AI stands out as one of the most promising tools to revolutionize operations, customer engagement and innovation. McKinsey estimates show the application of generative AI across commercial, R&D, operations and support functions in energy and materials can create anywhere from $80 billion to $140 billion in value.
Generative AI and other AI technologies can offer immense benefits, from advanced product recommendations for field sales to predictive maintenance for operations. However, for chemical and ingredient industry companies, there are many traps along the journey toward successful AI adoption.
Let’s explore the common pitfalls of AI adoption and how chemical and ingredient companies can navigate them effectively.
1. The Data Dilemma: Quality Over Quantity
AI systems are only as good as the data they are built on. The chemical industry’s reliance on unstructured data — scattered across PDFs, legacy systems, and email threads — can make it challenging to implement AI solutions effectively. Poor data quality or lack of centralization can lead to “hallucinations,” where AI generates plausible but incorrect outputs.
A Harvard Business School study found that test group members using AI were 19% less accurate than group members not using AI on tasks outside current AI capabilities. However, when the data foundation for an AI application is good, generative AI use can boost productivity up to 40% (Source: MIT Sloan).
Vertical AI software, where the data set and LLM training is tailored to a specialized field such as legal, finance or chemicals and ingredients, increases accuracy.
Solution: Establish a clean, harmonized data foundation with structured data management systems. Tools like Knowde’s Master Data Management (MDM) Platform can organize and centralize data, ensuring accuracy and consistency.
“I think chemistry is a killer application for AI. In some ways, it’s a big search problem, right? We’re searching over the possible ways to make a molecule. There’s a huge number of things you can explore. . . . It’s also a multiparameter optimization. You want your molecule to be good at this and this. That’s something that computation and AI are very good at doing: predicting across many diverse things simultaneously.” Iambic Therapeutics cofounder and CEO Tom Miller in an interview with C&EN
2. Overestimating Horizontal AI Tools
Generic AI tools trained on public datasets often fall short in specialized fields like chemicals. These tools struggle to process domain-specific data or handle complex workflows unique to the industry, such as regulatory compliance and intricate product attributes.
Horizontal AI applications are designed and trained on public data. They’re great for summarizing notes or rephrasing an email to sound more professional. The gap arises when you need to ask specific questions about your proprietary product data.
The gap between horizontal AI capability and real-world application also shows up when you need to automate a technical process that is unique to your industry. The automation needs to run on a data set that is not publicly available.
A great vertical AI tool needs to be built by people who understand the industry, the specific data sets and the complex operational processes that the tool needs to support.
Solution: Invest in vertical AI solutions tailored to the chemical sector. These models leverage industry-specific datasets and expertise, ensuring precision and reliability for tasks such as product development and regulatory compliance. Knowde AI is built and supported by a team of information specialists, chemical engineers and PhD chemists to organize complex product and technical data for the chemical and ingredient industries. This tool is the foundation for Knowde products such as Knowde Master Data Management Platform and Customer Experience Platform.
3. Ignoring Change Management
Introducing AI systems isn’t just a technical shift — it’s a cultural one. Resistance from employees, especially in data-driven roles, can slow adoption. AI tools may fail to integrate seamlessly into existing workflows without proper model training and support from the subject matter experts using the tools.
“The major barrier is to support change management. Most of the new technologies represent a clear advantage compared with previous implementations.”
Solution: Prioritize training and change management. Engage teams early, provide hands-on training, and highlight the long-term benefits of AI to gain buy-in across departments. As with all good change management programs, the key is clear and frequent communication across the organization. That communication should be two-way, including feedback and ideas from staff up to the executive level.
4. Misaligned Use Cases
One of the most common missteps in AI adoption is focusing on applications that fail to align with business goals. Many companies start with customer-facing tools like websites without addressing foundational issues such as data integration or internal inefficiencies.
The AI was implemented to take drive-thru orders in a series of tests. The AI made mistakes like asking if people wanted 260 chicken nuggets. This added frustration and extra time to the customer experience, as the existing one was already a well-oiled machine. McDonald’s and other fast food chains will still pursue AI implementation, but they’re re-examining which use cases are worth the time and effort.
Solution: Start with high-impact use cases. For instance, centralizing product data or automating repetitive tasks deliver quick wins and pave the way for more complex AI initiatives.
5. Talent Shortages
Implementing and managing AI systems requires specialized expertise in data science, machine learning and the chemical domain. Talent shortages can lead to suboptimal deployment or reliance on inadequate third-party solutions.
Solution: Collaborate with AI vendors that specialize in the chemical industry (like Knowde!). Their pre-built solutions and professional services can offset internal skill gaps while delivering tailored results. In Knowde’s case, we are built for chemists by chemists.
6. Overlooking Regulatory Implications
The chemical industry is heavily regulated, and AI systems must meet stringent compliance standards. Improper data handling or reliance on inaccurate outputs can lead to costly violations and reputational damage.
Solution: Choose AI systems designed to handle compliance documentation and regulatory data. These tools should provide traceability and transparency, ensuring outputs meet regulatory standards.
7. Underestimating Long-Term Costs
The initial setup costs for AI solutions often overshadow the hidden expenses of maintenance, integration, and scalability. Custom-built solutions, in particular, can spiral into resource-intensive projects. According to Upsilon IT, building an AI can cost from $5 – 500K, depending on the complexity and scope of the project.
Some of the hidden costs of building and training your own AI based on a foundational model like ChatGPT are:
Cost of Talent: AI model and prompting experts are in high demand and are costly to onboard and retain. The odds of finding specialists who are also familiar with or experts in chemical and ingredient industry data are slim.
Maintenance: Infrastructure costs and ongoing testing and maintenance can easily cost $10K annually.
Scalability: An increase in use volume or complexity and length of input data will raise the number of tokens needed (and therefore cost), plus the additional cost of maintenance and testing for added functionality.
Solution: Opt for subscription-based AI solutions that offer scalability and built-in maintenance. Platforms like Knowde’s enable companies to avoid the pitfalls of custom builds while delivering faster ROI. You get the benefit of a system that’s tailor-made for the complex needs of the chemical and ingredients industries without the price tag of a custom-built project.
Strategic AI Adoption for Sustainable Growth
AI holds transformative potential for the chemical industry, from automating customer support to driving innovation in product development. However, companies must address these pitfalls proactively. By focusing on structured data, choosing the right AI solutions, and fostering a culture of adaptability, chemical companies can harness AI to achieve operational excellence and competitive advantage.
Your central source of truth for master data.
Easily manage your clean master data from within your dedicated data platform.
