Re: [PSES] AI & Regulatory Compliance

[James Pawson (U3C)]

Hi Doug,

 

Thanks for the useful links.

 

FYI the link for Elicit should be https://elicit.com/ - the “elecit dot org” 
gave my antivirus webshield some conniptions.

 

All the best

James

 

James Pawson

Managing Director & EMC Problem Solver

 

Unit 3 Compliance Ltd

EMC : Environmental & Vibration : Electrical Safety : CE & UKCA : Consultancy

 

 <http://www.unit3compliance.co.uk/> www.unit3compliance.co.uk |  
<mailto:ja...@unit3compliance.co.uk> ja...@unit3compliance.co.uk 

+44(0)1274 911747  |  +44(0)7811 139957

2 Wellington Business Park, New Lane, Bradford, BD4 8AL

Registered in England and Wales # 10574298

 

Office hours:

Every morning my full attention is on consultancy, testing, and troubleshooting 
activities for our customers’ projects. I’m contactable between 1300h to 1730h 
from Monday to Friday.

For inquiries, bookings, and testing updates please send us an email on 
he...@unit3compliance.co.uk <mailto:he...@unit3compliance.co.uk>  or call 01274 
911747. Our lead times for testing and consultancy are typically 4-5 weeks.

 

 

 

 

From: Douglas Nix <0bb8ff993b10-dmarc-requ...@listserv.ieee.org> 
Sent: Tuesday, October 10, 2023 10:35 PM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

Hey, Ralph,

 

Listen, I get being excited about AI and the possibilities. The problem I have 
with this discussion is the approach that some people are taking using this 
tool. There are some excellent AI-based academic tools available:

 

▸Wisio.app (limted free version)

▸Jenni.ai

▸OpenRead.academy

▸Other research tools include:

-Paper Digest (https://www.paper-digest.com <https://www.paper-digest.com/> )

-Elicit (https://elecit . org)

- Litmaps visual literature search (https://www.litmaps.com 
<https://www.litmaps.com/> )

-Research Rabbit (https://www.researchrabbit.ai 
<https://www.youtube.com/redirect?event=video_description_token=QUFFLUhqbWpMaHlHNnd2bmoxbjhnTVpXVkQzSmZPTERBZ3xBQ3Jtc0traDZLeFJmQlBNeEpLYkE3NDdfa21LWnM4WDJxekUwYkV1bGxTX3VmSmYwUWRtXzlIVkdMaGM3Y2gzUmY2UnFmbUZ3VG0ySFZFbElNNlpObFFHTldKYUdkZEtrbUNTaVE0c1JkV3JBYnJKRVBvblBMMA=https%3A%2F%2Fwww.researchrabbit.ai%2F=8oEOa9wStjk>
 )

-Connected Papers (https://www.connectedpapers.com 
<https://www.connectedpapers.com/> )

-writefull (https://www.writefull.com <https://www.writefull.com/> )

-Penelope.ai (https://www.penelope.ai <https://www.penelope.ai/> )

 

 

All of these tools can be very helpful. I’m a fan of Litmaps and Elicit in 
particular. I’m just starting to poke around at Wisio, Jenni, and OpenRead.

 

Doug Nix

d...@mac.com <mailto:d...@mac.com> 

 

"The man who does not read good books has no advantage over the man who can't 
read these books." Mark Twain 





On Oct 10, 2023, at 17:01, Ralph McDiarmid mailto:rmm.priv...@gmail.com> > wrote:

 

That analogy was not intended, and I’m merely trying to point out that AI is 
here to stay and is expected to become a more useful tool in our industry as it 
has been proven to be in the medical profession.

 

From: Douglas Nix mailto:d...@mac.com> > 
Sent: Tuesday, October 10, 2023 1:03 PM
To: rmm.priv...@gmail.com <mailto:rmm.priv...@gmail.com> 
Cc: EMC-PSTC@LISTSERV.IEEE.ORG <mailto:EMC-PSTC@LISTSERV.IEEE.ORG> 
Subject: Re: [PSES] AI & Regulatory Compliance

 

You cannot compare ChatGPT to a medical AI that has been tailored for a 
specific task like medical image analysis. That’s like saying a 1968 Mini 
Cooper and Dodge 440 HEMI Charger are comparable because they are both cars. 
Nope.

 

Using ChatGPT to summarize a paper, produce an abstract from uploaded text, or 
produce a set of points as a starting point is perfectly fine. You are giving 
the software the specific input material from which to generate the output. You 
cannot ask it research questions an expect a valid response because ChatGPT has 
no parameters for correctness. It only wants to give you a plausible sounding 
answer. It will give you an authoritative answer with no reference to anything 
resembling truth.

