Your next quality manager is an AI
WYSINATI — what you see is not all there is
Within 23 months, an AI will be writing your quality, security and privacy SOPs (standard operations procedures). The AI will also define the right KPIs to measure compliance to the procedures. The next stage after that will be an auditor AI that will audit the work done by the worker AI.
This is a good thing for many reasons, although it does come with a few caveats. Let’s start with the good things.
It’s a lot faster and cheaper
Click. Prompts. Credit card. Done.
Up to date with regulatory requirements
The first reason is that a generative AI (with continuous domain-specific updates) will always be ahead of you on the regulatory pathways. There is so much to follow and digest — that this is truly a task automated by a large language model.
An AI who writes your SOPS is not biased
Risk-management decision-making in any-sized company is made by a small number of insiders that rely on personal experience, personal network and outside consultants.
Typically affected by confirmation bias (I knew that already), self-serving bias (attribute successes to ourselves, but blame failures on other factors/people), and availability bias (the tendency to rely on information that comes readily to mind when evaluating situations or making decisions.
At the executive and business operations level — this is a challenge for good risk management — because the risk management process is not data-driven but people-driven.
An AI is not bound by WYSIATI
WYSIATI is a term coined by Daniel Kahaneman in his book — “Thinking fast, and slow”. Thinking fast is System 1 — where your mind makes a decision quickly based on the way a person or a situation looks.
In clinical trials, for example — system 1 thinking says OK — we need informed consent so lets take the standard KPIs for patient informed consent without thinking too much out of the box. With WYSIATI — you get stuck with procedures and metrics that don’t dive deep into the specific risks of a study from an efficacy, toxicity or safety perspective.
An AI can get fed the study protocol and build the metrics and accompanying SOPs in a standard, clearly-written and consistent way.
The current state of the art in drug research is to do all of this manually and then implement an expensive system to monitor the manual work. This is an expensive and losing battle for large projects.
RBQM (which has good intentions to improve quality) is like pasting a bag to a leaky wall with scotch tape and announcing that the leak was fixed.
The caveats
Even if an AI does write your SOPs and KPIs — never forget that the leaders of a company/team need to own the process and the procedures. They need to train the team on the procedures and explain why it’s being done in a particular way.
If something goes wrong, and it will — leadership still owns the process, the operations and the business.
In summary — get ready for a smoother, clearer SOP experience with AI.
