Accenture’s report states that AI will be necessary to improve labour productivity in the chemicals industry
CHEMICAL engineers’ workloads will be transformed by generative AI, allowing workers more time for “complex judgemental tasks”, according to a report published by Accenture.
Several chemical companies, including BASF and Dow, have launched AI projects in recent years as part of their customer service output and research and development efforts.
Accenture says that AI will be necessary to improve labour productivity in industry and bypass the challenge of talent shortages, with around 30% of the current workforce expected to retire in the next five years.
The IT and consulting firm recommends companies use AI language models and machine learning to assist engineers and plant operators, and record and preserve expertise from retiring senior engineers, using it to build training models.
A change in workload
The report found production workers, including planners and supervisors, at chemical companies spend 90% of their time on simple judgement tasks in plant operation, like admin and documentation.
For planners, around 57% of their roles have the potential to be automated, and a further 15% could benefit from augmentation.
Automation relates mainly to the work of language models and condensing down large datasets like machine logs and production tables.
Augmentation involves an extra layer of sophistication, where operators would be able to ask questions of AI based on recorded data.
Stuart Prescott, an associate professor at UNSW Sydney’s School of Chemical Engineering, said: “Instead of having engineers sort through thousands of documents to understand a process, these documents can be given to a language model with a knowledge base to generate a response based on private company data.”
He added: “I imagine big companies will be building a team to test the waters and see what is possible with AI on a plant scale.”
Further research from the report shows that automation and augmentation will affect around 31% of working hours in the industry.
Prescott said that AI could help optimise plant operations by taking on vast volumes of sensor data, that would overwhelm human cognition, and allow more control over measuring and managing the risk of processes.
Sustainability is a conflict of interest
Accenture’s report states that AI would be essential to meet the global demand of sustainability-related products, including materials for and solar panels, which is expected to be worth US$570m by 2028.
However, there is a worry that the demand for data centres to facilitate AI eclipses the emission targets of many businesses – data centres and data transmission networks are responsible for 1% of the world energy-related greenhouse gas emissions.
Last month, Microsoft revealed that it was not on track to meet its 2020 pledge to become carbon-negative by 2030 due to rising Scope 3 emissions. These were “primarily” from the construction of data centres.
Prescott said the use of AI could address this problem to some extent thanks to its success in forecasting demand.
He said: “A lot of chemical processes are already automated and most plant control rooms are computerised.”
He added: “AI models are good at predicting and optimising processes, so there is a space where emissions can be paid back by first improving energy efficiency at plants.