Independent
Software Developer
HR Manager
Team Leader
Accountant
Copy Writer
Customer Service
Virtual Assistance
Evan Ratliff’s Shell Game
Every month, the market welcomes new artificial intelligence (AI) platforms that boast easier prompt customization, improved latency, or other new features. But once in action, these platforms still generate cold, robotic, and unconvincing responses.
The call center industry is collectively petrified of the arrival of the AI voice-bot technology that wipes away the need for their call centers. Despite optimistic pronouncements from AI enthusiasts, the current products are still falling short.
This was one of the takeaways from journalist Evan Ratliff’s podcast series, Shell Game, which explores the emergence of AI voice bots. As part of an “immersive journalism” project, Ratliff cloned his voice and tested it on friends, colleagues, family, call center agents, fellow AI agents, and even a psychologist. Most of them sensed something was wrong in seconds.
>> Podcast: Listen to my interview with Evan Ratliff, discussing his AI investigation.
We can instinctively decipher a human voice from an AI voice simply because we talk more to people. Varying inflections, abrupt pauses, and mispronunciations are all part of the human communication process.
When we hear someone talk too perfectly, we deem they’re either reading from a script, it’s some type of a pre-recorded voice, or it’s a programmed robot. Form follows content. AI voice bots also struggle with emotional types of conversations. As Ratliff told me in Episode 523 of the Outsource Accelerator Podcast, AI voice bots don’t laugh when they’re supposed to laugh.
It seems that despite rapid, incredulous advances in AI, AI voice bots still can’t convincingly capture the nuances of human language – yet…
The waiting game
These shortcomings in the technology have important implications for its business applications, especially in the call center industry.
A handful of business process outsourcing firms have deployed voice bots in their operations. These bots can handle very simple interactions, provided they are given a very tight prompt. But there’s still a risk involved when faced with urgent and unique inquiries that are not covered by the prompt.
Teaching bots how to feel and render emotion has become a great challenge. Analysts believe it’s the last step before we can fully have one computer handling thousands of queries at once, 24/7. Experts believe we’re at 80% to 95% perfect communication. But filling in the last 5% to 20% is an uphill climb.
How long will it take to fill that gap? Will a landmark product be launched in five years, or will this process take another decade or two? Will this ultimately decimate the call center and other industries?
The human contribution
Ratliff, a journalist for over 25 years, has similar fears. “I’m not looking to be replaced,” he told me. “I’m looking to find the meaningful part of my work that it can’t take and maybe to do more of that. And that makes me more foolproof against it.”
This is the mantra call centers and other industries should embrace. While we wait for AI to improve, we must find that trait that truly separates us from bots. Whether it’s critical thinking skills, soft skills, or the ability to empathize, we must look to develop and find the value of human contribution in everyday life.
Today, AI still needs us to learn and guide it through the ropes. Even as it reaches the highest level of sophistication in the next 20, 30, or 40 years, it will still need us to clean up the mess it will inevitably make.
The question for your business
What part of your operations that can only be done well by humans?
