Snoop and Trump — Why Our Coming AI Overlords Are Likely Doomed
But don’t get triggered, the rest of us will be just fine… probably…
Humans are unpredictable. Society is complicated. Humans are more complicated. Society is more unpredictable. Life is a strange feedback loop. Reality TV puts an exclamation mark on it. E! News exploits all of it and adds another exclamation mark… perhaps for irony? But what does this have to do with Artificial Intelligence?
Here is a headline from Eonline today. A bit interesting to say the least. How does a computer go about building context on this? The site also features a flashing image of the two celebrities. The story is a report on an internet meme? It offers that slowing down Adele’s song “Hello” reveals the voice of Sam Smith. A computer could score that… but why? Assuming it took the time to dig through the title, process the article, identify the concept, and then analyze the songs (at varying speeds) — all of which makes sense at some superficial level — why would it?
This claim/concept is making headlines because it interests people. The statistical accuracy of the claim is hardly important. Consider Fake News. Now realize that plenty of people aren’t fooled by it, but they spread it around anyway. Why? Because they have some interest in generating confusion. Is it political? Is it entertainment? Is it E! Entertainment? Humans are strange. Society is stranger.
Can we be honest? There is no truth about Fake News. Truth requires levels of context and verification that defy our current abilities to comprehend. Facebook and Google are going to use AI to root out Fake News. Hurray! Wait… how? Please identify for me the dozens of Hyperscale Data Centers and Tensor Flow algorithms that are going to ferret through a near endless array of information and data lineage to determine the veracity of millions of Fake News headlines generated each week. I’ll wait…
Better yet. Can someone show me the ROI analysis on that? I’ll wait a little longer… while I do the electric bill at these fictional data centers (they are real, they just aren’t working on any of this) is ticking into the millions. Aquifers are being drained either to cool the centers themselves or the nuclear power plants needed to support them. Does that seem like a good use of resources?
Why are our tech giants even pretending? Because AI is cool! But also obviously really stupid, too… or at least the folks pretending they would fund this.
Meanwhile, humans don’t care… most of us anyway. Fake News is a label. When our interests move on, we will entertain ourselves with something else. The folks in politics will gin that new thing up to their own benefit and, if it suits them, tech giants will pretend the new thing can be fixed with AI, too. Not that anyone will or could even try…
Okay, perhaps E! and Fake News aren’t that important. What about Russia and the US President? Thankfully, neither is half as important as anyone pretends. Free markets and constitutions have seen to that — sorry, I know that triggers some of you… calm down.
Image an AI trying to interpret Trump’s tweets? I think you can justify a return. His tweets move global markets — up, down, sideways, and in circles. It is all quite Big League… or is it Biggly? We all know — Trump tweets can be huge!
Yet it has nothing to do with what is in them and everything to do with how they are interpreted by other connectors in the giant network that is first social media, next the internet, and finally — society in general. To do this work right, AI will need to model how each of those connectors will react to his words. Even when they are typos.
Imagine the processing chaos of “covfefe”! Clearly a typo of “coverage”. It doesn’t take and AI to figure that out. Though an AI easily could. But what would it make of what society did with that term? So many connectors pretended it could have meant dozens of other things. Why? Because society is silly… well, people are. Society sometimes seems doubly so.
Imagine the number of Hyperscale Data Centers we would need to build just to interpret Snoop Dogg. Does he even know what he is saying? Honestly, I don’t think he cares. Why does Snoop’s nonsense make him endearing? Why does Trump’s? Why do so many of you take issue with one of those two statements but not the other?
People are silly. We like it that way. It keeps things entertaining. Society, doubly so. Our economy, our governments — at least those that last — have evolved to ignore or suppress all of it. Will our AI overlords do so as well? Likely…
You see artificial intelligence is no where near “the singularity” of Kurtzweil. Computers aren’t synthesizing… not well. They don’t even analyze with real context, yet. AI is about optimization. It is about doing things that humans already figured out — really, really well — and doing them better, faster, and more efficiently.
So perhaps only accountants should fear our coming AI overlords? Or perhaps not? I mean, have you seen the size of the tax code?
One can imagine a computer like Deep Thought pouring over the US tax code for a decade only to come to the same conclusion as the Artificial Intelligence in War Games… Would that be a breakthrough? Or should we all just stick to Tic-Tac-Toe? Perhaps that is a problem for AI too? But I doubt it… Thanks for reading!
