Chatbot Aims to Trap Suspected Sexual Predators

NegobotLast week there was news of a chatbot developed to ferret out suspected sexual predators of underaged victims. The chatbot, called Negobot, was developed by Carlos Laorden and other academics from the University of Deusto in Bilboa, Spain. Negobot is a AIML-based conversational agent that poses as a child on Internet chat forums and social networks and employs various sophisticated methods to draw out those who exhibit pedophile behavior. Not only does Negobot use natural language processing and machine learning, but it also leverages aspects of game theory to achieve its goal of inferring if someone has a high probability of being a sexual predator.

Negobot uses an AIML structure based on the Galaia Project to find appropriate responses to questions. The chatbot was primed with pedophile conversations from an existing law enforcement database. The database of conversations is stored in English, so Negobot translates all input into English before processing. Ongoing conversations are also added to the existing database.

Applying game theory concepts, the chatter bot views each conversation in terms of seven potential levels. In each successive level, the conversational partner shows more interest in the bot. At some point, the conversation transitions to one where sex is discussed explicitly.  Negobot analyzes the ongoing dialog and is aware of the level of ‘sliminess’ of the conversation. It bases its responses on this knowledge and adjusts its strategy based on the level, all the while continuing to play the game of  drawing out more information and damning conversational evidence from the suspected predator.

Negobot assigns a pedophile probability to the dialog partner based on the substance of the conversation. If the other person tries to end the conversation once it becomes clear that Negobot, or rather the child it’s posing as, is underage, then the probability of pedophile behavior decreases. If the person continues to ask questions of a sexual nature, the probability increases and Negobot poses questions to discover more personal information.

Laorden and his colleagues haven’t deployed Negobot into the real world yet, but they’re continuing to refine the chatbot. It may not be long before would-be sexual predators are being ensnared by virtual agent technologies. A full discussion of Negobot’s technology and capabilities is described in an abstract about the conversational agent published by the creators.

Kickstarter Chatbot Campaign

KickstarterLooking for a way to combine your enthusiasm for chatbot / virtual agent technologies and your addiction to Kickstarter? Well, you’re in luck! Mercer Engineering Research Center (MERC) has recently launched a Kickstarter campaign to gain funding for their Histochat project.  The campaign is set to end in 23 days and they’re a long way from reaching their goal, so have a look and see if this is something that deserves your support.

Based on the Kickstarter videos, it looks like Histochat is an educational tool that will provide students with an interactive experience with real historical figures. The MERC team is looking to use ChatScript to build out conversational chatbots representing possibly five historical personalities: Albert Einstein, Amelia Earhart, Mahatma Gandhi, Martin Luther King, Jr., and Susan B. Anthony.

Students will apparently use a smartphone to access the chatbot agent representing one of these historical figures. They’ll be able to engage the historical personality in a conversation by asking questions about the person’s life and accomplishments. Check out the Kickstarter videos. The demo isn’t particularly exciting, but the idea of creating chatbots to bring history to life for students is a good one. Hopefully the Kickstarter campaign will get more momentum in the coming days.

Ray Kurzweil’s Ambition and Musings on The Future of Virtual Agent Technology

Digital BrainSingularity Hub did an interview with Ray Kurzweil back in January, during which Kurzweil talked about his vision for an artificial intelligence that will act as a trusted personal assistant to humans. Kurzweil had only just started his stint at Google when the interview took place. He briefly shared his vision of constructing an artificially intelligent software system that mimics the hierarchical architecture of the human brain. It remains to be seen how successful Kurzweil and the team at Google will be in their endeavor. Whatever the outcome, the race to produce smarter and smarter digital entities is definitely underway. As Gary Marcus points out in his review of Kurzweil’s book on building a brain, there are many different machine learning techniques and cognitive systems that are being researched today in the public and private sector. Whether Kurzweil’s hierarchical approach pans out or not is really irrelevant. Advanced AI that can understand human intent and provide answers to human questions will happen. The progress made in this field is bound to influence commercially available virtual agent technologies, both in the mobile personal assistant and in the enterprise and customer support virtual agent domains.

Older conversational agent technologies will most likely be superseded by new ones. The work that Ray Kurzweil, the DeepQA team, and many other artificial intelligence researchers are engaged in today is producing techniques that far outpace the rudimentary pattern matching technology deployed in most simple chatbots. In the hands of dedicated and savvy bot masters, chatbot scripting languages such as AIML can be used to create impressive question answering agents. But unless it is combined with natural language processing, search, and machine learning algorithms, AIML by itself can’t produce a truly effective virtual agent.  It’s sporty to make any predictions when it comes to the future of artificial intelligence, but one pretty safe prediction would seem to be this: the intelligent virtual agent that one day passes the Turing Test won’t have been created using basic AIML pattern matching technology.

For commercially viable virtual agents in the field of customer support, incorporating strong search capabilities would seem to be a must. Search can be combined with pattern matching against a broad database of known frequently asked questions to provide web or mobile users with basic self serve information. Text and/or speech recognition and natural language processing would also seem to be non-negotiable skills for a virtual customer service agent. User profiling and targeted recommendations are capabilities that advanced virtual agents should also have in their toolkit. We could go even farther and list attributes such as a sense of humor, the ability to detect human emotion, and empathy. All of these would be desirable qualities in a customer-facing virtual agent.

