Businesses across the United States are moving beyond the artificial intelligence hype and starting to use it for practical tasks like inventory tracking and customer service.
Experts say the focus has shifted from broad, generalized tools to more specialized solutions for specific problems. While this marks progress, many companies still find their systems unprepared for AI integration, showing that the gap between AI’s promise and real-world use is challenging.
Deloitte’s third-quarter 2024 report on generative AI showed that companies are moving from testing to scaling their AI efforts. While 67% are boosting investments due to early successes, only 30% of projects have fully launched. Challenges include managing data, addressing risks and navigating unclear regulations. To maximize generative AI’s potential, businesses are focusing on integrating it into everyday processes, aiming for improvements in efficiency, innovation and customer relationships.
“Due to the significant hype around AI, many people have realized that it has specific and practical functions that help them in everyday life,” Ilia Badeev, head of data science at travel services firm Trevolution Group, told PYMNTS.
Businesses’ applications of AI have shifted, with more organizations opting for smaller, more specialized AI solutions over their larger counterparts, Max Vermeir, senior director of AI strategy at AI firm ABBYY, told PYMNTS.
“This reflects a maturation in the AI market that prioritizes value distinctly over the hype of more generalized, generic tools,” he said. “Instead of treating every business challenge like a nail to be struck with the AI hammer, businesses have focused on using it to enable intelligent automation of key workflows like processing high quantities of unstructured data from documents.
Yet observers say implementing AI presents challenges. Badeev said many companies are surprised to learn that their current processes and infrastructure still need to be prepared for AI implementation.
“For example, human business processes could be better described and better divided into specific stages; they rely on the skills of specific individuals rather than on a systematic approach,” he said. “In such conditions, it is difficult to implement AI because it’s unclear exactly where to integrate it.”
Despite AI’s positive aspects, the warning signs of hype are becoming harder to ignore. Business transformation expert Christopher Kaufman told PYMNTS that much of today’s AI implementation in enterprise could be less revolutionary than it appears.
“Most business use of AI is actually machine learning algorithms with sophisticated data structures to add insights to existing data models or slapped-on fourth-generation chatbots,” Kaufman said.
Few businesses are developing their own large language models (LLMs), he said. He identified three categories of organizations truly creating original AI models: “rampant large enterprise company in-house mavericks, tech entrepreneurs or core tech companies.”
One phenomenon shaping market trends is what Kaufman called “dysfunction force.”
“Dysfunction force is the amount of uneducated seeking to profit from that which is dysfunctional at pre-chasm states,” he said. “These curves — whether they are tulips or AI-generated animations — are in and of themselves effective to move markets, shape consensus and channel resources.”
He pointed to disparities in AI capabilities across platforms.
“There is a huge difference between what Google Labs’ NotebookLM is doing versus, say, Claude.ai,” Kaufman said. “There is a huge difference between what Copilot can do versus, say, Canva features with AI.”
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More