The CPG AI Gap: High Adoption, Low Impact

Large Consumer Packaged Goods (CPG) companies are adopting AI at scale, but the technology is failing to transform product innovation. While 9 out of 10 R&D executives in large CPG firms use AI in some capacity, only 19% have embedded it into routine workflows, according to a new survey by Turing Labs.

The adoption numbers suggest progress, but the utility is concentrated in low-stakes tasks. Of the 290 senior R&D leaders in the US and Europe surveyed, 91% are either using AI routinely or actively piloting/scaling it. However, most of this activity remains limited to concept generation, basic research, and summarization—areas where generic Large Language Models (LLMs) already perform well.

The real value is currently appearing in marketing, where firms use AI to reduce ad agency spend and increase in-house management. This does not translate to product innovation.

The disconnect between deployment and results is visible in the failure rates of substantive initiatives. The Turing Labs survey shows:

• 20% of AI initiatives are never implemented.
• 14% are implemented and later abandoned.
• 24% remain in use but deliver minimal or no business impact.

The barrier to entry for high-value innovation is the complexity of the domain. For many, GenAI formulation outputs are too generic to use without significant manual re-work and validation, a sentiment shared by 57% of respondents.

Effective formulation requires more than content generation; it demands integration of ingredient chemistry, regulatory constraints, manufacturability, cost, taste, texture, shelf life, and nutritional targets. Generic LLM approaches often fail this test. When users attempt to process thousands of technical white papers through ChatGPT, they frequently encounter inconsistent responses to the same questions.

The failure to move the needle stems from a fundamental strategic error: treating AI as a way to accelerate existing processes rather than a tool to restructure how innovation is organized. Companies that use AI merely to speed up current workflows will miss the broader benefits of the technology.

If you are leading an R&D team, ask whether your AI investments are simply automating old tasks or if they are equipped to handle the specific chemical and regulatory complexities of your product pipeline.

Source