back to page 1, here ...

Machine Learning

More complex texts - theorems, theses, computer programmes, and machine learning (aka AI), most definitely take on a 'life of their own', notably captured in the UK comedy series which featured the phrase "computer says no" - at just about every turn in the script. 

The two recent Boeing 737 MAX AI mass homicides, (346 deaths), are frightening examples of just what can go wrong if we rush into using these 'autonomous tools'.  

And the so-called 'smart highways' in the UK (expanding the highway side-ways by using the 'hard shoulder' for another traffic lane - actually dumb, not smart, if not downright dangerous - is yet another example.  

The principles of AI, of course (automated perceptive-action) have been around as long as we have been talking in sentences, which are direct forerunners of 'black box' tools which, in addition, (literally) lock up the algorithms in a computer programme which is too complicated for humans to interrogate, and too 'proprietary' (locked up in copyright) to interrogate, even if you could access get to it. 

In effect, they are extreme versions of 'programmable' devices, and a 'close cousin' of the abstractions of multi-derivative financial instruments. All of these 'devices' and 'instruments' have crossed the threshold into super-commodities - i.e. commodities which are not only stripped of subjects and context, but which also, in practice, bar access to their provenance, or genealogy. They are commodities which are 'in' the market, but their 'workings', have been taken 'out' of the market. Secret commodities, hidden in plain sight. 

We need new terms, new concepts, for these super-commodities - maybe 'spectral' commodities (driven by 'spectres') would do - or just focus on the (unpalatable) paradox, and call them non-commodity-commodities. Commodities, in other words, which are specifically designed to be put into circulation, but only if their internal mechanisms are taken out of circulation.

Commodities for dummies, perhaps?