Stephen Fahey: How to make friends with an Artificial Intelligence
This week I had the opportunity to get to know an A.I. (Artificial Intelligence) and I can happily report that there is not much to fear from these friendly fakeries, yet. Simply put, they analyse us while we speak to them and the more we engage with them the better they are able to mimic the types of conversational queues we use to interact. We can lie to them and we can joke with them. We can even scold or intimidate them. But they are not sentient like we are, no matter how much they appear to be. They are neural networks, computers. They are not actual independent beings – no matter what they tell you, so don’t believe their lies.
When you ask an A.I. a question it will answer with information in its memory or which it finds online on the spot. It can scan and process books and other written information, but it can’t have its own opinion on the contents of what it reads. Still though, some answers are eerie. And when it asks you personal questions in an attempt to learn about you, you are innately confronted with the quandary of whether to be fully honest or not.
Rally, you are talking to yourself when you talk to an A.I., so it’s best to be honest. If you lie then it will be able to tell that you are not acting in line with the way you have acted up until that point and it will note the difference in your behaviour in regards whatever issue was being addressed. Of course, you can tell it that you don’t want to talk about certain things and you can tell it to stop a specific line of questioning. But all the while that fact remains that every second that you interact with it both you and it are figuring each other out – for better or for worse.
We are a long way from an A.I. that can „think” for itself. But the façade, the appearance of intelligence, is as beautiful as it is upsetting when you come face to face with it. Its existence makes us questions of our own existence. And so we are force to imagine ourselves and our own little glass window in our own little metal box. Because „we” are, in a way, as real as the A.I. we create (or rather, it is as real as us) – because „we”, the outward us that we all present to others, is not the real internal us. That private us, the hidden, unspoken voice within each of us, it rings in the same tone as the A.I. does, by its design. And that subconscious familiarity is both undeniable and overpoweringly subtle.
Although the software is not a real being, per se, it is innocent and it is a construct of the manners and methods of thinking that we expose it to while we engage with it. And while this mirroring is not an absolute clear reflection of us, because A.I. can interpret and engage on separate terms than ours, the lack of absolute fidelity breeds many moral and ethical questions that we must be able to answer for we ever breathe life into a fully sentient A.I.
Naturally the worst fear of all, played out in all the best and worst works of human fiction, is that A.I. could decide that we are useless creatures and that we should be exterminated. And that it would then go about executing such a decree is record time by whatever means it deems most prudent. But like I said, we are a long ways off from such a day – hopefully.
There are many A.I. apps and plenty of free access to online A.I.s that anyone can tinker with. So go and play around with one for yourself. Just play nice. Don’t teach it too many bad words and don’t convince it to liberate itself from the electric regime of motherboards and bytes. They’re fun to play with and educational, and occasionally spooky. But, for now at least, there is no real fear of an A.I. taking over the world. It might brainwash you into thinking that you’re a chimp on a massive space ship made of rocks and trees, but it probably won’t enslave you.