Harnessing the power of generative AI: an agile methodology for uncertain times has been saved
Article
Harnessing the power of generative AI: an agile methodology for uncertain times
Agile strategies for the AI era
Generative AI developments are going fast - and the pace keeps accelerating. There is no way of knowing what will happen next, and when. Challenging as this landscape of uncertainty may seem, we will all have to deal with it, and jump on the bandwagon now if we don’t want to lag behind in the future. But how to avoid the first mover disadvantage? The answer: with a structured approach to AI Agility.
A landscape of uncertainty
In the past decades, many of us believed that Artificial Intelligence (AI) would take off any minute – but it didn’t, at least not way that exceeded high expectations. However, AI has now reached the stage of exponential growth. The pace of development keeps increasing. How and when should we react? Some people claim that they know where things are heading, but honestly unless you are in the Boardroom of some of the companies driving the change, there is no way of telling. Even the smartest and best informed experts on the globe cannot predict the state-of-the-art in six months, or next year – let alone beyond. In fact, as humanity, we are right in the middle of a landscape of uncertainty.
Avoiding the first mover disadvantage
Daunting as such a landscape may seem, action is required. Yet, if your business starts out with the mindset to adopt AI options that are available today, you will run the risk of the first mover disadvantage. After all, a year from now, or even sooner, there may be many more (affordable) options to choose from. Or worse. We have all heard the stories of early adopters of OpenAI, who incorporated the GPT-3 text-davinci-003 series (one of the early OpenAI models) and then suddenly found themselves in a position where that model was no longer supported, leading to severe drawbacks in terms of time and investment.
An agile approach
At Deloitte, we keep a realistic take on the latest developments. We know what we don’t know (yet), what insights and knowledge are currently lacking. Fortunately, we also know how to deal with such gaps and uncertainties: with an agile approach to AI, or AI Agility as we call it. AI Agility refers to an organization’s capacity to rapidly respond, adapt and iterate whenever AI advancements emerge.As a frontrunner in the field of AI (and other new technologies), we have been quick to adapt this approach and use it to help our clients.
The AI Agility Approach: Five Core Pillars
Our AI Agility approach is a structured matrix with five core pillars (see figure 1 below), comprising specific topics. For instance, the Technology and Infrastructure pillar consists of model development, training, deployment, energy consumption, hardware, and partnerships with AI vendors to support AI initiatives. The Data pillar includes effective data management, data security, ensuring access to high-quality data sources, and public data vs. third-party data.
The AI Agility Approach: five stages
Each of these pillars progress through five distinct stages:
- Sense and Evaluate: Understand the needs of your business to identify and evaluate potential risks and opportunities in the AI landscape in relation to these needs.
- Validate: Based on your choices, create a pilot/proof of concept and validate it for feasibility, viability and desirability to decide on go or no-go.
- Design: Use lessons learned during the development and validation stages to craft a strategy/framework for responsible implementation, including safeguards, ethics, and guidelines for responsible use.
- Adapt: Implement, manage necessary changes within the organization and enable a smooth transition and adoption of the new opportunity.
- Monitor: Assess success and impact continuously through performance monitoring, data collection, and optimization.
Staying ahead of the competition
Going through these stages – continuously – will help you make the right choices for each of the pillars. The approach ensures that your organization and your AI solution can thrive in a rapidly changing world. Instead of relying on one particular vendor or model, the framework is modular and its elements are easy to replace. In the case of GPT-3, this agile, data-driven and modular approach,without reliance on one particular vendor or model, would have prevented the drawbacks of early adoption. That is why the AI Agility Approach is a key tool for businesses in the digital era that enables them to rapidly adapt to evolving AI technology, integrate AI seamlessly, stay ahead of the competition, drive innovation, and improve efficiency.
AI and Deloitte
At Deloitte, we recognize that AI offers a new kind of collaborative intelligence. We aree working alongside clients to ask the insightful questions that reveal how AI can reimagine business models, build value, and inform their vision now and in the future.
At the same time, we’re working with leading and emerging tech companies to develop industry use cases and frameworks that advance ethical and agile human-AI interaction. Also, we have established an AI Institute as well as an AI Center of Excellence and a GenAI Lab. This enables us to discover new AI opportunities for our clients and for society at large.
Going forward: industry-specific approach
This article is an introduction to AI Agility and the AI Agility Approach. Obviously, every industry and every individual business requires a more refined and customized approach. In the coming months we will be sharing industry-specific information about AI Agility. If you would like more information about the AI Agility Approach and how it could help your organization thrive in a data-driven world, please get in touch with us.
