Of course we don't know what the AI algorithms do, that's the point of using them

We've yet another instance of the engineers' fallacy concerning the economy. Which is that we can understand it as a detailed level and therefore guide and even plan it. This is, as we all recognise, that Socialist Calculation debate all over again, the answer to which is that we can't, as in Hayek's The Pretence of Knowledge.

Here the fallacy comes from this insistence that we've got to know what all those algorithms are doing. For if we don't then we don't know what they're doing, do we? The answer to which is yes, quite, that's the point.

“In some ways we’ve lost agency. When programs pass into code and code passes into algorithms and then algorithms start to create new algorithms, it gets farther and farther from human agency. Software is released into a code universe which no one can fully understand.”

Entirely so, that's why we're using the algos. That real world out there is complex. Too much so for us to fully understand in detail. That's why planning doesn't work - the centre can never gain enough information in useful time to be able to process it. Which is exactly why we're using the algos, isn't it, just as we use markets to guide the economy.

The insistence that we've got to examine each and every algorithm, in order to see what it's doing, so that we can direct it, is exactly that socialist - or here that desire of engineers to, well, engineer everything - planning fallacy. We'll not be able to do it, doing it would be contrary to the very reason we're using the software in the first place.

Rather better for us to get on with what we do in that economy therefore. For the idea that we'll ultimately know it all is a delusion.