Executive Summary
The majority of TLT readers believe that artificial intelligence (AI) will make a significant impact on lubricant research and development (R&D). AI has the potential to improve efficiency, and data management and modeling are particularly promising applications. However, there are still barriers to the adoption of AI, and many are related to the quality of the data. Human subject matter experts (SMEs) will continue to be critical to effectively using AI as a tool for lubricant product development.
Q.1. What opportunities do you foresee for the use of generative artificial intelligence (GenAI) in the R&D process of lubricant product development?
In research areas AI can be used for data processing, results extrapolation and modeling. This can eventually be used for product development.
It could lead to major R&D leaps in lubricant formulations.
I do not see any sensible opportunity.
New base oil synthesis, additive component selection, tribological test selection and data analysis.
I’m not sure—generative AI involves some level of information curation and due to the proprietary nature of lubricant formulation, I’m guessing most of the available information would be in house.
Questionable.
Cutting the redundant recording and clerical elements of the development process could be a benefit.
Remote collection of data on the properties of lubricants used and signals correlated with the condition of distributed devices (may constitute dedicated research tests) in which they are used.
Reduction of time for processing experimental data and formulation according to specified parameters.
This will potentially revolutionize the way in which lubricants (and coatings and materials) are designed.
Can help in designing formulations which are more sustainable, like finding alternatives for actual biocides.
Potential for calculating predicted properties for lubricants under different conditions (lubricant viscosity for gas compression) and maybe assists in some formulation development work in the future.
Assisting with admin and reporting. Sorting through files and assisting in digitization of historical records and screening and summarizing.
The easy answer would be that generative AI opens up a new world of creativity for those people that are responsible for formulating new products.
Gathering and summarizing data that has been generated over the past 100 years and exists in various documents and formats.
GenAI provides great benefit to someone new to an R&D role. Getting a basis of formulation or providing understanding of interactions can give R&D workers an individual path to get the majority of start work done quickly. Of course, any AI generated understanding requires confirmation and investigation in its own right.
GenAI is very useful for developing new formulations, though basic understanding of the greases is required and do need SMEs to verify. It is a great starting point and reduces the number of trails. Developing standard operating procedures (SOPs), method of manufacturing and data analysis are other uses. GenAI helped immensely in enhancing product quality.
Predictive four-ball wear test results.
Almost unlimited opportunities to replace chemists with technicians.
Right now, GenAI can alleviate writers block when starting new work items. I have found it to give reasonable first drafts of white papers, proposals and executive summaries if the input parameters are sufficiently specified. The level of true technical expertise and correct language falls short of humans, so it cannot be relied on for final product submissions.
Might create some efficiencies concerning formulation development. Do not expect any significant help creating new molecules.
Medical services, manufacturing.
Not much at the moment. There’s a lot of work to do first on site with other significant problems.
Is the use of digital technologies such as GenAI in lubricant product development worth the hype?
Yes
74%
No
26%
Based on an informal poll sent to 15,000 TLT readers.
GenAI can vastly improve the R&D process of lubricant product development. It can take some guesswork out and guide the formulators and product developers to make decisions with running all the needed tests. It can also predict the impact of lubricant components on their performance in different tests. GenAI can help reduce the number of failed tests and associated costs and make the product development more economical. GenAI can also help meet the regulatory standards for lubricant components.
Product development.
To connect knowledge.
1.) Data analysis. 2.) Product development by using design of experiments (DOE) method—additionally if we use AI, we can save more time for developing optimal products. 3.) Generating experimental reports and other documentation.
We see opportunities especially in test data interpretation, not so much in the development phase itself (as most AI will in most cases base itself on history, or generally available information).
Time management and quality improvement.
Big data analysis and simulation models.
Predicting physiochemical and performance data from historical data, reducing the need to conduct the tests again, thereby shortening the product development times. Formulations can be predicted and generated using GenAI using suitable systems.
GenAI has a broad range of R&D opportunities. Think of it as a Google search on steroids. Unlike Google, Yahoo or literature searches, GenAI is more interactive and can be more targeted in answering the R&D research topic. The broad nature of GenAI may also provide R&D paths that were not obvious at first glance.
Not many. Any information that would be helpful in the development of lubricants would still have to be tested and proved before it could be used.
I see a lot of exciting opportunities for GenAI in lubricant R&D, especially in accelerating innovation and improving efficiency. One major advantage is formulation optimization—AI can analyze vast datasets of base oils, additives and performance characteristics to suggest new formulations faster than traditional trial-and-error methods. Another key area is predictive performance modeling. Instead of relying solely on physical testing, GenAI can simulate lubricant behavior under different conditions, helping R&D teams refine products before they even hit the lab. This could significantly reduce development time and costs while improving overall performance. I also see potential in failure analysis and troubleshooting. By processing historical test results, AI can identify patterns and predict failure modes, guiding engineers toward better solutions more efficiently. Additionally, GenAI could enhance technical documentation and regulatory compliance by automating reporting and ensuring formulations meet industry standards.
