Like a teenager who sometimes makes poor choices and can’t always explain the emotions that led them astray, artificial intelligence is experiencing growing pains.

Although the term was first coined by computer scientist John McCarthy in 1955, AI didn’t really get rolling until eight years ago when a confluence of faster processors and an explosion of data-gathering devices advanced the cause significantly. In 2011, Apple’s voice assistant Siri was introduced and IBM Corp.’s Watson out-answered contestants on a quiz show, which helped spark greater interest in the potential that AI offered.

The problem is that much of what is being created using AI, especially the kind known as deep learning that attempts to help computers learn a bit like humans do, still remains a mystery. As information gathered by sensors flows into a massive network of artificial neurons and algorithms and an outcome is spit out, no one can be quite sure how the machine reached its decision or took a particular action.

In current tech terminology, AI is encased in the “black box,” and that’s not necessarily a good thing.

One company — Kyndi Inc. — has set out to open the black box and explain what’s inside. The firm is building what it calls the first “Explainable AI” platform, a way to provide auditable reasoning behind the AI mystery machine.

“You can’t understand what the outcome is,” said Amy Guarino (pictured), chief operating officer at Kyndi. “We’re doing it in a way where you’re able to trace back to the actual raw data source to make sure that it really is correct.”

Guarino spoke with Peter Burris, host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, in Palo Alto, California, in two separate interviews. They discussed the technology behind’s Kyndi’s explainable AI, adoption of its platform beyond regulated industries, growing concerns that deep learning may be not be advancing rapidly enough, and what Guarino has learned from years of experience in the male-dominated corridors of Silicon Valley (see the full interviews with transcripts here and here).

This week, theCUBE spotlights Amy Guarino in its Women in Tech feature.

Parsing context from language

Kyndi is using language to unlock the black box. Its technology ingests documents in any kind of text and then applies natural language processing to parse out terms and context.

It can then extract a structure from the documents and provide domain flexibility, something that black box AI does not do especially well. Kyndi’s platform doesn’t need mountains of data to train its models either. The system can train on only 10 to 30 scientific documents with up to 50 pages in each.

“You don’t have to do a lot of work up front building out a taxonomy,” Guarino said. “The approach that we take really focuses on language.”

The company originally believed that the most widely used application of its technology would be in regulated industries where explanations were required. But Kyndi has found that its process of building a giant knowledge graph from unstructured text data has appeal across other organizations, such as human resources and marketing.

“It is also very relevant for any kind of decisions where humans are allocating resources and they have to explain why,” Guarino said.

Here’s the first video interview with Guarino:

Stalled progress for AI

Kyndi’s approach comes at a time when some tech experts are warning that AI may be about to hit a “wall” in its progress. Last year, Gary Marcus, a professor at New York University and longtime critic of current AI techniques, published a report that openly questioned whether the patterns extracted from deep learning were really all that deep after all.

That could pose problems for many enterprises down the line as AI and machine-learning adoption continue to grow. Deloitte Global recently reported that penetration rates of enterprise software with integrated AI will reach 87 percent by next year.

Kyndi’s own discussions with enterprise firms echo some of the uncertainly surrounding AI. Guarino recently spoke with one company that was using deep learning for water treatment plants, a process being questioned by water quality regulators.

“Their client said, ‘You need to explain to us what you’re doing to it and why,’” Guarino recalled. “They said, ‘We can’t.’ That’s a problem.”

Confronting the gender barrier

Guarino has spent more than 25 years guiding customer sales organizations. Prior to joining Kyndi, she ran business development for Marketo Inc., a marketing automation software company that was recently acquired by Adobe Inc. for $4.75 billion.

With her Silicon Valley experience, Guarino is well aware of the challenges confronting women in an industry that tends to be male-dominated.

“Especially in tech, there’s so many situations where you walk into a sales organization and it’s 80 to 90 percent fellows,” Guarino said. “It makes it tough for women to want to join that kind of an organization.”

Kyndi’s COO is no stranger to environments where women are in the minority. Guarino graduated from the University of Notre Dame in the 10th year after the school began admitting females. Her class ratio at the time was 80 percent men and 20 percent women.

As she pursued her career, Guarino developed a better understanding of the strengths that women can bring to a sales organization. “How do women look at sales differently?” Guarino asked. “A lot of it is focused on listening skills. Women do have a natural affinity to be able to listen and to really pull out, whether it’s a prospect or customer, what’s really important to them.”

The challenge confronting many tech companies is to diversify the workforce and take full advantage of the skills and attributes that different points of view can bring. This includes making sure that women are well-represented in a company’s leadership, according to Guarino.

“That can change the dynamic quite a bit in terms of the company culture,” Guarino said. “The expectation is that you’ve got different types of perspectives and different ways to look at customer acquisition, engagement and support. We can all help each other when you have different opinions and different ways of looking at things.”

Here’s the second video interview with Guarino, one of many CUBE Conversations from SiliconANGLE and theCUBE:

Photo: SiliconANGLE