Advances in Conversational AI

I gotta say, when AI is able to converse like humans, a lot of bad stuff will happen. People are so used to the other conversation partner having self-interest, empathy, being reasonable. When enough bots all have a “swarm” program to move conversations in a particular direction, they will overwhelm any public conversation. Moreover, in individual conversations, you won’t be able to trust anything anyone says or negotiates. Just like playing chess or poker online now. And with deepfakes, you won’t be able to trust audio or video either.<p>The ultimate shock will come when software can render deepfakes in realtime to carry on a conversation, as your friend but not. As a politician who “said crazy stuff” but really didn’t, but it’s in the realm of believability.<p>I would give it about 20 years until it all goes to shit. If you thought fake news was bad, realtime deepfakes and AI conversations with “friends” will be worse.<p>(If you go 50 years out people can start building sleeper bots with reputations to subvert community consensus, eg about science or politics. All our systems will be subverted. Don’t believe me? How far are we from that with “skeptical” blog comments and tweets or sorta-believable allegations ruining people’s reputations?)

Links to papers:<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1811.00207" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1811.00207</a> - Towards Empathetic Open-domain Conversation Models: a New Benchmark and Dataset<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1811.01241" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1811.01241</a> - Wizard of Wikipedia: Knowledge-Powered Conversational agents<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1810.10665" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1810.10665</a> - Engaging Image Captioning Via Personality<p><a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1902.08654" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;1902.08654</a> - What makes a good conversation? How controllable attributes affect human judgments<p>And a demo of a bot that they&#x27;ve produced: <a href="https:&#x2F;&#x2F;www.facebook.com&#x2F;Beat-The-Bot-212188996195556&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.facebook.com&#x2F;Beat-The-Bot-212188996195556&#x2F;</a>

<p><pre><code> Human: I&#x27;m a huge football fan - the Eagles are my favorite team! Knowledgable model: I&#x27;ve always been more of a fan of the American football team from Pittsburgh, the Steelers! </code></pre> While an impressive improvement from &quot;I like football too,&quot; this stilted attempt at incorporating relevant information ends up coming off as &quot;how do you do, fellow humans!?&quot;

It&#x27;s a bit weird that having the agent make up stuff about itself randomly (&quot;I&#x27;m a construction worker&quot; or &quot;I build antique homes and refurbish houses&quot;) is considered a good response. Something like &quot;I don&#x27;t need to make money, I&#x27;m a bot&quot; is what I&#x27;d expect. Or alternately, in a game situation, the bot could be playing a particular role it&#x27;s given beforehand and judged by how well it stays in character.

They created an Ai bot game as well<p><a href="https:&#x2F;&#x2F;www.facebook.com&#x2F;Beat-The-Bot-212188996195556&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.facebook.com&#x2F;Beat-The-Bot-212188996195556&#x2F;</a><p>They give you a character to play, you are given 2 answers one from a bot one from a real person.<p>You respond to the one you think is best, interesting, took too long to connect me to another &#x27;player&#x27; didn&#x27;t get a chance to actually play yet

Fascinating topic. Unfortunate that a voice in the back of my brain has to say &quot;Oh I bet if they good at this they&#x27;ll just use it to target ads better&quot;