For every line the machine may weave,
A second look you must conceive.
A twofold check, a watchful eye,
Lest error slip and truth run dry.
Yet prompt it well, and time you’ll save,
A guiding hand, both sharp and brave.
But haste beware—heed well this score:
To trust too much may cost you more!
What is Inkster’s Law?
Per Brian Inkster’s The Time Blawg on Inkster’s Law, Inkster’s Law is a term coined on LinkedIn by Gav Ward.
It all started with a LinkedIn post by Dan Hunter. Dan starts the post with:
Ok, here is my first pass at full syllabus for a comprehensive Legal GenAI subject…
(What have I missed? Will post explanations, teaching notes, and references for each seminar shortly.)
He then goes on to give us details of 11 seminars on Legal Gen AI. Two of these cover prompt engineering. I couldn’t resist telling Dan what he had missed:
If you are giving two full seminars to Prompt Engineering you need at least four to being a Legal Hallucinatory Detectorist.
Gav Ward then chimed in with:
The 2:1 ratio of time needed to check for accuracy of AI generated material – Inkster’s Law?
To which I retorted:
I hope a Suss Jnr is quoting Inkster’s Law in years to come
So, there we have it, Inkster’s Law: The 2:1 ratio of time needed to check for accuracy of AI generated material.
Asking AI What Inkster’s Law Is…
And, if we ask AI what Inkster’s Law is at Feb 2025, over a year since it was coined we hear:-
Inkster’s Law, as discussed between you and Brian Inkster, CEO of Inksters Solicitors, refers to the “2:1 ratio of time needed to check for accuracy of AI-generated material”, except where prompts are effectively tailored based on awareness of standard AI outputs.
This principle highlights the importance of quality control and human oversight in AI-generated content, acknowledging that reviewing and refining AI-produced material can take twice as long as generating it in the first place—unless the input prompts are carefully optimised.
Inkster’s Law aligns with broader discussions about AI accuracy, legal technology, and content validation, particularly in professional and legal contexts.
