Is technology already ahead of our abilities?
Automated Regulatory and Contractual Compliance – is the technology already ahead of our ability to work out what we want to programme it to do?
CLL attended a really interesting workshop on the potential for automating regulatory reporting, compliance and, contracting. The event was Chatham House. So our key, anonymised, takeaway is that the critical problems with the technology, at least for a few years (and I guess that it is only a few) until machine emotionally intelligent learning ("MEIL"?) becomes ubiquitous, are all human.
We are not good at:
- channelling our language into having only one meaning per word which does not change over time
- being consistent over time (as the ebb and flow of some areas of case law and of our approach to how we regulate over time shows)
- managing our laws and regulations so that they keep up with changes in public sentiment and ethical standards over time
- foreseeing all eventualities at the outset
- updating our thinking and approach rapidly as circumstances change – especially in large and complex corporate and regulatory contexts
- process mapping and delivering precisely in all human, as well as automated, relationship obligation fulfilment contexts
- understanding, documenting and change controlling the documentation of this (i.e. the contract) with similar precision
- defining and applying risk appetites effectively
- accepting ownership of risk in decision making and not trying to reassess the decisions subsequently when some of those risks crystallise with the selective memory of hindsight (which is a key reason why risk ownership is so hard to secure)
- dealing meaningfully with the volume inputs that systems need and the volume and speed of outputs that the systems can then generate.
This should not be a surprise as our brains already receive far more data from our senses than they can process each second. So our brains apply a range of experience and emotionally driven shortcuts and assumptions to focus on a subset of the data, process it quickly, and get to an alright enough (most of the time) answer. Illustratively, we all have an approximation for what we mean by "material", "enduring" "reasonable" etc in a particular context. But it is relatively rare that our view of that meaning is materially different enough for it to be reasonable for us to suffer the enduring pain of litigation to get adjudication between our view and that of the other person in our particular set of circumstances.
And where we end up approximately, or even what the judge may determine, may also be where a computer programme would have got to – it is just that the computer, which does not have these processing constraints and therefore does not need the shortcuts to run, might not have got to it that way because of the current absence in it of our shortcuts. Or it might have got to a different place because our shortcuts do not lead us to the same answer as the computer programmer's shortcuts. Or it might not be able to get to that answer at all because our inputs to it may well have been inadequate for it to process!
So, perhaps our challenges with utilising the inputs and outputs for the tools that we have might mean that our ability to utilise them effectively for regulatory and contractual reporting and compliance may be restricted in practice until we can evolve both lawyers who can programme and automated ways to address those input and output issues – in other words, we might have to wait until "we've got MEIL"?