How many of you used the Google Duplex? Yes, the latest Google assistant (passed Turing Test) that can book appointments for you. It is supposed to help businesses and interact just like a human, by understanding not just the language but also the nuances of human speech. So, you must be very excited to see Google Duplex hit the market and see if you can spot it amongst humans in service industries. But, how many of you have spoken to or have heard of Eliza, a Rogerian psychotherapist? When I first interacted with Eliza I was slightly frightened and rather mystified by it. I was barely a teen then and decided to uninstall her from my computer. It would be years later, that I realized that I had spoken to a chatbot, that uses words from your input to form an output and continue the conversation. In case you want to talk to Eliza, you can talk to her online. And for those who are wondering about its origin – the initial version of Eliza was released in 1966.
However, Eliza wasn’t perfect, because she failed the Turing Test. A Turing Test is passed when a machine converses with a human and by listening to the conversation, you are unable to ascertain as to which one of the two is the human subject, and which one, the machine. Another bot most of us might remember is Clippy, the helpful paperclip in MS Word which did anything but help, interrupting you in the middle of work and asking if you needed help in writing a letter or creating a spreadsheet.
Apple has built and refined Siri, Microsoft Cortana is out of the bag and this year Amazon’s Alexa is changing the dynamics of the table-top speakers in homes around the globes. However these are not specifically chatbots, but what you might call web-bots, that connects you to the web, without you having to type out something meaningful.
So, from the Xenophobic Microsoft Chatbot Tay, to lousy customer care chatbots that Banking Institutions are trying to build (they barely solve our problems), today we will be discussing some of the best and worst of bots and the technology that gave birth to them.
Although the word bot itself brings back memories of a bullet shooting maniacal robot from Terminator, bots are not all bad, and truth be told, they were developed mainly for good purposes, such as copyright bots that check for plagiarism, or spider bots that are used by search engines, such as Googlebot or Bingbot.
Sometimes you might Google things such as the weather or football scores and you might find the adequate results displayed right at the top. This is through the help of bots who track relevant information and keep that information updated with time so as to provide you with suitable results when you make some common searches.
The rise of the e-commerce wars has led to a new type of bots called trader bots, that collect pricing data of in-demand products by exploring the internet. These are used by several websites such as Amazon or eBay. More and more online retailers are using these to remain in business by collecting pricing information and then setting their own prices competitive to woo more customers.
Data-bots, which are also called media bots are the ones that help get the latest data related to specific events, for example, say there has been a natural disaster, an earthquake, that has affected some regions around the world. If you search “earthquake” in Google news, you will find everything from the latest articles on the disaster, as well as real-time reports, videos updated by people affected and more. This is possible with the help of these so-called media bots, who collect information in a manner so as to connect the dots and give a holistic picture of an event that takes place.
Customer Care Bot is one bot that generally takes the flak because of singular approach (displaying same link for different issues) while answering questions or solving a problem. These bots are in fact the bots that are developing at the fastest pace, and we will see most of in the near future. On a positive note – they are getting better at recognizing speech and text and have progressed far from showing results based on word placement in sentences. Customer care bots are even striving to understand our mental condition depending on the style of our speech or text and accordingly will help us. In fact, in future the initial conversation might get handled via bots in case of helpline numbers. This will help companies and the government to provide better support to a wider audience through the use of technological advancements.
Like everything else, including the nuclear technology, everything that could be beneficial to humans has also, always been slightly modified to achieve some evil gains. The same is applicable to bots. There are certain types of bots that cause the most harm, directly or indirectly, and we will look into some of them today-
What is website traffic? Well, it is just the count of people going through websites at any given time. In 2016, bad bots made up almost 29% of all website traffic whereas good bots hung to a measly 6%. Of the worst among these fraudulent bots are the click bots. Click bots, do exactly what their names suggest- they click on links and ads thereby disrupting advertising campaigns (the human traffic is wrongly calculated due to the actions of bots) and increasing wasted dollars by clicking on advertisements put by companies in PPC (pay per click) model on websites like Google.
Very similar to the click bot is the download bot, that downloads certain data from the internet to malign the results of analytics companies that are conducting a study or ranking algorithms. For example, you might build a game and put it in Google Play Store. Barely 10 people download it. Then, you build a self-cloning download bot that increases the download count of your app and increases its rank in the Playstore so that more and more people are shown the app when they visit Google Playstore.
Another type of dangerous bot is imposter bot, which mimics the behavior of humans (the reason why websites are using a captcha that is getting tougher and tougher with time). Their intent is to penetrate layers of defense in websites and then multiply quickly to increase traffic in your website at such speed, that the server is unable to handle the load and goes offline. This is called a DDoS attack, and several such attacks have happened since the beginning of the internet and they have caused widespread harm and taken thousands of websites down.
Some other bots that do not interfere with normal work, but instead are used by people for their evil benefits are spam bots. They tend to steal articles and data from the internet and stores them to be republished in another form. However such content is flagged by google search engine and it ultimately downgrades their websites. Spam bots also collect personal information from the internet, such as random mobile numbers and email ids, so that the data can be used to send advertisement links (sometime malicious content).
Bots have great potential, whether they are good or bad. In this section, I would like to talk about a famous bot that really went out of hand and had to be shut down. Yes, I am talking about Tay, Microsoft’s teenage AI chatbot that was built to learn from and converse with users in real time. Unfortunately, people took advantage of the possibilities and did the most unlikely thing by using its machine learning capabilities – people taught Tay slang, racism, sexism and made it develop an ugly-gangster like personality.
It was taken off very soon, but one can well imagine that if an entire network of interconnected bots got used in the same manner with someone having malicious intent, and in case the bots were in control of real-world electronics such as security cameras, door controls, etc, the result could be even been something out of a Terminator movie.
So what do you think? Will bots take over the world or will you just have to keep filling captcha in almost all websites by the end of this decade?