As AI gains enterprise 'citizenship,' it needs a foundation in trust, Accenture exec says
As AI systems gain increasing "citizenship" in the enterprise, organizations have to ensure the "brains in a box" are adequately trained with enough data sets and content to establish a "bedrock in trust," said Mike Redding, managing director of Strategic Technology Innovation for Accenture, speaking Friday at an Accenture Technology Vision event in Washington D.C.
Deployment of AI is complicated, particularly in legacy environments. This challenge has helped give rise to the popularity of robotic process automation (RPA), which is considered "entry level AI," Redding said. Companies can deploy a digital "agent" to the desktop of an employee to augment and streamline tasks, such as those requiring the processing of nuanced regulations.
As AI is rolled out, however, companies will need to understand what goes into systems before it is deployed as a way to ensure trust. With the pending GDPR deadline approaching, companies will have to soon explain what goes into AI, Redding said. While this does not mean unearthing underlying proprietary algorithms, companies will have to understand how an AI system arrived at its decision.
Experts are quick to extol the benefits of artificial intelligence, but companies are still working to apply it in combination with the existing workforce. Rather than replacing employees with machines, organizations want to enhance the workforce and make laborious tasks easier.
A baby step in what will eventually be advanced machine learning or even neural nets, RPA allows for organizations to have attainable AI that doesn't require large amount of computing capacity.
Trust remains an underlying imperative, especially as new regulations mandate that companies understand how deployed AI makes its decisions. That's one of the reasons deep learning has not taken off yet in organizations.
Serving as more of a "black box" for decision making, researchers often struggle to understand how the technology thinks, which poses a challenge in regulated markets, particularly in light of compliance concerns.
AI is one of the most popular buzzwords of the last year, and many companies will look to roll it out without understanding its full benefits or how it could best fit into a complex environment. It is up to technology leaders to thoughtfully deploy the technology without undermining the capabilities of an existing workforce.
Follow Naomi Eide on Twitter