Selling The AI Apocalypse

It’s funny how all the AI developers have come out recently to tell us that AI is going to kill us.  I think the time has come to start asking some very fundamental questions about this whole deal.  Let’s start with what should be a simple one – Why is all this AI development happening in the business sphere and not the academic one?  Think about it – these are all Silicon Valley start-ups doing this work, not the super computing lab at ORNL or some university lab.  Now ask yourself the follow-up – What is their business?  What are they selling?

Short answer – they are not selling anything, they are raising capital.  That’s the business they are in in – to raise money, very large piles of it I might add.  In Silicon Valley, raising capital is all about buzz…enthusiasm for the idea.  That means publicity, and lots of it.  Now, I cast no aspersions the way of AI developers, but the capital raising game has gotten out of hand before.  There is a fine line between raising capital for a risky new tech venture and selling snake oil.  I frankly think the idea of needing more money to make sure the AI does not kill us is a good way to keep the money flowing.

After the “Hugging Face” incident, I wrote a piece of just what did and did not happen.  I followed that up with a discussion of human fear and in it, I laid out some stats worthy of repeating:

The best AI’s require multiple data centers networked to achieve the necessary computing power, but single campus centers are currently in the works.  Such a single campus situation would cover 1000 acres, give or take, and consume roughly 14 billion (14,000,000,000) calories of energy per minute.  Now, compare that to an actual human brain which is about the size of your two fists put together and the entire human body consumes about 2 thousand (2,000) calories of energy PER DAY.

AI is a brutish sort of thing.  The science behind what is happening now is decades old – heck, I fiddled with it in the 1980’s.  These recent developments are not because of new insights into computers or the human mind.  These recent developments are simply because the price and size of hardware has been reduced to the point that sufficient computing cycles can be put together.

By analogy – computers have been beating chess grandmasters for decades now – but they are not smart.  A chess grandmaster can visualize the board and see things going on – deduce the opponents strategy.  A computer simply plays out countless future scenarios and chooses the one with the best probability of winning.  As soon as computers got big enough and fast enough to run sufficient future scenarios, they started beating the grandmasters.  The grandmasters are smart and elegant – the computers are brutish at best.

Now, since AI is really old news just now happening because computers keep getting cheaper and smaller, better AIs are more about getting more computing cycles than anything else.  And how do you get more computing cycles? – money and more money.

Therefore, every time an AI developer is on your TV telling you something, you need to remember the business they are in – raising money.  This latest round of fearmongering sounds to me suspiciously like they are trying to tap a very deep pocket – the United States government.  Remember, they not only need capital dollars but as the data center pushback continues, they need good will which means they have to get political.

I have been cynical about AI from the beginning of the buzz.  Not because nothing is happening, but because what is happening is nothing like what people are talking about.  This is not about computers modeling or mimicking the human mind.  That would be true artificial intelligence.  This is about massive amounts of data, computing cycles and a lot of probability and statistics.  This is not intelligence, this is math.  They slapped the “AI” label on it to begin with because it was sexy and helped them raise money – and now they are stuck with it.  Which is a problem.

The label is running out of steam and continuing to use it is going to get out of hand.  What’s happening is good and maybe even necessary given what is happening in the rest of the world.  But it is time to level with people about what it is and what it is not – most especially the capital markets.