What Earnings Calls Reveal About AI And The Future Of Collections

Fed researchers and ARM industry executives agree: AI investment is accelerating but the productivity gains everyone is expecting haven’t fully arrived yet.

A new analysis from the Federal Reserve Bank of St. Louis confirms what accounts receivable management (ARM) industry executives have been signaling for months: artificial intelligence is reshaping the language of business productivity, but the measurable payoff is still largely ahead.

The research, published July 31, analyzed roughly 490,000 earnings call transcripts from 5,198 publicly traded U.S. firms spanning 2000 through 2025. Its findings offer a valuable macro lens for ARM professionals that lines up closely with what the industry’s own publicly traded companies have been saying.

Productivity Talk Is Up and AI Is Driving It

The Fed study found that overall discussion of productivity on earnings calls has been rising in recent years, recovering sharply after a COVID-era dip and climbing to its highest levels of the study period by 2025. But the more striking finding is how that productivity conversation has changed.

Before ChatGPT launched in late 2022, virtually no earnings call productivity discussions referenced AI. That changed fast. By the end of 2025, roughly 15% of all productivity-related sentences on earnings calls mentioned AI.

The language itself has also evolved. In 2023, executives leaned heavily on the often-vague phrase “generative AI.” By 2025, more concrete phrasing had taken over: “using AI” and “AI tools” became more common, suggesting that the conversation is maturing from concept to application.

Almost All of It Is Forward-Looking

Here is where the data gets instructive for business leaders: approximately 95% of earnings call sentences that discuss both AI and productivity refer to future gains — not productivity improvements already achieved. That compares with roughly 75% of non-AI productivity discussions, which also skew forward-looking but notably less so.

This pattern held consistent across both 2023 and 2025, meaning firms haven’t shifted to reporting realized AI gains even as the technology has become more deeply embedded in business operations. As the researchers put it, firms appear to be “actively investing in, experimenting with, and expecting future gains from AI” — while the measurable productivity payoff remains mostly ahead.

That assessment aligns with aggregate economic data. Utilization-adjusted total factor productivity grew just 0.07% over the four quarters ending in Q1 2026, according to San Francisco Fed data — a muted signal that broad AI-driven gains have yet to materialize in the macroeconomic numbers.

What ARM Industry Leaders Are Saying

That same tension between investment and realized results shows  clearly in recent earnings calls from publicly traded debt collection companies. In Q4 2025 and Q1 2026 calls, executives at Encore Capital Group and PRA Group described a sector that is actively deploying AI, while also managing expectations about where the technology is and isn’t ready.

PRA Group CEO Martin Sjolund offered some of the most detailed public commentary on AI adoption in the ARM space. In the Q4 2025 call, he described a broad-based effort spanning multiple use cases: “We have already started testing a range of AI initiatives, from processing to interactive chatbots to using large language models to process massive unstructured datasets, to help us inform our collection strategies.”

He noted that PRA’s global footprint — including 70 million accounts, hundreds of millions of documents, and billions of call recordings — creates a data advantage.

“There is a significant opportunity to digitize workflows, serve customers digitally, and use virtual agents to transform customer service,” Sjolund said.

And on the Q1 2026 call, that commitment had deepened: “We will continue to leverage our massive amounts of data, customer insights and AI to drive improved processes, cost savings and enhanced customer service.”

Encore Capital Group CEO Ashish Masih offered a notably candid counterpoint in the Q1 2026 call. While expressing confidence in AI’s long-term potential, he acknowledged meaningful limits on the technology in collections specifically: “Our calls are very complex. It requires empathy and dealing with consumers. So, tools there, while there are a lot of voice-oriented tools, they are not quite ready to deal with that as well as we can with our account managers. There are some regulatory nuances … one has to be mindful of.”

What It Means for the ARM Industry

For debt collection and ARM professionals, the Fed’s findings carry a practical message. AI adoption in financial services, customer communications, compliance workflows, and operational efficiency is actively underway at publicly traded firms across sectors. The productivity gains executives are projecting will, when realized, reshape staffing models, process design, and competitive dynamics.

At the same time, the data is a useful check on hype. The fact that 95% of AI productivity talk is still future-tense — even three years after generative AI became mainstream — is a reminder that implementation takes time, and that the gap between investment and return can be significant.

The St. Louis Fed researchers noted they will continue monitoring whether earnings call language shifts from expectation to realized results in coming quarters.