Manufacturers See Low Success Rates for Early Gen AI Initiatives

Last year, manufacturers went full steam ahead on generative AI . This year, many lack the same enthusiasm.

How Etsy Approaches AI--Both Internally and Externally

Just one in every five generative AI initiatives have been successfully deployed by manufacturers, according to new data from Lucidworks. That’s lower than other industries, wherein one in every four generative AI initiatives has been deployed with some success.

About half—48 percent—of manufacturers using generative AI have reported higher cost savings because of those technology-led initiatives, the research shows. That’s just higher than the industry-agnostic average, which shows 42 percent of organizations have seen significant benefits from their work with generative AI.

According to Mike Sinoway, CEO of Lucidworks, manufacturers’ most successful generative AI initiatives have been those they’ve deployed to help with operational expenses.

“One of the most common gen AI applications appears to be in monitoring invoice elements and purchase price variance. [Manufacturers] seem to have plans to use the AI tools to automatically respond to vendors with requests for discounts or credits,” he said.

But despite those early results, which have been slow but show promise, manufacturers’ interest in increasing spending for generative AI has tanked between 2023 and 2024. In 2023, 93 percent of manufacturing leaders reported they had plans to increase investment in generative AI. In 2024, that figure dropped to 58 percent.

That figure shows manufacturers’ continued investment could be lagging behind other industries; 63 percent of global leaders across all industries reported continued plans to increase investment in 2024. In the U.S., nearly seven in 10 organizations said the same.

Sinoway said that gaining a better understanding of what it really takes to implement and run generative AI systems could be one of the reasons for the fallout.

“I think spending also dropped because people can see that this isn’t a quick fix—there has to be really thoughtful planning to ensure that the benefits, which may not be immediate, are worth the potential risks of accuracy and high costs. We’ve dropped out of the honeymoon phase of ‘anything is possible.’ Businesses are slowing down spending to get it right and not just do it fast,” he told Sourcing Journal.

Part of the reason for the decline in the spending surge stems from the manufacturing industry’s concerns about accuracy and reliability.

Up Close: In Conversation with Rithum's Ellie Williams

Lucidworks’ data found that manufacturers have a high rate of concern around response accuracy of generative AI tools, with 44 percent of industry-specific respondents identifying that concern, as compared with 36 percent of industry-agnostic respondents.

Sinoway said that could be because manufacturing requires a higher degree of precision than many other industries might.

“B2B buyers are looking for very specific parts and materials; something that’s ‘close’ isn’t going to cut it, so responses have to be extremely accurate. There’s also the added complication of dynamic pricing. Different contractors, for example, may have different costs for materials depending on their unique agreement with a manufacturer. Getting those numbers right is critical for maintaining trust and retaining business,” Sinoway said.

Though accuracy was the primary concern for manufacturers, they also face other major question marks; 37 percent of manufacturers worried about cost, and 32 percent said they have concerns around security.

In that sense, manufacturing companies seem to be the outlier. Across all industries, the most major concerns were data security and cost accuracy, nagging 46 percent and 43 percent of leaders, respectively. For cost, that figure is up 14 times from 2023, which Sinoway said was surprising to Lucidworks researchers.

Sinoway said manufacturers’ apparent lesser concern about cost could be due to major differences in customer-facing use cases by industry.

For instance, he said, 63 percent of retailers cited cost as a concern, which he believes might be attributable to “the sheer number of customer queries, [which] drives up the cost exponentially” when compared with other industries, like manufacturing.

Regardless of industry, Sinoway said, cost concerns may soon be quelled. He noted that the costs associated with large language models (LLMs), the technology often used to power generative AI systems, “will continue to drop quickly,” which could cause cost-related concerns to “level out” in the future.

“In just 12 months we have such a different understanding of what it takes to scale generative AI across hundreds of thousands of customers. It’ll be interesting to see how this shifts in 2025 assuming that a lot of commercial models will become cheaper as they get more efficient,” he said.

Tech Tactics: How Carve Designs' AI Quiz Simplifies Swimsuit Sizing