MLOps becomes IntelligenceOps: "You're gonna need a bigger boat"

"Exploring AI Futures with Bill Murray (not the actor)."

(This is the third article in a?series?on Artificial Intelligence. The first post is here . The?second?is here .)

In ‘Strategizing?AI… ’?I offered three patterns we found over nine months of watching?AI?evolve?through the lens of?Wardley Mapping :

· ? ? ? The?Expansive?Utilization?of?AI

·?? ? ? The?Mandatory?Operationalization?of?AI

·?? ? ? The Aggressive Industrialization of AI

In the?second?of the?series,?Practical steps with?AI?– The Stairway To Heaven we?expanded?‘The?Expansive?Utilization?of?AI’.

Now, I'll elaborate the next pattern:?The?Mandatory?Operationalization?of?AI. Be warned… “You’re gonna need a bigger boat”

MLOps evolves to IntelligenceOps (IOps)

Generative?AI?tops business priorities for 67% of IT leaders .?The vast?majority of workers across various industries and geographic locations have tried?generative?AI?tools at least once, indicating rapid adoption .?The net is this: we are ushering in a new era for organizations -- one that's more complex than traditional?AI?frameworks are used to handling.?

We touched on?Predictive?AI's existing strengths versus the newer, emerging ones from?generative?AI?in?"The Stairway to Heaven."? Predictive?AI?excels at analyzing?data,?making?predictions, and offering recommendations. It's the core of today's expressions of?AI.?Generative?AI?takes?things?up a notch. This extends beyond crunching numbers. Instead, the focus is on?creating, understanding emotions,?making?stuff happen, improving over time, and having meaningful chats .

Now, when we start mixing?generative?AI?into an organization's technology portfolio, it exerts an exponential influence on how much?data?you're dealing with.?AI?also changes the kinds of?data?you're using; quality becomes paramount.

MLOps has been handy at keeping?things?smooth with predictive?AI, but it's struggling to keep up with the more complex needs of?generative?AI. New frameworks and practices are cropping up to handle advanced concepts including?evolving?model architectures, expert?data?management, vigilant monitoring, swift updates, and agile governance for?AI's?evolving?nature.

To fill this gap, we?need?to think?bigger. Bells and whistles are good for MLOps but insufficient here. This isn't a hyperbolic evolution, it's real – DXC is designing solutions to handle?generative?AI's?extensive and varied challenges.

You’re gonna need a bigger boat

IntelligenceOps?is still in development, but it has the potential to revolutionize corporate?AI?strategies. An?Intelligence?Operations (IntelligenceOps) platform emphasizes growth and improvement. It builds on traditional?AI?processes and fine-tunes them into perfect alignment?with the dynamic and evolving world of?generative?AI . The objective is to?extend the operational scope ?to all aspects of?AI?management,?including?knowledge task management, data handling, model?optimization, monitoring, performance evaluation, deployment optimization, and governance oversight. With?IntelligenceOps, each of these elements undergoes an upgrade to enhance their effectiveness and adaptability:

• Model?Building and Evolution: This is not a path of trial and error. We begin by taking proactive?measures?to keep up with the rapid advancements of?generative?AI,?including?advanced?privacy?measures?and continuous?model?refinement.
• Data?Management:?Data?authenticity is paramount as?data?volume and diversity increase. Our evolving approach to?management?handles complexity with sophistication and reliability.
• Monitoring and Deployment: Shifting from basic accuracy checks to a holistic output quality assessment, we are now exploring the nuances of?generative?AI's outputs. Our focus is on a qualitative evaluation, prioritizing depth over sheer correctness.
• Governance?Model: The complexities of?advanced?AI?technologies require a robust?governance?framework to navigate the intricate nature of unsupervised?learning, content moderation, deep?learning?interpretation, intellectual property, and ethics with extreme care.?

