Large language models (LLMs) can "lie", which we define as outputting false statements despite "knowing" the truth in a demonstrable sense. LLMs might "lie", for example, when instructed to output misinformation. Here, we develop a simple lie detector that requires neither access to the LLM's activations (black-box) nor ground-truth knowledge of the fact in question. The detector works by asking a predefined set of unrelated follow-up questions after a suspected lie, and feeding the LLM's yes/no answers into a logistic regression classifier. Despite its simplicity, this lie detector is highly accurate and surprisingly general. When trained on examples from a single setting -- prompting GPT-3.5 to lie about factual questions -- the detector generalises out-of-distribution to (1) other LLM architectures, (2) LLMs fine-tuned to lie, (3) sycophantic lies, and (4) lies emerging in real-life scenarios such as sales. These results indicate that LLMs have distinctive lie-related behavioural patterns, consistent across architectures and contexts, which could enable general-purpose lie detection.

It can be difficult to make sure chats are responded to within the right amount of time after a customer asks a question. It can be expensive to staff live chat 24/7. However, only staffing during the day could result in lost interactions that drive revenue. Chat should not be put on every webpage on your site. Only place them on high-intent pages related to sales, or your sales team could get inundated with unrelated questions all day.

Chatbots don't have the best reputation. From creating their own secret languages to spouting racist and sexist comments to simply not being very helpful, the emerging technology still has its very public shortcomings. While increasingly being used by businesses for customer service, the technology is still in its rough, early stages, Forrester vice president Julie Ask said. But 2018 could be the year when the technology finally begins to mature. "(Chatbots) certainly have very limited functionality," Ask said, adding that they're a "work in progress." "Most are hard-coded decision trees.

Chatbots are not new, especially for this generation. In fact, back in 2000, if any of you had ever been an avid user of AOL Instant Messenger, you probably tried chatting with its bot, SmarterChild. I used it to request movie times and to be honest, called him names too (apparently, he didn't mind). SmarterChild was the only bot at that time and was loved for so many reasons. Who knew he would rest in peace so soon?