Legal risks and reputational setbacks fuelled by AI-based technologies could be devastating for businesses
“And when these (AI) models are exhibiting bias, chances are, their organizations have been exhibiting bias that they either didn’t realize or didn’t want to recognize,” she said.
Parsons said business executives who are used to computers providing the same answers when variables are kept the same are now “trying to wrap their heads around” generative AI because of its continually changing output.
“There are cautionary tales that don’t necessarily have to do with responsible tech, but more to do with whether companies understand this technology well enough, particularly in customer-facing situations,” she said.
No time to wait and see
But the time pressures around AI implementation mean corporations can’t wait for the technology’s kinks to be ironed out as they might have in the past.
“Even a year ago, what we thought would take three years is now here in terms of gen AI capabilities,” John Stackhouse, a senior vice-president at Royal Bank of Canada, said.
The technology is still evolving, but he said companies “need to get ready to get on the boat” if they haven’t done so already because it may be too late when generative AI reaches a larger scale.
While the ethical risks around generative AI are certainly substantial, RBC points to the need for businesses to start investing in the technology now, piloting their offerings early and starting small if necessary to create an environment that’s safe for exploration and experimentation by employees.
By delaying, the report’s authors warn that businesses will miss improvements that their competitors that started using the technology months or years earlier are taking advantage of.
Even a year ago, what we thought would take three years is now here in terms of gen AI capabilities
John Stackhouse
Stackhouse said having a data strategy that aligns with evolving technology and supports the overall business strategy is key.
“This is not something to be left to the techies in isolation … as part of that strategy; it’s also ensuring you are developing the right governance models and regulations,” he said.
As for curbing consumer apprehension around AI ethics, Stackhouse said companies need to ensure their customers understand what the potential risks are in sharing their data, as well as what protections are in place to minimize those risks.
“The general public has concerns, but is highly open to advanced technologies that make their lives easier and cheaper,” he said. “We have to communicate in a simple, non-technical way about what we’re doing with their data … and what the benefits are to them as users for opting in.”
Businesses looking to develop an AI strategy are at one of three maturity levels, said New York-based Lambert Hogenhout, chief of data, analytics and innovation at the Office for Information and Communications Technology in the United Nations Secretariat and a frequent advisor on responsible technology.
At the lowest level, companies are looking at responsible AI practices as a compliance issue best left with their legal departments to deal with. At the next level, organizations are approaching it as a risk-management issue, so they are being cautious to not use AI in any way that will alienate customers or partners.
The real leaders, Honeghout said, are the ones who look at responsible AI as an opportunity and a strategic differentiator.
“They will become trusted companies and it will help them in their branding and their marketing strategy,” he said. “That’s where companies need to be moving.”
In finding opportunities to put AI to its best use, Parsons reiterates the need for organizations to experiment with these technologies to determine where the benefits and risks are.
In addition to having skilled employees run controlled generative AI experiments and analyze their results, there is an opportunity to get valuable feedback throughout the organization.