Perhaps as Kurzweil / Google and others work towards recreating the human brain in digital form, advancements in cognitive computing, speech recognition, natural language processing, and other interrelated fields will be the outcome. It will be hugely interesting to see how software vendors in the virtual agent and personal digital assistant space capitalize on these breakthroughs to improve and reshape their commercial offerings.

Creating a Virtual Agent Chatbot for Your Business

Build Your ChatbotWhile it’s possible to develop a smart virtual agent from scratch, there are a number of software companies that provide easy-to-use and cost effective options for creating a customized virtual agent, or what is commonly referred to as a chatbot. I’m not affiliated with any of these companies. I’ve tried out some of the virtual agent products and I’ve chosen two at random to spotlight in this blog post so that you can get a feel for what’s involved in creating a conversational virtual agent for your business.

The chatbot companies we’re looking at in this post are MyCyberTwin and  At first glance, the companies seem quite different. MyCyberTwin presents itself as a virtual assistant vendor with products geared towards businesses. Chatbot4U looks more like a social site geared towards a younger crowd interested in creating and talking to chatbots impersonating popular teen idols. When I looked a bit further, though, I found that these two virtual agent providers have relatively similar business models, tools, and pricing structures.

Building Your Custom Chatbot

Both MyCyberTwin and offer fairly straightforward user interfaces that allow you to program, or ‘train’, your chatbot without having to write any code or markup.  Both offer chatbots that you communicate with via typed input. You don’t speak directly to the virtual agent, but rather talk to it via text messages. Once you’ve created your free user account, you can create a blank virtual agent chatbot and then start filling its knowledge base with input (questions) and output (response) phrases.

With MyCyberTwin, you can start by specifying a website that you want your chatbot to reference to see if it can find answers to questions. This is a great feature, because it means that your chatbot will have at least some limited ability to successfully respond to questions that you might not be able to predict, or exactly replicate, in advance. For example, if someone asks the chatbot how to contact you, but you didn’t think to include this question in the chatbot’s knowledge base, there’s a good chance the chatbot may be able to point the person to your website’s “Contact Us” page, since it will use keywords to locate the appropriate content.

MyCyberTwin allows you to select one of several very life-like avatars to represent your virtual chatbot. When you embed the chatbot code into your website, the animated avatar appears in a separate window and invites the visitor to engage in conversation by typing in messages.

With Chatbot4U, you can set your virtual agent’s avatar by uploading a photo. The avatar is not animated. As you train your chatbot, you have the ability to add multiple layers, or ‘go backs’, to an ongoing conversation. For example, if the visitor types in the question “How are you?,” you can train the chatbot to respond “Fine. And how are you?” Then you can have the chatbot say something more or less appropriate when the visitor responds. These types of meaningful threads are typical and essential to human dialog. They’re difficult to recreate with current chatbot technology, however, because a chatbot has a very limited memory and can’t remember what it said beyond its last utterance.

Another useful feature of the Chatbot4U platform is that it allows you to add knowledge modules to your virtual agent. These modules endow your chatbot with the ability to tell jokes, to provide current weather information for any location, and search Wikipedia.

Training your chatbot is a simple procedure on both of these platforms, but it’s a time consuming endeavor. I recommend that you pick an area of your business that you want to concentrate on, such as your FAQs. You can also use my blog post Does Your Business Need a Virtual Agent? for some ideas on how to train your virtual assistant. You’ll want to create as many possible questions as you can think of and provide the best answer to the question. Training the virtual agent is simply a means of defining what the chatbot will say in response to pre-defined questions. To liven things up a bit, you can provide more than one answer to the same question and the virtual agent will randomly vary which response it uses.

Testing Your Chatbot Trial Version

MyCyberTwin and both offer a free, personal version of their chatbot technology that you can try out with no time limit. The personal version lacks some of the features of the business version. Starting with a personal virtual agent will give you practice in using the platform and help you get a feel for creating conversational inputs and outputs.

Both vendors also offer you the option of running a 30-day trial of the full business chatbot. You can take advantage of the complete functionality available for training your virtual agent, including providing it access to web-based information sources, adding apps, and creating go-backs for more realistic conversations. I recommend that you converse with your chatbot at least a dozen times before you publish it to your website. Make a note of any questions you can think of that the virtual agent can’t answer. Converse with the agent as you would with an actual person and make sure that it has good responses for typical greetings and questions. You can also train the chatbot to direct the conversation towards its preferred topics about your business.

How Much Will Your Chatbot Cost?

At the time I’m writing this post, both MyCyberTwin and offer basic, introductory chatbot solutions at the low-end of the price range for intelligent digital agents. Both will host your chatbot for around $25 a month. The fee includes some level of reporting, which will allow you to track conversations and train your virtual assistant to answer questions that it missed.

In future posts, we’ll look at the process for coding your own customized chatbot from scratch.  We also recommend that you visit if you’re interested in seeing a thorough listing of virtual agent vendors available on the market today.