A decade ago, no one would have dreamed that commercial and other organizations would use a virtual assistant to answer queries which until then were the sole responsibility of a salesperson, an office telephone operator, or a receptionist. While organizations still deploy the latter in their front offices, taking cognizance that human interfaces are still preferred, perhaps the day is not far off when the receptionist would make way for a bot/robot, which would interact with visitors, providing answers, directions, and even accompanying/granting access to the visitor to access the desired destination.
What’s a Chatbot
A chatbot is a software solution that pops up the moment you visit a website. Powered by AI, it is designed to engage in a conversation with a user using a series of choices and prompts. Designed to be friendly and helpful in nature, it is used across industries to automate inquiry processes and guide the visitor to the right organizational resource or provide the desired information. In the event that none of the prompts serve the purpose of the user, most offer the possibility of being redirected to a live representative for additional help.
They are everywhere
Chatbot technology leverages Conversational AI to intelligently answer customers’ online queries around the clock. While ready or customizable solutions are available for smaller customers, tech giants Microsoft, Facebook, Google, Apple, Samsung, and Amazon have developed their own chatbots. They are everywhere – in smartphone applications, voice assistants, and more. The Covid-19 epidemic is cited as one of the major reasons for its exponential growth in the healthcare sector. One estimate puts the number of chatbots on FB Messenger alone at over 300,000 bots. 80% of netizens are estimated to have interacted with a chatbot at some point or another when browsing the internet. In 2023, cyber thought leader Gartner has predicted that almost 38% of all organizations (1) would plan to implement a chatbot solution in the next two years.
Why the chatbot
Chatbots are harbingers of our times. Organizations recognize that customers are kings. They want answers immediately, and are ready to move on should they not get a quick response. One study puts the percentage of customers willing to abandon a brand due to an unpleasant customer service experience as high as 30% (3). Organizations can’t afford a missed sales opportunity, especially with customers spread across the globe, and sales teams not being on hand to field inquiries due to varying time zones. Chatbots fill this need, offering a positive user experience most of the time.
The benefits of the chatbot
Chatbots offer many advantages but primary amongst them are:
- Positive and personalized user experience as a consequence of their analyzing ability, swift responses, language processing abilities, and documentation features
- Round-the-clock assistance, even at times when the offices of the organization are closed, and representatives are not available to offer assistance
- Single-point capturing of customer data and information due to the structuring of questions posed by the chatbot
- Strong sales focus due to their ability to speed up the selling process, as well as offer promotional material that may expedite buying decisions
- Optimized costs as a consequence of the automation of customer service team agents/representatives
The drawbacks and the risks inherent in chatbots
Despite their benefits, chatbots have their drawbacks and experts say they can be a security threat due to the technology that is used. Some of the major drawbacks of chatbots:
- Possible compromise of user data due to information being gathered through the chatbot becoming part of the data model, which can then be misused
- Violation of privacy norms and other rules for data compliance
- Malicious text and code generation features can be misused and put to unethical use
- Fake business opportunities via malicious apps can be created by bad actors
- Intellectual property and copyright concerns arising from conflicts with copyrighted property
- Open source license issues in the event that unauthorized use of open source libraries has taken place when developing the chatbot
- Restrictions on AI development due to the limitations imposed by ChatGPT, which mandate no further AI development can be undertaken on the chatbot
- Development takes time and testing before the chatbot is made available for use on the open web
- Chatbots need maintenance and are not a one-time solution. Over a period of time they need re-modeling and fresh feeds into the data model to make them relevant
- ‘Jailbreaking’ of chatbots via malicious instructions that cause the chatbot to misinterpret commands and produce irregular responses
- Facilitation of scamming and other social engineering acts with the opening of ChatGPT to the internet in general, a move that is widely seen as aiding and abetting scammer activity
- Data poisoning of the data sets in language learning models that power chatbots resulting in erroneous responses
- Legal issues can also be a cause for worry when using chatbots, especially with regard to contractual violations
Some best practices
Authentication and authorization are two of the main concerns when dealing with chatbots. Considering the security risks involved it is imperative that the user is identified by the chatbot before further access is permitted. The next step would be to ensure that only such access rights are granted based on this identification. That said, however, there are still some additional features that will help ensure a modicum of safety when dealing with chatbots.
- Multi-factor Authentication or Two-factor Authentication where the access code is sent to a handheld device and to an email ID
- Deployment of a user ID and password or biometric authentication of the user
- User timeouts from the active session to ensure that fresh logins are made
- End-to-end encryption of the interaction to prevent unauthorized use of the data
- Self-destructive messages using a timer of the kind presently being seen in many other apps like WhatsApp
The US Administration’s Oct 2023 Executive Order which takes a proactive approach to monitoring AI-generated tools is seen as a step in the right direction to check the relentless development of commercial and conversational AI.
Final words
As the debate rages about chatbots, the jury continues to remain out. The generation and use of chatbots however continue unabated. Perhaps some light on the pros and cons of the chatbot can be seen by taking a look at DebunkBot, an AI-powered bot that was started by a project team from MIT, Cornell University, and the American University. The site, which was tested with a large sample of internet users, is resolving doubts in users’ minds about conspiracies surrounding the use of bots by presenting facts and persuasive arguments. Reports show that almost 20% of that large sample (2) came away convinced about the genuineness of bots as a tool intended to collect information and answer queries that add value to users, rather than as a means for malfeasance.
Yet for all the valid arguments that DebunkBot propounds, the questions from those who cite incidents and pieces of evidence of its potentially harmful uses, won’t go away.
Good or bad, useful or harmful, it seems unlikely that the use of bots will abate. If anything, as projected by Gartner in their 2023 study, we are only going to see an increase in their numbers. Let’s hope that augurs well for cybersecurity!
References:
