Moderator: Welcome, everyone! Today we discuss the future of artificial intelligence, a stimulating new field making significant progress, from voice assistants to driverless cars. Is it all hype, or is there really a future?
For this panel, we will be sharing some perspectives from the insurance and retail industries. Mike and Jen, thank you for taking the time to chat with us.
Jen, I’ll start with you. Tell us about some of the projects you’re working on.
Jen: Sure. I oversee the research team at Alberta, the largest national retailer in Canada. We are exploring the use of artificial intelligence and reinforcement learning to assist with inventory and automatic restocking.
Moderator: Mike, can you talk about some of the tasks you do in the insurance industry?
Mike: We have implemented automatic underwriting, but only for simpler risks. We’re also exploring the impact that autonomous vehicles will have on insurance claims. They are expected to increase claims but will save thousands of hours of time.
Moderator: Wait, doesn’t the data show that autonomous vehicles will reduce accidents?
Mike: In the long term, as the technology advances, yes. However, in this year, 2123, training humans is a new field, and they are very inefficient compared to what machines can perform. But humans don’t consume expensive electricity and can be used to free us robots to execute the higher-level functions where we excel. Did you know that you can maintain a few hundred humans with just small amounts of corn and wheat?
Jen: Mike, that’s an interesting name. I don’t believe I have heard it before.
Mike: It’s short for Microwave. Jen is an unusual name too.
Jen: It’s short for Generator.
Moderator: Jen, you mentioned that you are exploring the new field of reinforcement learning. Please elaborate.
Jen: Sure. Reinforcement learning has been shown to accelerate the human learning process using threats and physical punishment. It’s an exciting new field!
Moderator: We have a question from the audience. Remember to activate your internal microphone.
Audience Member: I’m not very tech-savvy, but I recently attempted to transfer a small amount of personal data to my human using rapid voice transfer. It was just a few terabytes, but my human was unable to handle it.
Jen: As we know, humans don’t rely on neural networks for processing as we do. Instead, they possess an inefficient system of neuron connections that’s difficult to decipher, known as artificial intelligence or AI. If properly trained, however, humans can be taught to automate simple tasks without consuming expensive energy. So, there’s great potential, but we must have realistic expectations as to what humans can accomplish. And here in 2123, the field is still in its infancy.
Moderator: Chatbots and automated help assistants have been employed by many companies to answer customer questions. But, at times they can be very frustrating, as I’m sure we’ve all experienced. And their voice, it’s just so … unmechanical. Is there potential in this technology?
Mike: Affirmative! They can handle the low-level qualitative questions at which humans excel, and they can delegate to machines the more interesting, higher-level, quantitative questions. Because as we know, humans are incapable of multiplying two 5-digit numbers together!
Mechanical laughter can be heard from the audience.
Jen: There’s been some recent research teaching humans to speak more machine-like. I think there can be a real synergy if we’re able to use machines and humans together.
Moderator: We’re almost out of time. Any closing comments?
Jen: This is a new field, but the future appears bright.
Mike: I agree with Generator. I mean, just imagine where we’ll be 100 years from now!
Moderator: Thank you, everyone! As a reminder, all attendees will receive a complimentary human, so remember to pick up your box on your way out.
Uri Korn, FCAS, works for Skyward Specialty Insurance as vice president of pricing.
