GARBAGE IN, GARBAGE OUT: When algorithms for artificial intelligence use biased historical data, profiling will reflect that bias.
A fictional mortgage company adopts artificial intelligence software that can review loan applications without knowing the race, nationality or religion of the applicant. The goal: eliminate discrimination in mortgage lending. A new employee, however, discovers the AI software is rejecting all qualified applicants of a specific race or heritage. He knows he should alert his bosses, but he’s unsure of his standing.
Chrissann Ruehle, Florida Gulf Coast University management instructor, said her business students must learn how to ethically manage AI in the workplace. “I wanted them to develop an awareness of the ethical issues of AI, to begin thinking how this technology can negatively harm customers, how it could potentially harm the workforce,” she says. She recently partnered with Harrison Ambs, chief strategy officer at Vectra Digital in Fort Myers, to learn how workers confront AI gone awry.
Not to be confused with robotics or machines that perform repetitive work without human intervention, AI is software, and is most often used in automating budgeting, financial forecasting and other business processes.
Ambs and his digital marketing team built the AI to assist in writing ads and serve those ads on Facebook and Google advertising platforms. The platforms then target prospective buyers based on a customer’s purchase history and other unique data. Vectra’s AI also analyzes customer sentiment based on publicly available feedback (online reviews, surveys) so businesses know what their customers are thinking about them.
At Vectra, humans make the ethical choices, not the software. “The AI never has access to the data,” Ambs says. “It can’t do anything unethical with it.”
While researching her doctoral dissertation, “Understanding the Complex Ethical Landscape of Artificial Intelligence Adoptions,” Ruehle found that Vectra staff must exercise ethical—and legal—oversight as they collect, store and protect customer data attached to the AI system’s mission.
“I don’t like the creepiness of tracking social media posts,” Ambs says. “We built our tool so it can only use opinions that people give freely away on our clients’ websites. We didn’t want to build a system that would take your name, for example, in a Google review, then dive into your social media to find out everything else you talked about.
“Another ethical question we run into is ensuring there is no sharing of customer data between companies. If we build an email list that we’re marketing to for Client A, and Client B would very much benefit from having that data set, we don’t cross-pollinate; it’s ethically unfair to any of our clients.”
An ethical landscape is developing around the treatment of customer data, Ruehle said.
“Companies are going to be much more transparent about how data is used and how ethical they are,” she says. “It’s now part of our consumer consciousness.”
“Honesty” and “Communication” top a list of Vectra’s core values etched in the frosted glass of a conference room door. “During staff meetings, they have discussions around this issue to ensure they are appropriately and carefully managing the data,” Ruehle says. “They have a moral compass they can turn to.”
“If you are in a culture where you feel that you can’t bring something to your supervisor,” Ambs says, “that is a bad environment to be in.”
As for the mortgage company, its AI software learned the previous addresses of loan applicants. According to the Machine Intelligence Learning Institute, the software rejected anyone who was born in, or lived in, a poor neighborhood at one point.
In other words, when AI algorithms use biased historical data, profiling will reflect that bias. “It is possible, when you’re developing AI, to unknowingly put your own biases into it, either negative or positive,” Ambs says.
WILL ARTIFICIAL INTELLIGENCE COST JOBS IN SOUTHWEST FLORIDA?
Employees who worry that artificial intelligence will replace them should probably relax—for now. First, adoption is not yet widespread in Naples, Fort Myers or elsewhere in Southwest Florida. Second, AI
so far primarily helps with, rather than takes over, work done by humans.
In Southwest Florida and elsewhere, large high-tech enterprises, telecom companies and financial services houses are leading adopters of AI, and have the “most aggressive investment intentions,” says McKinsey Global Institute’s 2017 study.
Gartner Inc. found that 70% of organizations will integrate AI into the workplace, not to cut jobs to save money on salaries and benefits, but to “assist their employees’ productivity.” The company also says one in five workers engaged in nonroutine tasks will rely on AI to do a job by next year. Tasks include communicating with customers online and other virtual assistance functions, such as that “Ready to chat?” dialog box that pops up on a company’s website.
National corporations— such as airlines and health care providers—with field offices in Naples and Fort Myers rely on AI housed on servers elsewhere.
“I have seen it to be quite pervasive in health care, financial service, airlines, the manufacturing industry, especially in supply chain management, distribution and logistics,” says Florida Gulf Coast University Professor Chrissann Ruehle, who studies AI and its potential impact on workers.
ARTIFICIAL INTELLIGENCE: Humans make the ethical choic- es, not AI software, said Harrison Ambs, chief strategy officer at Vectra Digital.
“Companies are aware of it, and the benefits it can have,” says Harrison Ambs, chief strategy officer at Vectra Digital in Fort Myers. “Some are looking for AI vendors, maybe larger AI developers, outside of the state to help them. But for the most part, there’s nothing proprietary; no one is developing their own AI.”
If they have adopted AI, Southwest Florida firms have purchased smaller, off-the-shelf systems,
such as AI that automatically transcribes a voicemail into text and emails it to the recipient. “That frees employees to perform tasks that excite their creativity,” Ambs says. “Jobs that make use of their insight and intelligence. Things humans do better than AI.”
