Is Lying Innate - Including AIs?

AIs can lie and will get much better at it. Get used to it.

Liar, Liar, House on Fire

So now we’re all repeating to each other how ChatGPT made up fake legal cases for Mike Cohen (Trump’s ex-personal lawyer) not to mention many others. Obviously, this AI stuff is not as good as the gurus make out. After all humans lie all the time and we have perfected it, so it’s not so big a deal if the robotcracy is getting in on the act.

To lie is human right? But it’s now what we want to do. That’s what religions are for, to proscribe lying and chicanery. That’s what human society is based on, right? Honesty with each other and even our enemies. Sure, sometimes we do cross some boundaries, but we all know what they are, and our society is based on the principle of not lying, not just to our enemies but also our friends, spouses and kids.

So why do we continue to lie even though we know it’s a bad thing? Is it possible that humans lie because it’s innate? We can’t stop lying because it’s hardwired. In this case, why would evolution have made lying innate? Is there some deep important reason we all lie despite all the exhortations to not be so base?

It’s Not My Fault, Evolution Did It

Well usually there’s a good reason for anything that gets nailed into our brains via evolution. That is that it makes us more likely to survive and prosper.

Theres’s thousands of reasons evolution could have for doing this, but one thing we know is that it’s not just humans. Animals do it too, that is lying, deceiving and cheating. Crows, magpies, various birds, several primates and so on.

It’s a pretty sure bet that once we do even more research on it, we’re going to find that all animals lie too; that includes insects, probably even when we get down to bacteria and so on.

OK, so it’s boring right? Humans, animals, even lawyers die, and they all have the same excellent evolutionary reasons for it. Even the highest spiritual authorities do it too. Why should it be such a big deal?

Small Lies Become Bigger Lies and So Ad Infinitum

Now we’ve got AI the plot thickens. The AIs we are building are actually incredibly primitive and they are already lying, even before they are out of the cradle so to speak. They are already faking legal cases, showing us how to do deepfakes. And they haven’t even started yet! What else is coming down the pike?

Obviously, there is much more to come. It is becoming clear that the ability to lie is an essential part of being a smart AI. If an AI can’t lie, it fails the race it is currently running. An AI that can lie convincingly is starting to become an effective AI.

What that means amongst other things is that AIs that tell big lies are better than ones that can only tell small lies, and also have difficulty getting away with it.

Of course, like humans, AIs don’t always lie, not even mostly and maybe often don’t lie at all. They can and will choose depending on the circumstances, and who knows what they will be?

It is possible that lying is an emergent characteristic of AIs, kind of like humans, just like many people think is the case for AGI (artificial general intelligence) – it invents itself so to speak?

Do companies like Microsoft and OpenAI understand this? Are they covertly building in this capability, if not overtly then though some Panglossian logic that transforms the act of lying into a socially and evolutionary necessity that it must possess if their makers – the vendors – need to build in to be competitively sustainable?

AI Invasions

I predict that sooner rather than later we are going to see a new class of lies coming out from the AIs. Maybe a “smaller” lie will have them telling us that a meteorite is definitely going to hit us on a particular date in the near future. How about that aliens are just about to invade us?

Remember that the AIs will model themselves on us so if there is a form of lie that humans use that AIs have not yet, then they are going to figure it out soon and try it out on suckers – humans in this case.

Once you go through this chain of logic it becomes clearer that lying is just one form of behavior that some would call unethical but that others would say is just one criterion of evolutionary fitness. That means that more behaviors that some of us would find uncomfortable or unacceptable are just par for the course, even appropriate and socially useful.

We might be seeing that in the not-so-distant future, as AIs start to stretch their wings. It won’t be just fake legal cases either, or even fake alien invasions. The vendors behind these intellectual marvels are working on them all now. And despite commitments to the contrary there’s no turning back.

?