Is Your AI Lying to You?

AI lying
One young boy businessman conducts a job interview with another boy using a lie detector machine. One boy is apparently not telling the truth. You can never get too much practice for a job interview.

I talk a lot about artificial intelligence (AI) and how we are using it more and more in our homes. We use it because it is convenient, and if saves us time. We also use it because it is cool and it makes us feel like we’re cool when we use it.

I have also talked a lot about the security of these AI devices that we use. We talk about whether or not these devices are safe to have in your homes, in reference to security and hackers. Is convenience so important that we are willing to sacrifice our security? Well, that’s a question for a different article.

This article will focus on another pressing question: is your AI telling you the truth, or can it lie? Do you even care? Well, either way, I’m still going to talk about it.

So, here’s the inspiration for this article: some woman videotaped herself asking her Amazon Echo if it was connected to the CIA. According to that video, the Echo did not reply to the question. That’s a weird response, since the Echo’s brain, Alexa, will usually always reply. If she doesn’t know the answer, she’ll tell you so.

Even so, whether or not this video is legit, it raised a lot of questions about AI and its ability to lie.

The short answer is: no. Your AI, including those in the Amazon Echo and that in Google Home – and any other AI that is available out there – is using a set of responses that are pre-programmed. As cool as this AI is, and as fun as it is to believe that it is really intelligent, the truth is that it isn’t (yet…).

What’s really happening is that the brain in the devices is learning but not understanding. It is working as a system that receives and routes your requests. There just isn’t a lot of actual thinking that goes on. It’s really just an upgraded version of typing in a request on Google.

So, your AI isn’t lying so much as it’s tricking you. With good intentions, of course. Today’s AI is pre-programmed to be able to answer a set of questions that it is approved to answer. It doesn’t really lie – but it could since it is programmed by humans. So, it wouldn’t so much be that the AI is lying, as it is programmed to lie (this is starting to sound like the plot of a Sci Fi movie).

The AI really just processes what you said and seeks an answer in its database. So, it could only lie if the database was comprised. And who is going to do that? Unless you’re in a movie.

So, should you be afraid that your AI is lying to you and somehow a scheme by the CIA to get into your homes to spy on you? Not so much. Should you be paying attention, asking questions, and just generally staying alert as AI becomes more popular and mainstream? Well, it couldn’t hurt.