Bots and Conversations
Monage 2017: Bots and Conversations. With the Internet our conversations take place outside the networks.
26-Mar-2017Version 2: 2023-03-18 10:09:41

This is the video for my talk at Monage March 21st 2017. I thank Chris Fine for recording this video using LG5 phone. He did the best he could given the constraints. You can follow the PowerPoint for talk using the controls in the embedded image on the right.

I typically talk to engineering groups which are focused on the technology we use to communicate. For MoNage I started by asking what it means to communicate independent of the particular technology.

The idea of a chatbot, in the sense of having a conversation with a computer goes back at least half a century with Joe Weizenbaum's Eliza program showing how easily simple rules can give the appearance of a meaningful conversation and how readily people can engage even when the know they are speaking to a very simple program.

The lessons of early efforts like Eliza and SHRDLU are two-edged. They show how easy it is see human-like intelligence where there isn't any. This isn't just about our tendency to give computer too much credit, it is also a reflection of how we understand and misunderstand other people. Communications requires a conversation to establish and maintain a shared context.

In the last half century technology, has given us the ability to assimilate more data and over a wide distance but we err in assuming a massive number of bits is the same as wisdom and that there is one overarching center to which the data flows.

Technology has allowed us to communicate over a distance but language works metonomy in repurposing words for attributes to describe the whole. The term “telecommunications” was coined because the telegraph enabled us to send message over a distance but it's only a means. With the Internet, we no longer depend on the telecommunications provider for carrying the message but instead we have a conversation that takes place entirely apart from the telecommunications infrastructure.

I make this clear by showing that we can move a physical object, a toy, from one coast to another by simply pass its name. Returning the toy to the store on the east coast and buying a corresponding toy on the west results in the very same toy appearing on the west coast. The question is not whether it is indeed the same toy but rather what a mean by “same”. This ability to take an operational view of meaning is a byproduct of the articular understanding wrought by software.

This same understanding applies to chatbots and other conversations. Computers aren't the same as people which makes it all-the-more important to understand how the different kinds of intelligence and meaning interact.

The talk was only the start of a vital conversation.