 

Doug Nix

 <mailto:d...@mac.com> d...@mac.com

 <http://www.dougnix.net/> http://www.dougnix.net

(519) 729-5704

 

"All animals except man know that the ultimate joy of life is to enjoy it."  -- 
Samuel Butler

 

 






On Oct 10, 2023, at 14:15, Ralph McDiarmid < <mailto:rmm.priv...@gmail.com> 
rmm.priv...@gmail.com> wrote:

 

Physicians have used AI (expert systems) in their offices for many years as a 
tool to help them diagnose a problem more accurately and will greater speed.  
It’s a tool to speed productivity, and that’s how I use it today.

 

  _  

This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to All 
emc-pstc postings are archived and searchable on the web at: 
 <https://

Re: [PSES] AI & Regulatory Compliance

[Vincent Lee]

Hi all,
 
ChatGPT is just joining of words and characters using model called transformer 
that may sometimes, sound reasonable to human.
AI is certainly cannot replacing human after all human is much more intelligent 
with wisdom too.AI don't have wisdom, feeling and etc. Thus, human cannot be 
replace regardless.

Regards, Vincent 

On Wednesday, October 11, 2023 at 02:31:09 AM GMT+8, Ralph McDiarmid 
 wrote:  
 
 
I would add that about 9 out of 10 power engineers I’ve worked with over the 
years are equally confused with the “convention” of power factor and sign 
convention of real and reactive power flow.  Microelectronics engineers also 
often get the sign convention wrong when considering current source or sink 
from a logic gate or an op-amp.   I wouldn’t beat up on ChatGPT for struggling 
with that very specific topic.

  

Ralph

  

From: Brian Gregory  
Sent: Tuesday, October 10, 2023 9:16 AM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

  

 

Chat GPT is essentially a BS generator.  A very smart friend, a very successful 
entrepreneur, finds it quite useful for writing add copy, which I think proved 
my point.

 

The one time I tried challenging GPT with a question on power factor with 
generator convention (where positive power generation is represented as 
negative), it failed miserably. 

 

Echoing Dan's point;  what other AI's are more suitable to electrical 
engineering concepts, including testing and reviewing lists of standards?  If I 
had to pick one question for an AI I think might qualify would be:  " does this 
requirement apply to a residential unit? "  

 

 Colorado Brian 



-- Forwarded Message --
From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org>
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance
Date: Mon, 9 Oct 2023 22:21:26 -0400

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--

Dan Roman

danp...@verizon.net

On Oct 9, 2023, at 7:31 PM, Douglas Powell  wrote: 


I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

  

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt  wrote:


Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:

"How do you think AI will play a future role in our Regulatory compliance 
world?"

 

I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.…though I think we would have to include the usage of AI as one of the 
potential risks in the risk assessment. lol

UL touched base on this topic during their UL Innovations summit held in 
Fremont, CA yesterday.

I know there is the ISO/IEC DTR 5469 in development, but this appears, I 
believe, to only a

Re: [PSES] AI & Regulatory Compliance

[Douglas Nix]

Hey, Ralph,

Listen, I get being excited about AI and the possibilities. The problem I have 
with this discussion is the approach that some people are taking using this 
tool. There are some excellent AI-based academic tools available:

▸Wisio.app (limted free version)
▸Jenni.ai
▸OpenRead.academy
▸Other research tools include:
-Paper Digest (https://www.paper-digest.com <https://www.paper-digest.com/>)
-Elicit (https://elecit.org <https://elecit.org/>)
- Litmaps visual literature search (https://www.litmaps.com 
<https://www.litmaps.com/>)
-Research Rabbit (https://www.researchrabbit.ai 
<https://www.youtube.com/redirect?event=video_description_token=QUFFLUhqbWpMaHlHNnd2bmoxbjhnTVpXVkQzSmZPTERBZ3xBQ3Jtc0traDZLeFJmQlBNeEpLYkE3NDdfa21LWnM4WDJxekUwYkV1bGxTX3VmSmYwUWRtXzlIVkdMaGM3Y2gzUmY2UnFmbUZ3VG0ySFZFbElNNlpObFFHTldKYUdkZEtrbUNTaVE0c1JkV3JBYnJKRVBvblBMMA=https%3A%2F%2Fwww.researchrabbit.ai%2F=8oEOa9wStjk>)
-Connected Papers (https://www.connectedpapers.com 
<https://www.connectedpapers.com/>)
-writefull (https://www.writefull.com <https://www.writefull.com/>)
-Penelope.ai (https://www.penelope.ai <https://www.penelope.ai/>)


All of these tools can be very helpful. I’m a fan of Litmaps and Elicit in 
particular. I’m just starting to poke around at Wisio, Jenni, and OpenRead.