Q.2. What could be some of the barriers for adoption for the use of GenAI in lubricant product development?
Lack of computational models to pre-screen the performance of the suggested chemistries prior to acquisition of the most promising candidates and carry out of the experimental studies.
The method is not causal and not very useful in scientific research.
Company policy on adoption of AI, preparatory data safety.
Proprietary information, patents, trademarks, etc.
AI will never replace brilliant professionals as they research and develop lubricants for new challenges.
Lack of trust on the part of device owners and operating services, rapidly changing software and failure to keep up with appropriate employee training, as well as possible limitations in available energy and compatibility of devices produced by different manufacturers.
Unreliable and incompetent information provided by GenAI.
Data sharing.
Need to be trained on a lot of data. Data needs to be available and well organized and labeled.
Obtaining a data set of only good data without bad data. It is hard to always capture in datasets why certain formulas pass or fail certain tests that may be due to variables outside of the formula (contamination, blending procedure, minor differences in testing procedure, additive variance, etc.).
It is wrong a lot. Can’t be trusted.
Garbage in, garbage out. Creation (on paper/digitally) is not the same as creation on the bench. Generative AI is a computer code and has no consciousness, experience or creativity.
There are a lot of marketing claims in the public domain which simply aren’t fact based (or are embellished). The vast majority of data published is just the “good” data that supports the company writing it.
In lubricant product development or any field, new technology is slow to adopt and comes with failure. The earlier failures are bound to create more pessimistic viewpoints and barriers to AI adoption than any early successes it may have. On top of that, the managers and team leaders are often veterans in their field and are aware of the methods they used to create success over their career—why take the risk when you know of a proven method?
Prompt engineering and GenAI responses that don’t align with what the user thinks can create a gap.
Not having enough data to make a good AI-driven result.
Metalworking lubes have too many variables that cannot be fully tested in the laboratory. Will still need experienced chemist guidance.
Issues I have found from a technical standpoint are inventing false technical references and misinterpreting technical data, i.e., drawing incorrect and misleading conclusions.
Loading the data that allows systems to develop models in a more efficient way than those used today. Improvements would be incremental and even the best models today need validation.
None.
Lack of data. First, we need to have lots of information regarding specific situations we are trying to solve.
What use cases do you think GenAI has the opportunity to have maximum impact?
Unlocking knowledge from historic data in different formats (files, images, tables, databases, etc.)
88%
Pre-testing virtual experimentation recommendations
53%
Auto generating reports, tender or OEM approval documentation
53%
Predicting engine tests or field trials
38%
None of the above—GenAI doesn’t have a place in lubricant product development
8%
Other not mentioned above
23%
Based on an informal poll sent to 15,000 TLT readers. Total exceeds 100% because respondents were allowed to choose more than one answer.
One barrier is the lack of stored historical data in a structured form so that it can be extracted and used in GenAI. The historical data may not be in a digital form. These historical data need to be formatted in an electronic form to be used in GenAI. There can be hesitancy to adapt to GenAI and it may be considered a threat to the jobs. There would be an apprehension until the GenAI produced products that meet the specs and regulatory standards.
Trust. Intellectual property (IP).
Complex human neural network.
None.
AI does not have the capability to make the right judgement of the experimental results.
Finding data analysts with tribological knowledge is challenging.
In the initial phase, upgradation of the machinery.
The lack of AI regulations. There is also a loss of opportunities for trainees and junior employees to build experience due to the use of AI.
Lack of awareness and understanding among users. Also those working are worried about secrecy of the data that will be used for generating the insights.
GenAI is too broad, unless supervised. It could be useful, but there is a trust issue that should be mitigated for the task at hand.
The downside of GenAI is that there is no verification, so the information retrieved must still be understood and verified. GenAI cannot perform physical testing, which must be conducted outside of AI. If testing and verification require more effort than GenAI uses, then it will not be time effective.
Many, I have more confidence in our engineers and technicians than GenAI. Education, experience and common sense are very important in the development of new products or improvements to existing products. Another concern is the amount of deceit that is in the lubricant market. Many lubricant companies have very appealing advertising, which lures people into buying products that do not live up to the hype that these companies are putting out. How much worse could it be with GenAI generated information that would be misleading?
Not enough allocated budget. The significant risks around data security, intellectual property protection, regulatory compliance and privacy concerns that can dramatically impact an organization’s willingness to fully embrace GenAI technologies without proper safeguards in place.