To deliver AI outcomes safely and at scale within your organization, you must first determine your AI pace.
- Gartner client? Log in for personalized search results.
Scaling AI: Strategies for AI-Steady and AI-Accelerated Organizations
By Mary Mesaglio and Hung LeHong | November 4, 2024
Scaling AI requires inventorying your organization
AI has made significant strides, with new GenAI foundation models being released every two and a half days. Despite this rapid innovation, nearly half of CIOs report that AI hasn't met ROI expectations. This dichotomy creates a unique challenge for organizations: balancing the hype and potential of AI with the reality of achieving tangible outcomes. The urgency is underscored by the fact that 74% of CEOs believe AI will significantly impact their industries in 2024, up from 59% in 2023. Needless to say, understanding and implementing AI strategies is more critical than ever.
The keynote presentation at Gartner IT Symposium/Xpo 2024 — from which this article is sourced — highlighted the dual nature of the AI landscape. With AI technologies evolving rapidly, organizations interested in scaling AI must decide whether to adopt an AI-steady or AI-accelerated approach based on their pace and ambitions.
To scale AI and fuel AI adoption, determine your AI ambitions
While your organization is likely running only the second of these races, it’s important to be aware of both.
Tech vendor race: This race is driven by relentless innovation from technology vendors, with new AI models being released every few days.
AI outcomes race: This race is about delivering AI outcomes safely and at scale within your organization. The focus here is on achieving real business value from AI investments.
Setting the pace for the AI adoption journey
Presuming you are running the AI outcomes race, start by determining whether your organization falls into the “AI-steady” or “AI-accelerated” category.
AI-steady pace
For organizations with modest AI ambitions or those in industries not yet heavily impacted by AI, an AI-steady pace is appropriate. This approach allows for a more measured adoption of AI technologies.
Focus on employee productivity: Achieving productivity gains from AI can be challenging, and employees must integrate AI tools into their daily workflows effectively.
Understand AI costs: AI investments can add up, and it’s easy to overspend. Organizations must understand their AI bills and monitor costs closely, similar to how they manage cloud costs.
Build a tech sandwich: A “tech sandwich” approach accommodates both the centralized data and AI typically managed by IT on the bottom and the data and AI coming from everywhere (business departments, enterprise software vendors, data science and engineering team, etc.) on top.
Lean into governance and trust: Establish a responsible AI team, central committee and a few communities of practice to ensure AI governance and safety. These mechanisms are crucial for managing AI initiatives effectively.
Modify change management: Be mindful of and include strategies to respond to the myriad ways in which employees react to AI.
AI-accelerated pace
For organizations with ambitious AI goals or those in industries being transformed by AI, an AI-accelerated pace is necessary. This approach requires a more aggressive adoption of AI technologies.
Go beyond productivity: Seek benefits such as process improvement, business model innovation and new revenue streams.
Implement real-time cost monitoring: Use it to manage AI expenses effectively. This is crucial for scaling AI initiatives without incurring unexpected costs.
Build a custom tech sandwich: Design a tech sandwich that accommodates AI and data from various sources. This includes building AI in-house, leveraging embedded AI and managing decentralized data.
Use TRiSM technologies: Trust, risk and security management (TRiSM) technologies are essential for ensuring safe AI at scale. These technologies enforce AI policies programmatically and in real time, beyond what human governance can achieve.
Consider behavioral impacts on employees: This includes managing emotions like jealousy and anxiety and ensuring employees are comfortable with AI-driven changes.
Scaling AI to achieve AI outcomes
Regardless of your pace, achieving AI outcomes involves focusing on three key areas:
Business outcomes: AI should deliver tangible business benefits, such as increased productivity, improved processes and innovative business models.
Technology outcomes: Prepare your technology environment for AI by managing structured and unstructured data effectively. Ensure data access rights are correctly set to avoid security issues.
Behavioral outcomes: Address the emotional and behavioral impacts of AI on employees. Involve employees in the AI journey and manage their experiences to ensure successful AI adoption.
Scaling AI FAQs
What is scaling AI?
Scaling AI — and generative AI — across the enterprise involves such steps as establishing a continuous process to prioritize use cases, creating a decision framework for build versus buy, piloting use cases for scalability, putting responsible AI at the forefront and investing in data and AI literacy. These emerging industry best practices enable CIOs to bolster their strategy and execution.
How do you scale an AI model?
Designing a composable platform architect is key to scaling an AI model. To do this:
Embrace composability in your architecture and decouple the models from engineering tools, infrastructure and the UX layer.
Determine the right model fit for your use cases based on model performance, cost of ownership, and security and privacy principles.
Invest in AI engineering tools that are agnostic to the underlying models but are tightly integrated with them.
Avoid expensive infrastructure build-outs, as well as expensive model customization upfront, unless there is a clear business case for it.
Connect with an Exclusive Community of CIOs to Kickstart Your Role
Experience the Gartner CIO Leadership Forum, an exclusive event designed for Chief Information Officers to acquire the insights needed to lead IT beyond its conventional scope, collaborate with C-suite peers, and drive business growth. Perfect for CIOs new to their roles, this forum offers unparalleled opportunities to learn from fellow leaders, engage directly with Gartner analysts, and gain access to personalized tools that ensure a fast and strong start in a new role.
Find your Conference: