As generative and conversational AI continues to take the world by storm, questions abound about how to use these new tools in a meaningful, impactful, and – most importantly – ethical way.
One approach? Keep human beings at the center of the conversation. Speaking at the most recent League of Leaders gathering at the SAS campus in Cary, North Carolina, SAS’ VP of data ethics, Reggie Townsend discussed how businesses today can harness the sociotechnical space – generative AI, automation, machine learning, and more – while making sure there's ample space for the human to show up too.
Meeting up four times a year, the League of Leaders draws together a cross-functional suite of business and IT leaders in retail and consumer brands. Conversations between attendees are candid, collaborative, and focused on the hottest topics of the moment.
That topic today? ChatGPT.
"You don't need access to everything."
Kicking off proceedings, Townsend’s demo with the chatbot resulted in informative, extensive, and specific answers to a query about where to purchase a gray suit. For retail and consumer goods companies, Townsend said, ChatGPT offers immense possibilities for personalization, as well as looking for the optimal route to get a manufactured product from point A to point B.
“What I predict is these generative APIs will be most useful when they are domain-specific. Right now, a tool like ChatGPT goes across the entirety of the web, and it crawls for books and publications and social media sites, and popular datasets. But sometimes you don't need access to everything.”
In addition, Townsend cautioned that these answers are only based on the greatest probability response and may not be true and accurate.
To illustrate this point further, Townsend demonstrated how he could provoke ChatGPT to give wildly inaccurate information – in this case, on the population of Myanmar – simply by disagreeing and directing the conversation in a different way. The human factor becomes, then, a moral imperative. We need to step in to make sure the information generated is not only accurate but used for good.
An Ethical AI Blueprint
In terms of SAS’ approach, Townsend offered a look into the company’s research-backed ethical and responsible AI blueprint:
“At the very top, before you start writing a line of code, [you have] to activate a trustworthy AI environment,” Townsend explained. “You've got to start with this some measure of oversight. You’ve got to think about what your operation is going to look like. You've got to make sure you've got adequate performance and risk mitigation in place and you've got to work on building a culture that is ethical by design.”
This story was originally published on CGT, a RIS News sister site