Original Message:
Sent: 20-Jan-2024 07:51
From: Naveen Agarwal
Subject: How is AI/ML changing the nature of work in QA/RA?
Hello Ram - thank you for sharing this resource. Very cool!
I am quite interested in a Smart Search type of capability for warning letters, recalls, 510Ks, PMAs etc.
Best regards
------------------------------
Naveen Agarwal, Ph.D.
Problem Solver | Knowledge Sharer.
Let's Talk Risk!
@https://naveenagarwalphd.substack.com/
Original Message:
Sent: 20-Jan-2024 05:43
From: Ram Balani
Subject: How is AI/ML changing the nature of work in QA/RA?
Agree with most comments here, i.e. 'somewhat agree' seems the predominant answer, i.e. that AI/ML will impact QA/RA work given some routine while other efforts require that 'human intelligence' touch while recognizing that AI/ML is not always accurate and reliable..
Attended a recent event on Generative AI in Boston recently--the overwhelming conclusion was that AI/ML will become 'smarter' and generative LLM (large language model) will have less and less 'hallucinations' in time specially when the base language model i(Open AI LLM) is trained with an enterprise's internal data like SOPs, PDF policy documents, etc.
That said--AI/ML - regardless of how large the pre-training data is ingested will always have a predictive element to it-it isn't human thinking after all! That's the nature of machine learning until Star Trek like 'cyborgs' become real.
We came up with a sort of hybrid solution -see videos below and our take on this further in the attachment with DSCSA as a sample topic.
It's a hybrid app solution with an Open AI ChatGPT-3.5 app/bot COMBINED with a full-text indexed every word on US FDA Guidance PDFs page contents from FDA.Gov on medical devices, drugs and biologics--a total of 27664 items for our search app. That said--FDA Guidance on DSCSA are only 31 PDFs in total while supply chain references with on pharma drugs are peppered across multiple 21 CFRs!
Idea is simple--not rocket science--i.e. on the same app that runs on your mobile, laptop or browser--the app( bottom screen) allows one to prompt ChatGPT to discover possible search keyword hints from generative AI replies with your prompt... BUT also allows on the app top screen (no externa link switching) to search full-text--and here's the rub -- you're searching the ACTUAL FDA.Gov contents collected, e.g. on DSCSA for example -i.e. you're searching full text much like Google without the web noise ads or clutter whatever the FDA has actually published on DSCSA compliance.
No one can debate that FDA is the single source of 'regulatory truth' on a topic, so if the FDA publishes it, you'll discover it. No predictive AI/ML bias here.
If you're curious--see short video recording below.
Interested to hear anyone's feedback-but above is my 2-cents worth.
--------------------Video recording clips of SmartSearch+ on a Samsung Galaxy mobile phone:--------------
https://lnkd.in/g4-vchMN
πApp integrated with OpenAI ChatGPT-3.5 Turbo Chatbot & illustrates Why US FDA metadata or tags assigned based search is ABYSMAL!
http://tinyurl.com/278p3mwz
π App Sample search use case with a DSCSA stakeholder label
--find FDA Guidance or CFRs referencing 'DSCSA' with 'repacker' & 'labeling'
------------------------------
Ram Balani
CEO
FDASmart Inc. /eSTARHelper LLC www.estarhelper.com
Amawalk , New York
rbalani@fdasmart.com
2019130558
https://tinyurl.com/3jbjcd7v
on US FDA eSTAR for 510(K)
Original Message:
Sent: 18-Jan-2024 08:35
From: Naveen Agarwal
Subject: How is AI/ML changing the nature of work in QA/RA?
Dear colleagues, hello! π
It is now clear that rapid advancement in AI/ML is transforming life around us, including the nature of work in QA/RA space.
I have been curious about this trend specifically how it is affecting the nature of our work and what it means for career advancement.
According to a recent report from McKinsey Global Institute, generative AI is poised to drive automation even in professions that involve "expertise, interaction with people, and creativity". They estimate that by 2030, automation could account for upto 30% of the hours worked in the US economy.
Recently, I asked this question to my colleagues on LinkedIn in a poll -
To what extent do you agree with the following - "In the next 5 years, AI will significantly impact both the number and type of QA/RA jobs."?
Nearly 70% of the respondents said they either somewhat or strongly agreed with the statement. (Note: this is not a scientific poll; only an indication of how our colleagues are feeling about AI)
I had a follow up conversation with a colleague about this topic in my weekly Let's Talk Risk! discussions on LinkedIn. She has been recognized as a Top LinkedIn voice in recruiting, Here is an excerpt of our conversation. She encourages us to be early adopters of AI in our profession and not fight it!
Dear colleagues - I would love to hear your opinion. Do you see AI as a challenge or an opportunity for your career advancement in QA/RA?
Best regards
------------------------------
Naveen Agarwal, Ph.D.
Problem Solver | Knowledge Sharer.
Let's Talk Risk!
@https://naveenagarwalphd.substack.com/
------------------------------