Artificial Intelligence (AI) is having a big impact on how companies approach innovation, but are companies achieving big innovation impact from AI? The evidence is mixed. Is using AI to support innovation really achieving impact?
The question isn’t whether AI can have an impact. While technical challenges that may prevent achieving full value persist, many companies have realized some benefit— and often with minimal investment. It’s more a matter of whether companies are using AI properly and for use cases with the potential to drive real business value.
This year, the Boston Consulting Groups (BCG) in it’s global innovation survey digs into the critical role of artificial intelligence (AI) in innovation as in many other areas of business today.
Defining AI in innovation
Artificial intelligence (AI) is at its core a set of algorithms that classify, process, and manipulate data. “Random forests,” “neural networks,” and “generative AI” are all examples of the algorithms on which AI use cases are built. While AI is continually evolving, the taxonomy is subject to change. BCG categorizes AI use cases into five broad domains:
- Computer Vision and Hearing. These capabilities in- volve identifying objects and individuals and extracting text from images, speech, and handwriting (finding a specific element in an image, for instance)
- Natural Language Processing (NLP). This domain involves translating human (or “natural”) language into machine-understandable inputs—and vice-versa—and un- derstanding and extracting meaning from natural language (such as identifying positive or negative sentiments in user product reviews). Combining capabilities across multiple domains unlocks additional use cases. For example, apps that allow users to point their camera at a street sign in a foreign language and translate it into their own utilize both computer vision and NLP capabilities
- Machine Learning. Gartner describes machine learning, the backbone of many companies’ applications, as a set of algorithms and techniques that operate guided by lessons from existing information. Commonly used techniques are unsupervised, semi-supervised, and supervised learning
- Decision Making. This set of capabilities makes decisions based on databases of similar cases or sets of human-defined rules. Decision making can be combined with machine- learning techniques to yield a self-learning expert system
- Responding and Generating. This domain includes, among other techniques, the generative AI family of algorithms (used by such systems as DALL-E and ChatGPT). They are capable of generating seemingly new, realistic content, including text, images, or audio, from the vast quantity of unlabeled data they are trained to interpret. They also use reinforcement learning techniques (also used in machine learning), which are based on “learning by doing,” as opposed to learning by observing
More companies are Investing in AI than in any other technology today. A majority of companies (61%) have invested in AI. In fact, AI is attracting investment from more companies than any other technology.
Implementing AI with impact
In the 2023 innovation survey, BCG asked respondents to rate their company’s use of AI across six use cases:
- Revealing market trends and competitor activities (such as domains, topics, and technologies)
- Making portfolio prioritization decisions
- Identifying players with external innovation potential (such as alliances, partnerships, venturing, and M&A)
- Informing innovation investment decisions (such as starting an R&D project in a particular field)
- Identifying new innovation themes, domains, adjacencies, and technologies
- Providing input to support idea creation (such as surfacing or validating ideas)
BCG also asked respondents to rank their companies’ skill at leveraging AI from one to five, with one meaning that they hadn’t implemented that use case and five signifying they had implemented the use case with impact.
The top three use cases with the most AI impact are making portfolio prioritization decisions (40% of companies implementing AI for that use case achieve impact), identifying players with external innovation potential (38%), and revealing market trends and competitor activities (37%).
The use cases with the most untapped potential across all industries include identify-ing players with innovation potential, supporting the idea creation process, and identifying new innovation themes.
Four Impact Success Factors
Is using AI to support innovation really achieving impact? AI alone won’t supercharge an innovation program. Companies must have the right platforms and practices in place. Companies that successfully scale AI typically dedicate 10% of their investment to algorithms, 20% to technology, and 70% to people and embedding AI into business processes. AI can also accelerate the ability to access external sources of input, speed the generation and evaluation of ideas, and inform decision making. Through research and case work, BCG identified four key factors to help companies successfully incorporate AI into their innovation efforts.
A Strong Digital Backbone and Data Pipeline. To enable AI, companies must have established digital and data capabilities. Leaders make data and technology accessible across the organization, avoiding siloed and incompatible tech stacks and standalone databases that impede scaling. BCG’s research into AI scaling has shown that companies that make more than 75% of technology and data widely available have a 40% greater likelihood of realizing AI use cases at scale than those that make 25% or less widely available. The survey reveals that among companies that have invested in AI, those that have successfully implemented AI in their innovation processes are 1.5 times as likely to prioritize full data transparency and effective digital tools for decision making. They are also twice as likely to have an active digital innovation laboratory, and they invest in an average of 2.5 times as many complementary technologies (such as IoT and augmented reality). These complementary technologies in turn generate data to enhance AI algorithms, creating a positive feedback loop as algorithms learn from what they do.
Multiple Technology Initiatives. Companies that achieve the most impact from AI also invest in other advanced technologies and combine these initiatives with AI. Since AI algorithms learn from what they do, uses cases that require technologies such as robotics, automation, or high-throughput sensors that facilitate data collection fuel the AI learning and development process. The most common technologies to pair with AI investment are robotics for process automation and IoT. The success rate from investing in additional technologies rises fast: each additional technology increases the likelihood of realizing impact from AI by 26%.
A Robust Innovation Funnel. Companies that realize impact from AI become idea generation powerhouses with robust innovation funnels that both incubate new ideas and rigorously prioritize investment. These firms generate more than five times as many ideas than others and incubate more than twice as many minimum viable products. With more than three times as many people working in innovation- related functions, these organizations apply demanding feasibility criteria for new ideas and are twice as selective as companies that don’t realize impact from AI. The best AI innovators develop a virtuous cycle or flywheel effect: more ideas mean greater likelihood of finding the best use cases for AI, and implementing AI helps generate more ideas. Some companies have begun incorporating generative AI into the idea creation phase, leveraging the technology to stimulate their teams’ creativity.
Clear Articulation of Purpose and Responsibility. Visible leadership from the top matters. BCG’s research shows that when the CEO personally articulates the purpose of innovation for the organization, the likelihood of the company achieving impact from using AI nearly doubles. Driving impact from AI includes not only the immediate results, but also the downstream implications. Companies need to factor into their decision making how the use of AI will play out over time. Leaders are already instilling the policies and practices of responsible AI (RAI). Successful RAI begins with a set of clearly articulated principles and is operationalized by translating principles into policies and training, establishing clear governance mechanisms, building tools for end-to-end case reviews, and integrating RAI considerations into existing tools and methods. RAI is not just a risk mitigation tool but a mecha- nism to accelerate innovation. By providing a framework with clear guardrails, RAI allows innovators to experiment with confidence and catch failures faster.
Is using AI to support innovation really achieving impact?
Given the accelerating pace of advancement in AI, it is imperative for corporate leaders to establish mechanisms to monitor these new capabilities, evaluate risks and opportunities for their business, and ensure that they are leveraging the technology in the most promising use cases.
