Latest in (open source) chatbot/fake AI?

Question

What is the lastest in open source chatbot/fake AI \'technology\' ? Is ELIZA/ALICE/MegaHAL still \'current\', or have there been made any advances in the past decade ?

Answer 1

I created a tool for myself as I wanted to build simple chatbots and manage the whole lifecycle easily: create, test, deploy and analyse. It gained some popularity so I turned it into a startup: https://www.hubtype.com/products/chatbots.html

It's useful to get a chatbot up and running fast using rich interactive elements that messaging apps provide like quick replies, carrousels, etc, which make interaction very easy without the need of fancy AI. However, if you need more complex text interactions you can also integrate state of the art Natural Language Understanding services like IBM Watson or api.ai.

Answer 2

ChatScript is the best open source tool that I've found for developing conversational agents (BSD License):

https://github.com/ChatScript/ChatScript

I have used it for several years and it's extremely programmer friendly (written in a way amenable to how programmers think and use tools). Most importantly, it was written to solve the truly messy problems of parsing natural language sentences with a powerful input pre-processor and a flexible pattern matching style rules engine for matching text.

From the SourceForge project page:

ChatScript is the next generation chatbot engine that won the 2010 Loebner Prize with Suzette, 2011 Loebner with Rosette, and 2nd in 2012 Loebner with Angela (a bug I introduced in the Loebner protocol, not the engine). The technology behind Outfit7's mobile app Tom Loves Angela and ESL chatbots at Japan's SpeakGlobal. 3rd place winner Chatbot Battles 2012 and awarded best 15 minute conversation prize. 3rd place Loebner winner for 2013 and 1st place in 2014.

Also has useful ontology files for nouns, verbs, adjectives, adverbs. Stand-alone or server modes. LINUX (32 & 64 bit) and Windows (Visual Studio 10) and Mac/iOS.

Answer 3

In terms of up to date conversational agent FAQs, see my Meta Guide webpages, "Yahoo! Answers - Best Answers" & "Quora Answers". In terms of non-pattern matching, statistical dialog systems, I suggest looking at the USC Virtual Human Toolkit NPCEditor & the IBM Watson precursor QA system, OpenEphyra from CMU.