Artificial intelligence is about to transform Valerie Ingold’s industry – debt collection – and she says it is about time.
“When you started a collections agency, you used to need a phone and a filing cabinet. We’ve progressed from there, but we’re slow adopters,” said Ingold, managing director of Commercial Collection Corp. in New York.
ADVERTISEMENT
CONTINUE READING BELOW
“With AI, ChatGPT and the advances, I think it’s really going to level the playing field for us,” she said.
“We’re on the cusp of it. This is the future.”
More than 73 million Americans – more than 20% – have a bill in collections, according to the Consumer Financial Protection Bureau (CFPB).
Almost 60% of those are for medical debt, followed by phone and utility arrears, the agency said last year.
Ingold works with commercial rather than consumer collections, which deals with the particularly sensitive issue of individual debts, but both are quickly looking to tap into the new efficiencies promised by AI tools.
Ingold sees the prospect as both exciting and worrisome.
“It’s a scary prospect, certainly, and there’s a lot to be concerned about. We’re kind of living in the future here.”
Only a little more than 10% of collections companies are currently using AI-driven tools, but 60% are considering or working toward doing so, according to the TransUnion credit reporting agency.
Companies are looking to predict who is willing or able to pay, virtually negotiate payments and “segment and profile customers”, the company found.
AI is “the hot new topic,” said Joann Needleman, a member at law firm Clark Hill who advises the collections industry, noting the sector is seeing an “explosion of technology”.
“Those algorithms are very exciting to financial services providers – they’ll tell you who will pay and who will not,” said her colleague, senior counsel Aryeh D. Derman.
Needleman and Derman said the new tools could prove a major help to companies in complying with consumer protection rules, while some could see AI-driven chatbots as less embarrassing to talk with about one’s debts than a human.
“The beauty of AI is you can throw a bunch of factors into a machine, and it spits out – here’s a list of who you should call,” Derman said of collections companies.
But he warned it would be extremely important for firms to be able to monitor what exactly is going into those algorithms.
“You have sort of entrusted a machine to tell you what (accounts) to work.”
Racial disparities
AI holds promise as “a vehicle for better debt collection,” said Ronald S. Flagg, president of Legal Services Corporation, the country’s largest funder of civil legal aid.
“If (debt collectors) were able to efficiently review the data about the cases they have in greater depth,” thereby reducing the number of cases filed against debtors, “they could use AI to make debt collection smarter and easier,” Flagg said.
Still, he worries about exacerbating longstanding concerns and introducing new ones – bias in the data, possible harassment of debtors and even overwhelming an already heavily burdened court system.
“It could be a vehicle for good, or a vehicle for further abuse and unfairness in the debt collection justice system,” said Flagg.
Researchers have long found racial disparities in who faces debt collection, including who is sued or even who gets contacted in the first place, said April Kuehnhoff, a senior attorney with the National Consumer Law Center.
AI tools could exacerbate those concerns.
ADVERTISEMENT
CONTINUE READING BELOW
“Is the algorithm going to learn that this particular Zip code needs to be sent directly to litigation and be sued?” she said, referring to postal areas.
“That could result in disparate treatment, because of how divided geographically the country is in terms of where populations live.”
She pointed to other potential problems, too: “generative” AI tools making up facts, raising legal liabilities for collections companies; or increased efficiencies allowing collectors to pursue smaller debt amounts or contact people more frequently.
The CFPB finalized a first-ever rule on the collections industry in 2021, making it illegal to use threats or harassment to push for debt repayments, though it did not touch on artificial intelligence or machine learning.
A spokesperson said the agency was monitoring how AI was used in collections, and pointed to a report last year on chatbots in consumer finance that underscored that companies using AI had to comply with consumer finance laws.
Flood of tools
Numerous companies are already offering AI tools to the collections industry, including generative products.
Skit.ai offers AI-driven voice tools for use by collections and other financial services companies that can be used to call debtors or receive calls from them, and which it says automates a million consumer conversations each week.
“Our products include voice, text, and email bots, capable of handling two-way, intelligent conversations with consumers,” CEO and founder Sourabh Gupta said in an email.
“Bots eliminate wait times and instantly respond to inquiries, provide information about due balances and payment options, and assist consumers with dispute handling, settlement negotiations, and more.”
He cautioned that AI had to be handled “ethically and responsibly”, noting Skit.ai tools came with “rigorous filters” to ensure compliance with regulations.
“Our technology doesn’t go off-script and requires prior express consent for its use. We are very aware of the potential for bias in AI, and we have established a system of checks and balances with transparency measures,” he said, including “protected attributes” that can’t be used by the tool to make decisions.
For an industry such as collections, AI tools are here to stay, said Bard Myrstad, co-founder and CEO of Simplifai, an Oslo-based company that seeks to automate a range of basic repetitive office work.
The firm’s tools can typically automate the processing of 70 to 90% of routine tasks, Myrstad said, with average response times going from days to minutes and possible cost savings of more than 50%.
Part of his job is also to caution clients against where to aim their new tool.
“Anything that’s decision-related – that’s something I’d be careful with, knowing in (the collections) industry in particular there could be significant amount of bias in the historic data.”
Still, AI’s eventual effect on office work will be similar to how robots changed factories or machinery changed agriculture, he suggested, with the industry currently being “flooded” with new technical solutions.
“A few years from now, we’ll see if you’re working in a labor-intensive industry like debt collection, you can’t be a significant player unless you’ve embraced automation tools.”