IntelligenceOps?equips us to?reshape our?AI?operating models; making?AI?better, faster, and cheaper to deploy, at enterprise and ecosystem?scale. Crucially, we are doing this?without increasing our Technical Debt? and?using Agile at?scale. This enables new meaningful paths to?apply intelligen ce, reshaping IT and the organization.

See what I mean:?“You’re gonna need a bigger boat”.

Ethical AI evolves through stages

Transitioning from MLOps to?IntelligenceOps?requires attention beyond technological?advancements; this is a?comprehensive re-evaluation and adaptation of?ethical?practices. ?This evolution of?AI?Ethics mirrors the maturation process of generative?AI, beginning with an intuitive, exploratory phase and progressing toward a more structured and value-driven approach as the?technology?matures.

Employing a Wardley Map to chart AI's development reveals a significant transformation in?AI?management and governance. This progression is marked by?three distinct roles , each playing a pivotal part in shaping the?future?of?AI:

• AI?Pioneers(**): These innovators lead the way in the early?stages?of?AI, where intuition and pioneering spirit guide the exploration of ethical and technical boundaries. Their?role?is crucial in laying the groundwork for?future?developments.
• AI?Settlers (##): As?AI?technology?progresses,?settlers?take over, building upon the foundations laid by the?pioneers. They start systematizing and formalizing the initial explorative approaches, establishing reliable?practices?and methodologies.
• AI?Town?Planners:?Town?planners?assume responsibility in the advanced?stages?of?AI's?lifecycle. Their focus is solidifying these established?practices?into stable, scalable operations. It creates a robust policy framework, secure and adaptable to?future?advancements.

Each?role?is integral to the?ethical?evolution of?AI. The guiding policies evolve in tandem with the rapid growth of the technology in organizations, and the?Settlers?are the bridge.

A final thought – Sam Altman and AI (again)

Altman's impact on?AI's?trajectory is substantial and undeniable, regardless of personal opinions about him - he is a pioneer . In the practical realm of?AI?development, entities such as?OpenAI and Microsoft?act?as?settlers . Their efforts in incorporating?ethical?measures into ChatGPT and its broader ecosystem demonstrate the transition of?ethical?theory into tangible practice. This is essential in grounding?AI's?ethical?considerations in real-world applications.

Regulatory authorities like the?US government ?and the?EU ?act?as town planners for?AI.?These?teams construct the legal framework for?AI's?responsible integration into society and setting global standards for its operations. The trick is for?Settlers?to?‘steal’?nascent?frameworks?from Pioneering?AI?projects and?Regulation Sandboxes ?to?‘steal’?road tested?frameworks?from scale ups and mature organizations and?convert the findings into Policy .

With the increasing adoption of generative?AI?in?organizations, the?role?of?Ethical?AI?representatives has once again become prominent.?These?individuals?act?as pioneers within their?organizations, working in cross-functional teams to adapt and apply?ethical?frameworks?first devised by committees.

The Settler?role?is critical in bridging the gap between innovative, pioneering practices and?established?scalable operations.?Settlers?'borrow' from?both?ends of the spectrum—integrating fresh ideas from the?pioneers?and?established?methods from mature organizations into a coherent policy?framework. This?‘living ethical?framework’?is adaptable, keeping pace with?both?the swift innovations in?AI?and the stringent demands of regulation.

For those involved in?AI?development, whether as?Pioneers,?Settlers,?or Town Planners, adopting?IntelligenceOPs?is more than a recommendation—it's a strategic imperative. With Generative?AI, we have a Cambrian Explosion of Use Cases that?every?business function,?every?division, and?every?ecosystem partner will be piloting, deploying, and scaling. To manage?AI?at scale, robust frameworks like?IntelligenceOps?are crucial.?

"You're gonna need a bigger boat".

Editor’s note: for those of you who are not movie buffs, we offer a handy reference to this pop-culture chestnut

** - recent updates to this framework now position this role as "explorer." We have left the original name to tie to prior reference work

## - recent updates to this framework now position this role as "villager." Again, we have left the original name to tie to prior reference work

=/=/=