People are more into texting than talking. About 65 percent of us would prefer to have a long and painful conversation via Whatsapp than to have one-minute phone call or face-to-face meeting. We text while driving, even though we know it’s extremely dangerous. Families may still sit together at the dinner table, but often, they’re texting other people.

There is simply too much going on — too many people to talk to, too many contacts to maintain — so it can feel easier to live as a digital introvert, dealing with others at your own pace, than to waste energy on real-world conversations you don’t know how to start or how to end.

With chatting being so popular, it’s no surprise that our world is full of chatbots. Sometimes they are made just to replace real-world company, sometimes they have specific functions. Chatbots are also in great demand on the corporate side: They are workers that you don’t have to treat well. They have no emotions, and they do exactly what they’re programmed to. Chatbot hype went through the roof when Telegram’s bots platform got a full-scale API. Lots of businesses immersed themselves in making chatbots then: customer service bots, support bots, training bots, information bots, porn bots, whatever.

The question is, do people really need that many chatbots in addition to real people? We humans are tricky pieces of meat: We experience empathy even when dealing with something that is lifeless by design — your car can have gender, your iPhone a name, and the list goes on.

With chatbots, we seem to project certain emotions onto those scripts and in a way start to think that they are alive, granting them personality they don’t have. This phenomenon is not new — consider Apple’s Siri, the mother of modern bots.

Consider the possibilities. Let’s assume that bots are in fact a new trend and every company in the world will develop one or two — or a hundred, one for every marketing activity. That, by the way, is exactly what is happening.

What will the ecosystem look like? Huge contact lists, with artificial connections a strong majority and real people a minority. The chatbots will talk to you, send you emojis, stickers, funny cat pictures, and links. But don’t kid yourself — they will also learn your behavior and do everything they can to sell you what their masters want you to buy.

That is what happens when you have futuristic technology managed by old-school mentality. And it’s just the tip of the iceberg: Cybersecurity issues are right there. Chatbots are a goldmine for social engineering and crime; they analyze people’s behavior and learn from it.

Phishing; ransomware; theft of credentials, identity, and credit cards — all will be way easier for hackers when they obtain amazing new tools capable of talking people into trouble using their behavioral patterns. Basically, an infected bot would tell you exactly what you want to hear, right when you expect it, so you’d have no reason to be suspicious.

And that’s just basic hacking. The worst techniques have yet to be developed. What about identity duplication? Chatbots are a two-way street: When you’re talking to one, it is learning fast, and with the right settings it can learn not only to be more effective in talking to you, but also to copy your behavior to talk with third parties. By the way, that is exactly what the chatbot in Google’s Allo messaging app is doing.

What could a rogue bot with access to your credentials and a knack for imitating the way you chat do with your mobile banking account or an enterprise chat system such as Slack or Lync? Well, certain elements of science fiction start to look a lot more realistic.

Now think for a moment — is this what we need in our chats?

My personal opinion is a firm No. I’m not saying that bots are an entirely bad idea, nor that they are doomed to fail. Quite the contrary! However, I’m absolutely sure that as long as technology is free of the limitations of the human body, you don’t really need hundreds of them. One bot can do it all: provide support, chat with you, manage your meetings, and all manner of other things. I believe that a machine-learning personal assistant is totally viable and actually has amazing potential, both scientifically and economically speaking.

Here is where we come to another “but,” probably even more troubling than the one of cybersecurity. Realistically, only five companies in the world today can make a “global bot” and benefit from it: Apple, Google, Microsoft, Amazon, and Facebook.

These guys process big data (huge data!), and that helps them make any bot service best in class. With access to petabytes of personal data, these companies’ bots are going to be the smartest and the most adaptive, accurate, and fast-learning. And the chatbots out there today? We’re all just beta-testers improving today’s methods for the big guys.

Unfortunately, as you can see, chatbots are not really about chatting; it’s their function, but not their purpose. Their purpose? Massive marketing research that will ultimately be very profitable to a very few businesses — cybercrime being one of those.

This post appeared originally on the Kaspersky Lab blog, used with permission.