Generative artificial intelligence (AI) has garnered significant attention for its remarkable advancements, leading some executives to mistakenly believe that all existing AI technologies will soon become outdated. This focus on generative AI to the exclusion of other forms can be detrimental, as it results in the segregation of AI expertise and resources, ultimately constraining the full potential of AI since generative AI alone cannot address every type of problem.
The future of AI evolution does not solely rely on groundbreaking technological innovations but rather on the adoption of a unified strategic approach to AI, known as a One-AI approach. This approach involves integrating the latest generative AI with other AI forms to achieve outcomes that surpass what each can accomplish independently.
Fundamentally, all modern AI technologies share the core capability of recognizing and learning from complex data patterns. Their varied applications stem from their suitability for different use cases. For example, large language models leverage pattern recognition to predict the next likely word, primarily used in content creation and creative problem-solving. In contrast, predictive AI, which has been in existence for over a decade, leverages historical data to forecast future events, anticipate behaviors, and provide recommendations for informed decision-making.
The synergy between different AI forms is essential, as highlighted by Ken Moore, Mastercard’s chief innovation officer, who advocates for combining generative and predictive AI to achieve superior outcomes. However, the compartmentalization of AI resources by companies hinders this synergy, slowing down AI adoption and jeopardizing investments in predictive AI despite its demonstrated return on investment.
The future of AI deployment is envisioned in the form of end-to-end systems capable of performing a wide array of tasks seamlessly. Leading companies are already preparing for this shift, with mature AI firms being twice as likely to embrace a One-AI approach to scale applications, according to a December 2023 BCG survey of C-suite executives. Organizations that fail to establish the necessary architecture to support a One-AI approach risk falling behind once this integrated strategy becomes mainstream.
Insilico Medicine, a biotech company, exemplifies the efficacy of the One-AI approach by accelerating drug development while reducing costs significantly. By combining generative and predictive AI, Insilico achieved the milestone of designing the world’s first AI-developed drug in just 18 months at a fraction of the traditional time and cost. This success underscores the complementary nature of predictive and generative AI in expanding the realm of possibilities for innovation.
For companies aspiring to adopt a One-AI approach, understanding the diverse modes of AI interaction is crucial. The three primary modes include sequential mode, feedback-loop mode, and standalone mode, each offering unique benefits depending on the business problem at hand. Companies must choose the most suitable mode of AI interaction to address specific challenges effectively.
To embrace emerging best practices in One-AI implementation, companies should focus on unifying AI teams, employing model-agnostic problem-solving approaches, ensuring data integrity during AI interactions, and managing risks associated with One-AI solutions. By organizing around a unified AI strategy and staying adaptable to technological advancements, companies can harness the full potential of AI technologies and drive innovation in the age of permanent AI revolution.
François Candelon, Leonid Zhukov, Namrata Rajagopal, and David Zuluaga Martínez are experts at Boston Consulting Group (BCG) and contributors to the discussion on the strategic integration of AI technologies through a One-AI approach.
