Robotic process automation (RPA) vendor Nice has developed an ethical framework for building bots that it plans to share with its customers, the provider revealed during a webinar last week.

The framework will be distributed along with training and explanations, said Orr Rachlevsky, director of the Nice Advanced Process Automation Line. During the webinar, “The Path Towards Unbiased Robots: Guiding Principles of Ethics in RPA That Every Enterprise Should Think About,” Rachlevsky said the framework and its underlying logic were inspired by science fiction writer Isaac Asimov’s three rules of robotics, and include an ethics assessment created by Emergence Partners, a U.K.-based IT research firm.

“We’ve developed an ethical framework we now distribute to our customers. Every license that they receive comes with the ethical framework,” Rachlevsky said. “Customers can take these ethical frameworks as a guidance to how to develop an automation.”

Emergence Partner’s ethics in technology assessment framework looks at several dimensions, including trust in technology, data privacy, managing data and ethical AI, all of which are top concerns as automation takes on a greater role at financial institutions.

Asimov’s rules of robotics prohibit robots from harming humans directly or indirectly. “Robots must be designed for positive impact,” Rachlevsky said, adding that Nice requests partners and developers be able to articulate a positive impact for the bot, in terms of the customer or employee experience.

The Nice framework outlines the vendor’s five rules for bots:

1. Robots must be designed for positive impact.
2. Robots must be designed to disregard group identities, including race and gender.
3. Robots must be designed to minimize the risk of individual harm.
4. Robots must be trained and function on verified data sources.
5. Robots must be designed with governance and control.

But is bot automation still too nascent for such guidelines? Webinar attendees asked how to address bias in algorithms, and whether organizations can assess the potential harm of a bot, particularly if automation leads to layoffs.

However, Sarah Burnett, founding partner of Emergence and an RPA expert, advised that now is the time to act. “It’s really important that we start tackling this today rather than leave it for effectively the genie to be out of the bottle and causing loads of trouble in 10, 15 years’ time,” Burnett said.

Eliminating bias in algorithms begins with a diverse team and ethical use of data, Rachlevsky said. Training data must be a large enough sample size and comprehensive enough to build trust in the AI, and there are also tools and techniques for checking algorithms for bias, he added.

As for the potential for automation to lead to job loss, Rachlevsky said it’s still not clear whether that’s the case. He pointed out that shifting from horses to cars led to short-term job loss, but workers later found new jobs in new industries.

Ultimately, ensuring bots and AI are ethical is up to the humans involved, Rachlevsky said.

“If people don’t care, all these documents are meaningless,” Rachlevsky said. “But if they do care, I believe that this can help in shaping the thinking and implementation of the RPA technology.”

Bank Automation News would like to know what our readers think about bots. We invite readers to take part in a short BAN reader survey.