Technological singularity is the theory that artificial intelligence (AI) will overcome its human masters, i.e. its trainers. The idea was proposed in 1993 by researcher Verner Vinn. In a follow-up paper, Vinn estimated that technological singularity will happen "... not before 2005, and not after 2030."
"So, we have a few years left," said Diego Gosmar, CEO at AI development company XCALLY. Gosmar addressed a crowd gathered in the Solutions Theater at February's ITEXPO 2023 in Fort Lauderdale, FL.
Gosmar spoke about his belief that humans can successfully manage AI beyond Vinn's expected expiration date, as well. "We don't know when it will happen," he said. "It's a singularity, which means nobody's seen it before."
While technological singularity may be inevitable, Gosmar said that he believes the events can be appropriately guided, if not prevented. His company XCALLY is geared toward developing and implementing the human management controls that proper AI implementations need to succeed. (And he wrote a book on machine learning.)
Today, AI-influenced operations appear mostly in customer service applications. Gosmar said the most popular applications of AI for customer service include:
- Conversational AI - Using AI to either converse with customers or to provide support for team members.
- Conversational analytics - Gaining useful information by analyzing customers' conversations.
- Agent recommendations - Leveraging AI to provide better information and guidance, and far more quickly.
- Workforce management - Scheduling the proper services to proper;y scale workforces.
Next, Gosmar showed a three-minute video describing how AI operates in a theoretical business; i.e. an autonomous, self-driving water taxi. AI was used in the demo project to pilot the boat, manage boat operations, and process online reviews. AI also was employed in the registration process, where technology provided the answers to customers' queries. The idea was to show the audience where the future is headed.
Gosmar then described a non-theoretical project. By adding AI-influenced customer service processes to a public bus line in France, XCALLY was able to push half of the queries to AI within two years. In three years, the AI had deployed machine learning tactics to increase its capability to nearly 70% of customer questions. "That was really a success," Gosmar assured.
He then moved on to today's challenges, as he sees them. Firstly, data aggregation and management. Due to privacy concerns, "It's not easy to get a lot of customer data today," he said. Secondly, AI finds it difficult to incorporate nonverbal communication, such as voice tone, gestures and facial expressions. Thirdly and finally, there's a definite lack of skills among workers when it comes to using and benefitting from AI technology. "People need training to deal with AI," he asserted.
"What we really need is an artificial intelligence we can trust." To that end, tools are being made available to check for bias in AI scripts, solve the "black-box" dilemma of explainability, and format governance rules. Gosmar said that being able to explain precisely how AI makes its decisions to all stakeholders - c-suite, staff and customers - is important because each segment requires their own explanations.
So yes, technological singularity can be managed, Gosmar believes. "Our goal is to combine human governance with the boldness of the top-edge technology," he concluded.
Edited by
Alex Passett
