When you use ChatGPT or other related technologies, you need to practice “safe AI.”
Is AI going to take over the world? Will AI make us smarter? Will AI make us dumb? Will AI eliminate jobs? Will AI help us grow? These questions and more continue to make AI, ChatGPT and similar technologies the topic of the day.
In the past several months, I’ve been asked to respond to numerous questions related to AI and ChatGPT and how they impact customer service and CX. These have been at conferences, on social media and webinars. Here are some of the most asked questions:
QUESTION: Is it time for more businesses, especially customer service leaders, to embrace generative AI?
ANSWER: The short answer is, “Yes!” AI has been around for many years, but the improvements in the last year, along with lower costs, make it easier than ever to embrace. Let’s go straight to the customer support and contact center. Agents are concerned, if not outright scared, about the possibility of losing their jobs. We must show our agents how generative AI can help them in their jobs, not replace them. Generative AI can help answer basic questions better than anything we’ve had up to this point and allow agents to focus on higher-level issues, as well as building relationships, creating confidence and making suggestions to upsell/cross-sell customers into more products they need.
QUESTION: What are you seeing as the most useful and productive application for AI and ChatGPT technologies?
ANSWER: As mentioned, let AI support lower-level customer concerns, answering basic questions like, “Has my order shipped?” or “What’s my checking account balance?” For the customer support agent, let it support them by quickly providing the agent with answers they need for their customers’ difficult questions. Train agents in two areas. First, they must learn to create prompts to get the best answers. And second, how to communicate that information and simultaneously build stronger and more confident relationships with their customers.
QUESTION: Are there security concerns around using AI and ChatGPT with customers, patients, employees and anyone else using the technology?
ANSWER: Not only are there security concerns, there are also ethical concerns, which are just as important. Let’s move from a company using AI to support customers to the general public taking advantage of ChatGPT-type solutions. Here are the areas to pay attention to:
· Confidentiality: Anything a customer asks in an OpenAI application can potentially become public information. Imagine a patient asking about their health. How can that be kept confidential? In an open forum (versus a company’s private website with its own AI solution), confidentiality doesn’t seem to be maintained. People must be informed how to protect their personal information, and it can’t be hidden in a website’s fine print.
· No Parental Controls: This has been a chronic problem of any search engine and website. We must keep this in mind as the future of ChatGPT and OpenAI continue to improve.
· Inaccurate Information: Sometimes, the answers provided by ChatGPT are wrong. For example, I asked, “How many customer service books has Shep Hyken written?” The answer ChatGPT returned was four, and it included titles. Yet, the answer was wrong. The correct answer is eight (but who’s counting?). So, I followed up by asking, “Are there other books Shep Hyken has written, and what are the titles?” The answer came back, and this time it was 12. Again, the answer was wrong; however, the extra titles included books in which I had written a foreword or was quoted. Imagine if you had a serious health issue, asked ChatGPT a question, and the answer was wrong. Unfortunately, ChatGPT can’t be sued for malpractice—at least not yet!
QUESTION: ChatGPT was introduced to the world in November 2022, barely eight months ago. The adoption and use of this technology is like nothing we’ve seen before. What safeguards should be put in place? Can it—or should it— be regulated?
ANSWER: Yes, the adoption of this new technology is overwhelming. It took just two months for ChatGPT to reach 100 million users! Compare that to Facebook, which took four and a half years. If the government wants to regulate ChatGPT, it will find a way. More than anything, it’s about making the technology safe and reliable. We are still in the infancy of this new technology. ChatGPT will dwarf Moore’s Law, which is that the number of transistors in a microchip doubles every two years. In other words, ChatGPT’s capabilities may double every two months, not every two years, and maybe even faster! So, use caution. When you get in a car, you want to practice safe driving. When you are out for an evening with friends, you want to practice safe drinking. When you use ChatGPT or other related technologies, you need to practice “safe AI.”
New tools and technologies are often met with concern. Short-term AI, ChatGPT and related technologies will be surrounded by controversy. There will be promoters, skeptics and those who may want to eliminate it altogether. For now, accept it as is, be cautious when using it and take advantage of the good it offers.
