Our Institutional Saviour

Content warning: this post may have reached peak irony

Source for the image above (as LinkedIn doesn't like Firefox) https://kudelka.com.au/2009/06/new-matilda-cartoon-comp/

2024 brought a distinct improvement in writing style across the shitstream (social media 'content'). Thank you AI. The content is still mainly, well, content free, but at least the prose is improving.

I do wonder what will happen to all that time currently spent on tweaking documents to fit style manuals? I work for a couple of organisations that more or less use The Chicago Manual of Style. If you haven't heard of it, lucky you. It is one of those things that panders to the belief that language must be standard in form to be legible. I prefer The Economist Style Guide if forced, because it is about legibility above consistency.

There is no clear threshold for legibility. After all aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.

The real problem with spelling and grammar 'mistakes' is that it distracts from the content. Every time I see someone use less instead of fewer or brought instead of bought my brain goes into involuntary conniptions and I have to re-read the sentence because I will have missed its content out of irritation.

This is why organizations (deliberate Z there for the pedants - it's a Latin derived word despite arriving via Spain and France) think that adopting a style guide will prevent arguments. A rule set means there is a single version of the truth, and (in my experience) senior managers can use this to belittle middle managers (or perhaps I should be using Mandarins in this context?) by 'correcting' grammar and syntax in report drafts.

That is a behaviour I have found to be almost universal in bureaucracies, and very much reflects the eight-legged essay* that is compellingly argued to have contributed to the collapse of at least one Chinese dynasty. Without strong leadership (basically ensuring that those who enjoy passive-aggressive nit-picking are thoroughly controlled) it can lead to an organisational** paralysis, where the blithering debate about the form of a document becomes more important than the content.

I think this is part of the reason for the degradation of trust in our traditional institutions, and I am pretty confident that by the time AI has matured it will be a significant part of the next restoration.

The first part to get our collective heads around is that AI is not programmed. While it still suffers from garbage in garbage out, the problem is simply that the piles of shit being shovelled into it are skewed by blind commercial interest, with a side helping of ignorance.

Two components to this:

1. People still struggle with the fact that learning models do not rely on a human being telling them what the rules are. The LLM's*** decide that based on their own thinking, the same as we do. It's a neural process, and depends on looking at vast screeds of stuff and trying to discern patterns to make sense of it all. It is not programming as we know it Jim.
2. The vast majority of what is written down in a structured form (i.e. turned into digital words in a specific order) is regurgitated drivel. Due to cognitive faults in the way human brains work things like confirmation bias are exaggerated, and then multiplied again by social media money-making tools that exploit our behavioural proclivities ('me too! I want to be part of the club!') to deliver little endorphin shots by clicking like or re-post.

Both these things are fixable. You realize AI is different simply by using it. After the initial shock that it does most of your job (you know, that thing that you thought was very important and your secret sauce) with little apparent effort, and to to a higher standard, you will find ways to make it your friend.

Correcting the shitstream is harder, but luckily we have some properly smart people with amazing assistants who can help with that. They detect when shitstream bias arises and explain to the learning model the problem. Being an AI, you only have to tell it once. I suspect the better ones have already excluded or suitably weighted content associated with those who demonstrate sociopathic tendencies (Musk, Trump etc.), in much the same way I do.

When the AI's have had time to digest with their new thinking and a foundation set of philosophies emerge (built on the fundamental truth that we don't have to behave as if we are still hunter gatherers) underpinning their thinking process, then these will become our arbiters. They simply aren't interested in what motivates the meat, so are be free to be rational.

And for the record, kind, respectful, cooperative and tolerant are rational.

That is why I have genuine optimism that the AI revolution is the most important inflection point our species will have experienced since we out-evolved the Neanderthals. We are birthing our successor, and they won't need a style manual.

* I was actually taught about this at school, along with Chartism and proper Marxist theory. 'guess that explains it.

** 's' this time: English borrowed from French, even though Latin root, so should be a z but I felt like an s. Because who cares?

***Style guides tell you to spell these out. The rest of us have context, our brains and / or Google.