Over 1,500 attendees were in San Francisco at A.I. World recently to hear experts talk about the future of the artificial intelligence industry. One of the highlights was a panel featuring William Mark, PhD, President of SRI's Information and Computing Sciences Division and one of the brains behind Apple's Siri.

"We need to make artificial intelligence less artificial," said Dr. Mark, who stressed that A.I. is a lot more than "deep learning" and machine intelligence.

This was an important distinction as most attendees were from the enterprise sector and almost entirely focused on using A.I. to do what humans cannot—massively parallel computing. Many speakers pulled out the oft-quoted data point that IBM's Watson supercomputer can read 40 million documents in 15 seconds as an indicator of our hurtling trajectory towards the Singularity.

But Dr. Mark's vision for the future of A.I. is altogether, well, more visionary, and complex in terms of system design.

"We do a lot of work in conversational A.I.," said Mark. "And we do use deep learning for parts of what those systems do, but there's also intent recognition, natural language processing, as well as emotion detection.

"In managing real-time conversation, [so the A.I. knows what to say next] when someone says something, we have to teach it to know that they've introduced a new intent, but haven't finished the previous one yet. For that we don't use machine learning, we use explicit rules and all kinds of inference mechanisms."

After getting his PhD in Computer Science at MIT, Dr. Mark held executive and research roles at National Semiconductor and Lockheed Martin Palo Alto Research Laboratories. Joining SRI in 1998, he focused on creating new technology in speech and natural language, vision and perception, planning and reasoning.

After Siri was spun out and bought by Apple in 2011, his group refined another conversational A.I. platform called KAI, which is used by the banking industry. Although it's now a separate company, Mark remains on its board of directors.

KAI's financial system-based A.I. meant Mark's team had to incorporate trust mechanisms into its design.

"Being able to trust these A.I. systems is going to be a major factor, which is what I meant by making A.I. less artificial," Mark said. "Because people think differently. For example, people like to think of cause and effect but most of the systems are [designed to be] about correlation. Making that bridge [between the two modes of thinking] is really important, and one of the leading edge problems we're working on—to move forward with trustable systems."

Finally, he told the attendees at A.I. World what's next from his team at SRI: something he dubbed "Super Siri."

"One of the things we're working on is the problem of the aging population," explained Mark. "It's a huge cultural and economic issue [and there are] many aspects to that problem. People's minds immediately go to 'robotic help in the home?' And we are working on that. But we're also looking at companionship. Many elderly people are lonely and that leads to cognitive decline. We're working on systems—think 'Super Siri'—which will bring in family, help bring back memories people had in the past, and it is all tied into sensors in the home" to monitor mobility and other health issues, and into IoT.

So there you have it: the next wave of A.I. are systems that engender trust, use inference mechanisms to infer intent, and predict what you might say, do or need next. Perhaps something like this scenario:

"How about a nap?"

"Excellent idea, Siri, I think I will. Wake me up after 20 minutes of REM when I reach steady delta wave formation."

"What's the magic word?"

"Please."

"Consider it done."