Name: 判斷你是否在和機器人哈拉的5種方法 (5 Ways to Tell If You're Chatting with a Bot)
Uploaded: 2021-01-14T09:28:25.000Z
Duration: 9 min 29 s
Description: 【看影片學英語】數萬部 YouTube 影片，搭配英漢字典即點即查，輕鬆掌握單字發音與用法，長久累積看電影不必再看字幕。

Technology has brought us a lot of wonderful things, such as ordering a pizza with a single

分詞構句

button, or receiving text alerts -- about how close your pizza is.

轉承詞

However, not everything is sunshine and pizzas in the digital age.

Once upon a time, you could say with confidence that the person you were talking to is, well,

That's no longer the case as robocalls and chatbots, computer programs designed to simulate

human interaction over the phone or on the web, are more common than ever, and getting

If you want to know whether the person on the other end is flesh and blood, or silicon

and circuits, you'll want to hear about these five tricks for catching chatbots in the act.

Number One: The Account is too Active While you're poking around for your favorite

game, Pizza Master Pro, you notice one account seems to be posting quite a bit of content

While it's understandable that people would be enthused about the best pizza simulator,

twenty posts a day seems like a bit much.

My point is that since bots don’t need to sleep, they can spend all night spamming your

So, if one account is posting comments 24/7 365, you've probably got a bot on your hands.

While we’re on the subject of samey posts…

Number Two: Suspiciously Similar Comments Not everyone has a direct line to the muses,

so maybe you shouldn’t be too harsh on people for leaning on the same old jokes.

It might get annoying, but it doesn’t mean they’re a machine.

Now, if they’re posting the exact same comment all the time and every single one has a link

to the same website, you probably have a bot on your hands.

An annoying bot, which is the worst kind; and I’m including the ones designed to steal

Chatbots may be complicated machines, but they’re still just machines.

Creative thinking isn’t exactly their forte, or their pommel either.

And if you get that joke, you deserve a prize!

Bots pick words out of comments posted by real humans and run them through an algorithm

The complexity of the algorithms and number of responses will vary depending on the resources

available to whoever programmed it, but it doesn't take that much programming knowledge

This is easier to pull off in a text-based medium like Facebook, Reddit, or YouTube’s

comment section, but it can work over the phone too.

The weakness here is that there are only so many responses the programmer can plan for,

so a similar question or statement will usually result in an identical response.

This means that if you're ever not sure what you're dealing with, you can try asking them

If they respond with the same thing over and over again, and don't notice you're asking

slight variations on the same question, then you've got a bot in your chat.

Number Three: They Have Trouble Answering Simple Questions

Chatbots are pretty good at pulling trivia out of thin air.

They're basically just a search engine with delusions of humanity, so it would be pretty

embarrassing if a bot designed to pass as a human can be outsmarted by Siri or Google

Conversations are more than just listing facts and answering yes or no questions.

They require a little bit of common sense, a field that computers don't exactly excel

Bots often don't know what to do with jokes and have no understanding of context or the

You can take advantage of this failing through a trick known as “Two Step Dissociation.”

This is where you ask the potential bot a question, then follow it up with a second

question that only makes sense in the context of the first.

When explaining this concept, digital marketing expert Anthony Helmstetter uses the following

The human caller asks the bot where it’s located, and it replies, "Seattle."

The caller then asks, “What’s the weather like outside?” and the machine replies,

“Can you please rephrase the question?”

A human would understand that the caller was asking what the weather was like in Seattle,

but the bot just can’t connect those dots.

If your customer service rep can’t make a simple connection like that, it probably

On the subject of things that aren’t real people, why doesn’t Google’s voice have

Amazon, Apple, and Microsoft all named their digital assistants, so why not Google?

Don’t feel bad, Google voice, I think you’re fantastic.

Tell me in the comments if you prefer Siri, Alexa, Google, or Cortana when you need to

look something up, but don’t want to get your hands dirty on that filthy keyboard.

Number Four: Sparse Profiles Do you remember the first time you logged

They wanted you to set a profile picture, a background picture, enter a birthday, job,

school, hometown, relationship status, and plenty of other things.

While not everyone fills out every single field, most people will at least cover the

Just enough for people to get an idea of who you are beyond the pictures of cats, eggs,

Since humans are big on oversharing, you can probably guess that bots aren’t that into

Most websites that allow users to leave public comments require them to register an account.

Not every site expects a full bio, and not every person is willing to give it, but if

the account is nothing more than a name and placeholder profile pic than you might have

While this might not be conclusive evidence, it’s a pretty big red flag.

If you notice a whole bunch of these barren accounts, that's a much bigger red flag.

In that case, you don’t just have one bot, you have an infestation.

You see, the automatic responses bots give out tend to be chock full of keywords.

Those keywords attract other bots, whose responses attract more bots.

This can lead to a feedback loop of automatic responses to automatic responses.

Even when things don’t escalate that far, bots still love to follow each other around.

Particularly nefarious programmers will go as far as creating bots specifically to upvote

字幕列表影片播放

判斷你是否在和機器人哈拉的5種方法 (5 Ways to Tell If You're Chatting with a Bot)

bunch

sophisticated

context

tend