
common questions around Ai and how does
this topic even relate to hybrid Cloud
first off can current technology provide
a strong foundation for today's
unparalleled AI capabil ities absolutely
it can an intentionally designed hybrid
Cloud architecture allows you to
facilitate rapid scalability by tuning
and deploying AI wherever your data
resides this is a central step towards
more efficient cost effective value
creation second what does that
infrastructure look like it looks like a
hybrid Cloud environment but with a
consistent platform and a consistent set
of AI capabilities and governance across
that full span it's a strategic coupling
of public and private cloud scalability
and on premises like security how can a
hybrid by Design architecture help
businesses gain the most value from AI
well an intentionally designed hybrid
Cloud architecture allows you to train
modify and integrate AI capabilities
into your business right alongside your
most critical data that means that
Solutions like Advanced automation
real-time analysis and around the-clock
customer engagement are right at your
fingertips ready to be deployed when and
where you need the most you're doing AI
where your critical data is in all the
places that you're doing business and
where your customers are organizations
need an integrated open and intentional
hybrid Cloud architecture to enable AI
deployment at scale hybrid by Design is
all about deliberately building a
cohesive it architecture that offers the
combined benefit of multiple systems and
services when your hyb Cloud
architecture is strategically selected
carefully structured and you leverage an
AI vendor who meets you where you are
your Enterprise can run AI workloads
with the operational agility and
scalability of the cloud underpinned by
the robust security resilience and
governance traditionally expected of on-
premises
infrastructure that is how you unlock
the full value of AI