Embracing the Power of AI in R&D: A New Frontier for the CPG Industry

In today's rapidly evolving world, the Consumer Packaged Goods (CPG) industry faces increasing pressure to innovate and stay competitive. The rise of Artificial Intelligence (AI), provides a game-changing technology that has the potential to revolutionize research and development (R&D) processes across the CPG landscape. However, for many of these R&D teams, there is confusion around what AI means for them, as well as concerns and doubts about embracing AI. In this blog, we will delve into the world of AI, address fears and concerns, and set realistic expectations for CPG companies considering adopting this transformative technology.

‍

Challenging the status quo

While most R&D teams have established some foundational processes for leveraging AI successfully, it's important to recognize that classic methods like designs of experiments (DOE) are limited in their adaptability. Unlike AI, which operates as a closed-loop system, traditional DOE follows an open-loop technique. This implies that a predetermined set of experiments is carried out, and outcomes are observed and analyzed, but without the incorporation of feedback to guide subsequent experiments. Once this set of experiments concludes, the process ends.

The key to AI lies in its capacity for data and continuous learning. The model's proficiency grows with more data, resulting in better accuracy over time. Through ongoing learning and adjustments based on observed outcomes, AI can rapidly converge toward optimal solutions, particularly in intricate and uncertain contexts. Classical DOEs, while retaining value in various scenarios, may necessitate more rounds of experimentation to attain similar levels of optimization. This has posed a notable challenge for numerous R&D teams we've engaged with.

Within the realm of R&D teams, some have an abundance of data, but struggle with the task of organizing and processing it. Meanwhile, others possess limited data and might dismiss AI platforms as unfitting. A pivotal advantage of AI platforms lies in their forward modeling ability. While they effectively utilize historical data to build robust models, they also accommodate organizations that struggle to harness their existing data or possess meager amounts. This proves to be a substantial competitive edge compared to non-AI platforms. As emphasized before, continuous enhancement is intrinsic; modeling forward integrates data from ongoing formulation runs to progressively fortify the platform's robustness.

Though certain R&D teams conduct an extensive array of experiments throughout a year, the reality is that 1) not all of this data is captured digitally, and 2) often only successful experiments are documented. For a model to genuinely learn, it must be exposed to the complete spectrum of data—both successes and failures. Consistent data ingestion and practice are imperative for sustained improvement over time.

‍

The misconceptions about AI

For many organizations, there are a lot of misconceptions, fears, and anxieties about implementing AI into their business.  That is why education is important to help these organizations overcome their fears. Let’s discuss the reality of some of these misconceptions.

Job displacement: One concern often associated with AI is the fear that it will replace human jobs. This can be particularly concerning for CPG R&D product developers who have been reformulating products for 20-30+ years.  All of that knowledge and experience can’t be replaced, and in some instances that knowledge will be able to translate success or understand issues, faster than AI. While AI can automate certain tasks, it is important to note that it is designed to augment human capabilities, not replace them. By leveraging AI employees can focus on more complex and creative tasks, leading to higher productivity and innovation.

Complexity and Cost: Integrating AI into existing R&D processes, or simply using the technology may seem complex and costly. However, advancements in technology and the availability of user-friendly AI tools geared specifically towards CPG R&D teams and built to support the way they work and think, have made adoption and implementation more accessible. Start with small-scale pilot projects to demonstrate the value of AI before scaling up.

‍

The Benefits of AI in R&D for CPG

The benefits of implementing AI into an R&D’s teams process, procedures, and culture can be endless, but there are a couple of areas where we have seen some consistent success.

Overcome Resource Limitations: R&D teams are overloaded with projects and requests, and in many cases, they lack the resources to complete these projects and requests.  This could be a lack of physical people, limits in funding, or even a lack of ingredients and supplies. AI can help overcome these limitations by allowing teams to conduct virtual testing and simulations- explore multiple scenarios without the need for physical prototypes and extensive lab work.  Thus, saving time, reducing costs, and optimizing their resources.  

Accelerated Insights: AI can process vast amounts of data, extract patterns, and generate actionable insights at an unprecedented speed. AI can create models that optimize formulations, by suggesting the most promising ingredient combinations and concentrations. This can reduce the number of experiments required to achieve meaningful results.  For example, instead of running 25 weeks of experiments and tests, AI can help achieve the same results with 2 weeks of testing. This expedites the R&D cycle, enabling companies to bring innovative products to market faster, gaining a competitive advantage.

‍

“The AI digitization of the R&D formulation process will disrupt the industry.  Those that do it will see breakthroughs in formulation in terms of product performance, speed to market, product cost, and R&D productivity.  It has the potential to rewrite who are the winners and who are the losers in the ever-competitive CPG categories for product performance, product cost, market share, and top and bottom line performance.  Those that don’t will simply get left behind!” - Alan Maingot, former CPG R&D Executive

‍

Setting Realistic Expectations

While AI offers significant benefits, it is crucial to set realistic expectations. For many CPG R&D teams that are new to AI, there is a perception that with a click of a button, AI will provide all of the answers they need or want.  Unfortunately, that is not the case.  Here are some considerations for successfully leveraging AI for your R&D needs.

Continuous Learning: As I mentioned earlier, data is the key to any successful implementation of AI.  The more data that is fed to an AI model, the more it learns and the more accurate it becomes.  Even in the world of LLM (large language models) like ChatGPT, they just didn’t happen overnight, it took time for those models to learn. That is why it is critical to capture ALL of your data from current and past testing and experimentations as I stated earlier.

AI systems improve over time through continuous learning. Initial results may not be perfect, but with feedback and refinements, AI models become more accurate and reliable. Patience and ongoing investment in AI capabilities are key to unlocking its full potential.

Human Expertise is Vital: AI should be seen as a tool that complements human expertise, rather than a replacement. Back to the example earlier of R&D formulators with 20-30+ years of expertise, their knowledge is irreplaceable, but it can be augmented. The human touch is still essential for decision-making, creativity, and critical thinking, but it also allows for the capture and input of tribal knowledge that an AI system may or may not understand or know. Collaboration between AI systems and human experts leads to the best outcomes.

Have a Goal in Mind: As I stated earlier, AI isn’t a magic button that can answer all of your questions and solve all of your problems.  Too often teams want to use or leverage AI, but don’t know where to start, and when they do, they don’t know what success looks like.  What this means is that any implementation of AI is set up for failure, resulting in unmet or unrealistic expectations.  That is why it is important to look at those AI projects that are measurable and repeatable- this will ensure whether it was successful and if continued success is possible.

‍

AI represents a remarkable opportunity for the CPG industry to transform R&D processes, drive innovation, and deliver superior products. By addressing fears, and concerns, and setting realistic expectations, businesses can start to better leverage AI to better impact their business.