You’ve probably chatted to a robot more recently than you think. Artificial Intelligence-driven language models are now a standard feature of customer service, online banking and everyday research. They’re also a regular fixture on the dating scene, with as many as 1 in 5 accounts belonging to bots, according to some estimates.
In an ideal world, chat bots make our lives easier, lines shorter and business costs lower. In an alternative dystopian version, however, they’re using increasingly sophisticated algorithms to trick us, catfish us and rob conversation of its human touch. Here are five ways to spot if you’re talking to a bot online.
1. They Speak Like “The Terminator”
The average American adult might read at 7th-to 8th-grade level, but most are able to raise their game when it comes to talking or chatting, especially in a dating scenario. Not so with chat bots, which are usually stuck in a middle school hallway, fumbling their lines under pressure and getting their wires crossed. In particular, bots struggle with idiom and nuance, find it difficult to switch fluently between formal and casual registers and revert to set phrases when they’re confused. In short, they can sound robotic.
Unfortunately, that doesn’t mean they struggle with offensive language. AI models have already shocked researchers with their ability to learn toxic stereotypes relating to race and gender.
2. The Use of Repetitive Language
Humans can choose from infinite ways to say nothing of any substance at all, but bots will always be botting. For now, at least. The original chat bots (for example, customer service or frequently asked question bots) relied on target phrases or keywords to trigger set responses. You could rephrase the question or try a different angle, but the bot would repeat the same response. However, the new generation of “generative pre-trained transformer” (GPT) models can sound scarily human, so be warned. Programs such as ChatGPT, which was introduced in November 2022 and has already passed 100 million users, predict the next best word rather than relying on pre-set responses.
3. They Reply Too Fast
Any experienced online dater will know that replying to your Tinder match too fast sends the wrong message. Your scammer bot, on the other hand, is in a hurry to share you a link to a crypto-trading program or malware-infected website. So if you’re getting almost instant replies, no matter how complex or fact-heavy your questions, there’s a good chance that you’re talking to a bot. As soon as that link drops, or a request to take the conversation to another platform, that’s a sign that you’re finally about to connect with a real human (cybercriminal) ready to close the deal.
Which is why you should take your time to check our 11 ways to outsmart a romance scammer, first.
4. Bots Have No Hot Takes to Offer
You’ve probably been shocked at those YouTube videos of spring breakers or college students struggling to name a single foreign country. But some bots take ignorance of current affairs to another level. Try dropping in a topical reference or contemporary news item to the conversation and see what happens. Bots don’t read the news, understand memes or stay up to date with the celebrity gossip. So if your chat partner seems to be living in their own world, cut off from current reality, it could be because they’re not human.
Having no hot takes is how bots talk.
5. Bots Take Language Literally
If you were to read a transcript of everything you said yesterday, you’d be struck by one thing in particular. A lot of it makes no sense whatsoever. Humans might have the unique gift of language, but we also have an innate ability to speak in fragments and nonsequiturs. Because of that, we also find it easy to make sense of the chaos and extract meaning from each others’ mangled sentences.
Bots, on the other hand…not so much. Throw a random sentence or reference into a conversation with a bot and it will seize up. It may even apologize for its lack of understanding and request clarification. That’s your cue to turn up the crazy and go full stream of consciousness. If your bot can’t keep up, it’s the code talking.
How to Catch a Bot and Protect Yourself Online
It’s getting harder each day to spot bots on language characteristics alone, such is the speed with which artificial intelligence and machine learning are evolving. Programs such as ChatGPT seem eerily sentient already, and are designed to learn from their mistakes. You might need to look at the surrounding context to tell if you’re talking to a bot. For example, if you’re enjoying a stilted conversation with an exceptionally pretty woman who wants to talk crypto, that’s a good sign you’re being fattened for a Chinese pig butchering scam. As a general rule, make your excuses and leave as soon as any conversation online turns to emergency funds, money or sweepstakes.
You can also research a claimed ID with Spokeo to check if it belongs to a real human. Simply run a search of the linked email address or phone number to see if the public photos attached to them match the ones sent to you. You could save yourself a lot of heartache in just a few seconds.
Sources
Privado VPN – Avoid Tinder Scams
Center for Plain Language What is readability and why should content editors care about it?
Futurism – Scientists Alarmed When Robot Immediately Becomes Racist and Sexist
Tech Target – Pig butchering scam example