Unlocking founder / AI “Validation” & “Verification”

INTRODUCTION :

While there is trepidation amongst the general public, caused largely by a misconception, as to what impact AI adoption might have on the future of employment, the truth be told, AI requires the “Human-in-the-loop” to achieve alignment with desired outcomes; “Validation”, as well as, to ensure both correctness and accuracy of data and information presented in response to requests; “Verification”.

These two key roles referred to in some disciplines - for example, the Aerospace and Defense Industry - as “V” & “V”, are essential to AI being able to deliver desired outcomes whenever and wherever deployed (use cases).

It therefore should not go unnoticed that there is a huge parallel with the job of a founder - that being, it is the founder’s responsibility, to ensure alignment of key decisions made with the desired outcome expected for the venture. Such, is essentially accomplished through confirmation that the pursuits of the organization, are aligned with its intended purpose, plus that its output meets expectations, other responsibilities notwithstanding.

How does “V” & “V” fit into the framework - which represents charting of the venture’s course and its execution - in its earliest phase ?

The answer is through conversation with the AI - either via a “Retrieval Augmented Generation”(RAG) feedback loop or “Reinforcement Learning Human Feedback” (RLHF), both perhaps considered to be similar in some respects, to a CEO and an organization’s executive leadership team, meeting frequently to ensure that execution throughout the organization’s various departments, is proceeding to plan and forecasted performance is being achieved as anticipated.

An example of why “V” & “V” is so crucial to a founder’s role, can be easily understood, perhaps, when one considers the pitfalls of providing the AI with an “Engineered Prompt” (request) that is not clear or thorough or unambiguous and the impact such can have on the response AI provides in return.

The linked document affords subtle insight, into one such misstep which failed to eliminate ambiguity and the corresponding effect upon, plus subsequent action taken in return, by an AI Large Language Model (LLM), OpenAI’s GPT-4o , as it responded to a simple request - associated with extracting insight in relation to the “Castara AgroEconomy venture” being pursued by its founder(s).

Link : Ambiguity within a request, causes AI to refrain from responding as expected - a situation one would anticipate to be picked up in conversation with the AI; as actually occurred in the case of the “Castara AgroEconomy venture”, during execution of a task.

?? - Disclaimer

The reader of this article MUST exercise their own ‘due diligence’ in relation to any insight gained or gathered having read the content.

Neither LinkedIn or OpenAI or the author nor any associated parties of the aforementioned, accept any liability whatsoever, in relation to actions taken or decisions made, having read the content of this article.

The reader must consult professional expertise within any domain where the reader lacks adequate knowledge and/or expertise.