We sit down at New York’s Bryant Park Grill to discuss how the combination of humans and machines will lead to improved accuracy, relevancy and trust.
Dennis Mortensen: So any good assistant will have some understanding of the boss she works for, and Amy works for me.
Joie Chen: Amy is not a real person?
Dennis Mortensen: And she is not a real person.
Joie Chen: So we are in this era where machines are making more and more decisions even in a place like a restaurant, the Bryant Park Grill. Data drives what happens here, too, whether it's menu planning or produce buying…what’s the future for that?
Sreekar Krishna: Artificial intelligence is now coming into play to bring together data, analytics, and the human decision process. They're all coming together under the umbrella of AI and we are now transforming the way KPMG is doing our business.
Traci Gusher: There's a factor of trust and quality in the application of AI. I would say that that goes directly back to the trust and the quality and the data under the AI.
Dennis Mortensen: So we're a tiny company, right? We're a 150-man band, but we've got a hundred people doing nothing…
Sreekar Krishna: …But data cleaning
Dennis Mortensen: …But data cleaning.
Sreekar Krishna: Garbage in, garbage out. Right? If you don't clean the data, we are going to end up creating machines that learn the noise rather than the signal. How do we take the mistrust out of the loop … by building the confidence chain.
Adam Devine: So taking it out of the loop…by putting humans into the loop.
Sreekar Krishna: Into the loop.
Traci Gusher: Into the loop.
Adam Devine: You build the confidence in the machines.
Sreekar Krishna: Yes.
Traci Gusher: There is no such thing as I'm going to plug this in and, BAM, you've got great AI and automation. We're not at an autonomous self-learning appliance yet.
Dennis Mortensen: What we do at x.ai is we run a network. That means that when we service one person, we can do kind of okay. If we service 100,000 people, then all of a sudden we have this superhuman assistant that can learn from 100,000 people.
Adam Devine: The whole principle is pairing intelligent automation, machine learning robotics with people to handle the exceptions that machines can't do, and then making the machines smarter incrementally by taking in the information from a human interaction.
Traci Gusher: Yes. The more we interact with these devices, the smarter they will get and the better they will get over time. I don't think we've even begun to touch where it could go.
Adam Devine: So if you’re able to generate good quality data at scale, and you can automate the process of applying that data to an AI system, suddenly, if you're a giant bank, anti-money laundering is solved.
Traci Gusher: Right.
Adam Devine: Know-your-customer is solved.
Sreekar Krishna: At KPMG we are seeing the flooding of data, it is fantastic. I mean, this is the time to look at what AI can do and how it can bring greener pastures on the other side.
Traci Gusher: The most beneficial use cases for AI are in the back office and the middle office and machine-to-machine. There are financial benefits. There are creativity benefits. There are strategic benefits. There's so much there.