Purpose: Difference between revisions

3,297 bytes added ,  3 June 2023
no edit summary
No edit summary
No edit summary
Line 46: Line 46:


Now if we contrive to  take a pace back from the canvas and look at it through the lens of the disembodied entity who pays all the wages, some agendas appear more important than others. The ultimate clients care a lot about creating a relationship of trust and reciprocity in which mutually beneficial commerce can flourish; they care somewhat about safety mechanisms should that relationship break down, but they will put these behind maintenance of that healthy relationship in the first place, and they will care little about the ongoing livelihood of the professional advisers who claim to help them achieve it. If they could achieve it without agents, they assuredly would.
Now if we contrive to  take a pace back from the canvas and look at it through the lens of the disembodied entity who pays all the wages, some agendas appear more important than others. The ultimate clients care a lot about creating a relationship of trust and reciprocity in which mutually beneficial commerce can flourish; they care somewhat about safety mechanisms should that relationship break down, but they will put these behind maintenance of that healthy relationship in the first place, and they will care little about the ongoing livelihood of the professional advisers who claim to help them achieve it. If they could achieve it without agents, they assuredly would.
===Case history the NDA===
Let us apply this “[[behavioural economics]]” lens to a workaday [[contract]] that draws constant gripes from practitioners and constant efforts from [[Legaltech entrepreneur|legaltechbro]]<nowiki/>s and disruptors to solve the legal problem it poses: the [[non-disclosure agreement]].


{{Sa}}
Our proposition is this: the impulse towards [[confidentiality]] at the start of a commercial relationship is universal. The overarching ''legal'' purpose of an NDA is well understood, yet the [[NDA]] has resisted standardisation and automation.
 
This has not been for want of trying: any number of nascent [[legaltech]] providers<ref>Our [[legaltech roll of honour]], as ever, refers.</ref> have brought serviceable AI review tools to market, and while they handle many NDAs, the reality is they do not save much time or money, let alone move the dial on legal risk. Oh, they will argue differently, but come on.
 
Similarly, excellent standardisation and centralisation initiatives such as [[OneNDA]] have localised take-up but have not revolutionised, nor really made a dent in, the wider NDA industry.
 
The mystery of why this is so disappears when you look at it from a ''purpose'' point of view. The point of an NDA, notwithstanding formalistic legal theory, is ''not'' to “manage legal risk to proprietary information”.
 
Rather, an NDA is a ''commitment signal''. It shows you are serious and you care enough to make an effort. It also speaks to exactly how special you consider your sauce to be, and therefore ''how much it is worth''. Its role is ostentatiously to display your preparedness to invest time and resources in beginning a business relationship and sharing a precious secret. It is a bit like sending expensive embossed wedding invitations rather than a to-all email.
 
An NDA is largely ''performative''.
 
That being the case, an important part of the NDA process is ''incurring expense to put it in place''.  Engaging expensive humans to perform a ritualised “negotiation” of no real legal moment. You must haggle, a little bit. A “[[for the avoidance of doubt]]” here, a “[[without limitation]]” there, some undertaking to inform the discloser in this circumstance or that.
 
That being the case, no [[legaltech]] solution that ''transparently'' makes the process easier, cheaper, quicker or less bothersome fulfils that primary function. A ritual that can be completed in a trice loses its essential force. It becomes meaningless.
 
Therefore the only role for AI in the process can be behind the scenes: the interface between parties must still be human to human, and must carry the gravitas of grit. It must be somewhat ''painful'' (Think embossed invitations).
 
Therefore (since there isn’t really any ''legal'' purpose to an NDA “negotiation”) the AI (or any other “cheapifying” strategy) cannot afford to make any part of the process more efficient. What it must do is make it ''seem'' more efficient to the chief operating officer inside the organisation without actually ''being'' more efficient to the counterparty. Its only application can be to further complicate a formal process.
 
And this is exactly what we are seeing, l. We should expect this to be the basic template for any AI deployment in a commercial negotiation{{Sa}}


* [[Agency problem]]
* [[Agency problem]]
* [[Design principles]]
* [[Design principles]]
{{ref}}
{{ref}}