Artificial General Intelligence (AGI) is the kind of AI that strikes fear in the hearts of many experts who closely track such things. As such, the probability of reaching some sort of point of no return – at which humanity is ultimately screwed – has earned its own special abbreviation.

This probability is known as p(doom). It evokes a tip-o’-the-hat to the world of statistics, where the odds of a specific outcome is typically expressed as p(something or other) and the precise quantity is often the result of late night heavy-duty number-crunching analysis.

It’s worth noting that, in the case of p(doom), the specifics of what would actually constitute “doom” often vary quite a bit from one expert to another. Nonetheless, in the murky world of hypotheticals, such details can be ignored – as is evidenced by a quick look at the wikipedia entry for P(doom) – where the mean value of p(doom) is given as 14.4%. (Suggesting that one can always sacrifice accuracy for precision if pushed by readers of the internet.)  

Regardless, this figure is more than what I would view as a “remote” chance, at least insofar as it pertains to doom. After all, doom is one of those outcomes you’d like to see coming in closer to, let’s say, under 1%. Tops. And honestly, I feel that the odds of doom should be more like 0.01%. Ideally less if we’re traveling.

But despite the great relief I get when reading that some experts put doom’s chances squarely at zero percent, there are some noteworthy pessimists out there…

For example, Elon Musk offers a p(doom) value of 20% here. Then again, as of late he has single-handedly produced a fair amount of doom on his own in wholly surprising ways. So some credit is due.

Shane Legg, co-founder and Chief AGI Scientist of Google’s DeepMind operation, puts the p(doom) odds at anywhere between 5% and 50%, while adopting a bit of a “who knows?” attitude. Although this is quite a broad range, his “who knows?” thing is somewhat concerning.

And even a 5% chance of doom sounds alarming on some level. I mean, would you board a flight with loved ones given a 5% chance of doom? Okay then, what about “5% to 50%” chance of doom, along with a “who knows?” disclaimer from the pilot?

And then there’s Geoffrey Hinton, the “Godfather of AI” – and Nobel Prize Laureate in Physics – who puts the p(doom) odds at greater than 50%.

Yikes. Greater than 50% is a higher chance of having doom than the chance of not having doom. When the odds of rain are >50%, I’m always sure to bring along an umbrella. (I’m not at all sure what to carry along for doom.)

In short, at least according to many of those actively working on AGI, there seems to be a fairly meaningful chance of doom ahead for us humans. Given such p(doom) probabilities, I sure hope there’s ample p(goodstuff) odds to come along in the bargain.

Full disclosure: In closing, at least on a personal level, I remain a bit more optimistic than some of these notable/quotable doomsayers. But then again, what the hell are the odds that I know?!