A gathering of (AI) knuckleheads

Yes, once upon a time, it was a Steven Spielberg movie (and an underrated masterpiece on top of that). But today “AI” is a buzz phrase that pops up with more and more frequency in all kinds of social, political and especially technological contexts.

AI stands, of course, for “artificial intelligence,” also often referred to as “machine learning.” Ask most folks to define AI, however, and it’s much like asking people how a toilet works: You think you know, until you have to explain it to someone else. Then you’re not so certain.

Artificial intelligence is a moving target, which makes it a challenge for even professionals in tech to fully understand it. To help the rest of us, Santa Cruz Works is convening a special event on Wednesday, May 4 (which, coincidentally or not, is Star Wars Day). The event is called “A.I. for Knuckleheads,” a sure-to-be lively panel conversation with four people on the frontiers of the technology, moderated by author, entrepreneur and Silicon Valley gadfly G. Craig Vachon.

Vachon is particularly suited to lead an event called “A.I. for Knuckleheads.” The onetime Santa Cruzan who now lives in Carmel is the CEO of a startup called AI Redefined, or AIR, plus he’s the author of a comic novel titled “The Knucklehead of Silicon Valley.”

Vachon will be discussing AI technology, and how it pervades, or will pervade, our lives with a group of professionals in the field who cover a wide range of how the technology is being used, mostly in the realm of spoken language. Drew Meyer works for Amazon in the field of “voice tech,” most notably in the familiar consumer product we know as Alexa. Jim Whitehead is an engineer and professor in UC Santa Cruz’s gaming department. Beth Ann Hockey is a data scientist with a background in linguistics and spoken-dialogue systems. And David Colleen is a software designer and entrepreneur whose business, SapientX, develops platforms for conversational AI in consumer products.

We discussed AI, its potential, and what the audience might expect from the May 4 event at the Kuumbwa.

The following conversation has been edited for clarity.

Lookout: What do you expect general audiences know about AI?

Craig Vachon: Artificial intelligence implies the ability of a machine to self-learn. The audience we’re hoping to attract to this is people who are curious about what AI is and what we hope to achieve with it. In my realm, AI is just another kind of tool in the toolbox, much like, 20 years ago, someone might mention a database. And I think AI is a kind of tool set, because it is so broad. It’s easy to have it fit many different scenarios. And even the panel, which is an excellent panel that we’ll be presenting next Wednesday, is a broad swath of the AI spectrum. So you have people who are in natural language processing, like the Alexa team. That’s just one component of artificial intelligence. But AI is much broader than that.

Lookout: It is very broad. Maybe like 150 years ago, electricity was broad, or the broadcast spectrum is broad, or the internet 30 years ago. It seems that there are two questions on most people’s minds: One is, what is AI doing now that I’m either dimly aware of, or not aware of at all, in terms of influencing and shaping my life? And the second is, where is it taking us? Is that kind of the orientation of the evening?

Vachon: I think that’s exactly right. But maybe even a little bit more of, where should it take us? The greatest challenge is that it’s very limited in its narrowness. So we can teach an AI to find a cancer cell in the liver, but point it 3 centimeters away to the pancreas and the AI needs to be completely retrained. And that narrowness is kind of the bane of our existence today, where we have things like Facebook’s AI algorithm for engagement. We’re now, as social scientists, hugely concerned because that AI has become successful because it has exploited people’s desires, and has created, in some instances, polarization, hate, and all the rest. AI isn’t smart enough to realize that, while it’s highly successful in creating engagement, it’s wildly inappropriate for humankind as a species, because of the base elements that it’s exploiting.

Lookout: Right. You’re talking about moral intelligence, or emotional intelligence, as opposed to something task-oriented.

Vachon: Where AI has been hugely successful is in a very narrow, task-oriented scope. Where it has failed, enormously, is in understanding how it impacts humans. [In the case of social media], the ecosystem of engagement is highly successful because that’s what it was tasked with. But the other ecosystems — human wellness, mental health, getting ourselves out of misogyny and hate and racial bias — all those things suffer because AI has this narrowness about it.

Lookout: We’ve seen this happen all the time, where the low, base kind of impulses of humanity create an opening for technology — how, say, hardcore pornography created the video revolution of the ’80s, for instance, which opened the door for all these other kinds of applications. Is that what’s going on here? Or is it something more complicated than that?

Vachon: I think it’s a little bit more complicated than that. I think this is one of the sort of the greatest challenges of any new technology, understanding its strengths. And then, as a human species, coalescing on which direction we want to take it in. AI shouldn’t be replacing human beings, but instead AI should be looking at how it augments humans. It’s good at what it does, but it lacks contextual learning. And it lacks reliability. And it has problems with things like hallucinations, not being able to know what the truth is. And so, a real solution with AI is not replacing humans, but augmenting them, and having AI as an apprentice to the human expert, and can apply what its greatest strengths are to humans’ ability to understand full context and rules-based systems. When we taught AI how to play a game, how to play “Go,” for example, the first thing it did was it cheated, right? Because we didn’t explain all of the different ramifications of what cheating was. And so, literally, we have to not rely solely on AI without some of the other aspects of humanity.

Lookout: It’s like raising a child, in a lot of ways. Tell me a little about the panel you’ve put together. Did you want to choose people with diverse viewpoints and diverse experience to try to get the widest possible interpretation of what AI is? Was that the idea?

Vachon: No, I think with the premise of “A.I. for Knuckleheads,” we’re assuming that people are coming to this without a great deal of knowledge, and hence the breadth of the panelists that you see. I think we want to talk a little bit about how our own personal efforts are bettering humankind. The desire of this particular panel is to show the breadth of this. In some instances, AI is being used to solve real problems that exist today. And we’re trying to find solutions that may occur in the future. And so the breadth is what we wanted to mostly talk about: how do we share this opportunity of this amazing toolset while still having the discussion around what is our responsibility to this tool set, so that it doesn’t get misused, that it doesn’t cause human harm, and that it doesn’t have unintended consequences as well.

“A.I. for Knuckleheads” takes place Wednesday, May 4, at the Kuumbwa Jazz Center from 6 to 8 p.m. Advance tickets are $20-$50.