Sunday, November 27, 2022

Addressing the Challenges Associated with Deploying AI at scale

By Prangya Pandab - February 11, 2022 4 mins read

The enterprise is rapidly discovering the numerous ways AI can expedite and optimize processes, but most of these successes are taking place at a limited scale so far.

AI, like any other technology, works best in controlled environments, but pushing it far and wide over an increasingly diversified data ecosystem is fraught with dangers.

The enterprise becomes a loose collection of platforms, processes, and cultures at scale, rather than a unified, fully integrated digital environment. Of course, AI promises to change all of that, but it won’t be able to function at scale unless it achieves scale, which means there’s still a lot of work to be done before companies can fully exploit AI’s value proposition.

Also Read: CIOs Should Prioritize Environmental Sustainability Going Forward

Assembly-line AI

This issue is compared to an organization that builds each product from the ground up, with no standardization or consistency among processes, components, or quality control, according to  McKinsey & Co. researchers. To scale AI across the enterprise, it needs to be put on a digital production line where teams from all over the company can turn out risk-compliant, production-ready, and reliable models.

Thankfully, AI tools and platforms have progressed to the point where more regulated, assembly-line approaches to development are now available, with the majority of these being harnessed under the MLOps model. MLOps is already helping in the reduction of the development cycle for AI projects from years and months to as low as two weeks. Organizations can produce consistently reliable products with all the inbuilt security and governance standards needed to scale them up quickly and effectively using standardized components and other reusable assets.

Of course, full scalability will not be achieved overnight. AI implementation is divided into three stages. It starts with a proof of concept, then moves on to strategic scaling, and lastly to the point when the company is industrialized for growth. Three fundamental elements will need to be refined in each phase:

  • Managing expectations and providing clearly defined objectives, operational models, and timetables to drive “intentional AI”
  • By carefully managing both internal and external data used to train AI, data noise can be turned off. AI, like any other technology, has a garbage-in/garbage-out issue.
  • Treating AI as a team sport by promoting multi-disciplinary and cross-platform deployment

Unique issues for unique enterprise

Because no two businesses are alike, everyone’s journey to full-scale AI will be unique. The key to a successful scaling strategy is identifying the obstacles ahead of time so that the solution can be worked upon. A series of actions can be taken to speed up this process, such as beginning with the best use cases and then developing a playbook to lead managers through the training and development process. Then, institutional skills in critical functions like process automation and data and security analysis must be honed. Improving data delivery and quality assurance, as well as predicting the cultural shift that will occur as a result of this new working environment, will all be important. All of this, of course, will necessitate continuous assessment and monitoring to ensure that the program stays on track to meet its goals and objectives.

Also Read: Three Strategies for IT Leaders to Advance Gender Equality in 2022

Clearly, this will not be as simple as it appears. It entails not only a change in the organization’s operating model, but also a bi-directional implementation framework that allows essential activities to flow from the top down as well as the bottom up.

All of this isn’t possible on a shoestring budget. Few companies are embarking on this path from an ideal place. Most are constrained by a lack of data structure, siloed work cultures, dedicated resources, and a slew of other issues.

This transformation, however, is not optional. To delay is to risk being left behind as the world moves in this direction. As a result, the dilemma for today’s business leaders is not whether or not to scale AI, but how to do so in the least disruptive way possible while maximizing ROI.

Check Out The New Enterprisetalk Podcast. For more such updates follow us on Google News Enterprisetalk News.


Prangya Pandab

Prangya Pandab is an Associate Editor with OnDot Media. She is a seasoned journalist with almost seven years of experience in the business news sector. Before joining ODM, she was a journalist with CNBC-TV18 for four years. She also had a brief stint with an infrastructure finance company working for their communications and branding vertical.

Subscribe To Newsletter

*By clicking on the Submit button, you are agreeing with the Privacy Policy with Enterprise Talks.*