Doug Nix
d...@mac.com

"The man who does not read good books has no advantage over the man who can't 
read these books." Mark Twain 

> On Oct 10, 2023, at 17:01, Ralph McDiarmid  wrote:
> 
> That analogy was not intended, and I’m merely trying to point out that AI is 
> here to stay and is expected to become a more useful tool in our industry as 
> it has been proven to be in the medical profession.
>  
> From: Douglas Nix  
> Sent: Tuesday, October 10, 2023 1:03 PM
> To: rmm.priv...@gmail.com
> Cc: EMC-PSTC@LISTSERV.IEEE.ORG
> Subject: Re: [PSES] AI & Regulatory Compliance
>  
> You cannot compare ChatGPT to a medical AI that has been tailored for a 
> specific task like medical image analysis. That’s like saying a 1968 Mini 
> Cooper and Dodge 440 HEMI Charger are comparable because they are both cars. 
> Nope.
>  
> Using ChatGPT to summarize a paper, produce an abstract from uploaded text, 
> or produce a set of points as a starting point is perfectly fine. You are 
> giving the software the specific input material from which to generate the 
> output. You cannot ask it research questions an expect a valid response 
> because ChatGPT has no parameters for correctness. It only wants to give you 
> a plausible sounding answer. It will give you an authoritative answer with no 
> reference to anything resembling truth.
>  
> Doug Nix
> d...@mac.com <mailto:d...@mac.com>
> http://www.dougnix.net <http://www.dougnix.net/>
> (519) 729-5704
>  
> "All animals except man know that the ultimate joy of life is to enjoy it."  
> -- Samuel Butler
>  
>  
> 
> 
>> On Oct 10, 2023, at 14:15, Ralph McDiarmid > <mailto:rmm.priv...@gmail.com>> wrote:
>>  
>> Physicians have used AI (expert systems) in their offices for many years as 
>> a tool to help them diagnose a problem more accurately and will greater 
>> speed.  It’s a tool to speed productivity, and that’s how I use it today.
> 
>  
> This message is from the IEEE Product Safety Engineering Society emc-pstc 
> discussion list. To post a message to the list, send your e-mail to All 
> emc-pstc postings are archived and searchable on the web at: 
> https://www.mail-archive.com/emc-pstc@listserv.ieee.org/
> 
> Website: https://ewh.ieee.org/soc/pses/  <https://ewh.ieee.org/soc/pses/>
> Instructions: https://ewh.ieee.org/soc/pses/list.html (including how to 
> unsubscribe) <https://ewh.ieee.org/soc/pses/list.html>
> List rules: https://ewh.ieee.org/soc/pses/listrules.html
> 
> For help, send mail to the list administrators:
> Mike Sherman at: msherma...@comcast.net <mailto:msherma...@comcast.net>
> Rick Linford at: linf...@ieee.org <mailto:linf...@ieee.org>
> For policy questions, send mail to:
> Jim Bacher at: j.bac...@ieee.org <mailto:j.bac...@ieee.org>
> To unsubscribe from the EMC-PSTC list, click the following link: 
> https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1 
> 


-

This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to 


All emc-pstc postings are archived and searchable on the web at:
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/

Website:  https://ewh.ieee.org/soc/pses/
Instructions:  https://ewh.ieee.org/soc/pses/list.html (including how to 
unsubscribe)
List rules: https://ewh.ieee.org/s

Re: [PSES] AI & Regulatory Compliance

[Ralph McDiarmid]

That analogy was not intended, and I’m merely trying to point out that AI is 
here to stay and is expected to become a more useful tool in our industry as it 
has been proven to be in the medical profession.

 

From: Douglas Nix  
Sent: Tuesday, October 10, 2023 1:03 PM
To: rmm.priv...@gmail.com
Cc: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

You cannot compare ChatGPT to a medical AI that has been tailored for a 
specific task like medical image analysis. That’s like saying a 1968 Mini 
Cooper and Dodge 440 HEMI Charger are comparable because they are both cars. 
Nope.

 

Using ChatGPT to summarize a paper, produce an abstract from uploaded text, or 
produce a set of points as a starting point is perfectly fine. You are giving 
the software the specific input material from which to generate the output. You 
cannot ask it research questions an expect a valid response because ChatGPT has 
no parameters for correctness. It only wants to give you a plausible sounding 
answer. It will give you an authoritative answer with no reference to anything 
resembling truth.

 

Doug Nix

d...@mac.com <mailto:d...@mac.com> 

http://www.dougnix.net

(519) 729-5704

 

"All animals except man know that the ultimate joy of life is to enjoy it."  -- 
Samuel Butler

 

 





On Oct 10, 2023, at 14:15, Ralph McDiarmid mailto:rmm.priv...@gmail.com> > wrote:

 

Physicians have used AI (expert systems) in their offices for many years as a 
tool to help them diagnose a problem more accurately and will greater speed.  
It’s a tool to speed productivity, and that’s how I use it today.

 


-

This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to 


All emc-pstc postings are archived and searchable on the web at:
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/

Website:  https://ewh.ieee.org/soc/pses/
Instructions:  https://ewh.ieee.org/soc/pses/list.html (including how to 
unsubscribe)
List rules: https://ewh.ieee.org/soc/pses/listrules.html

For help, send mail to the list administrators:
Mike Sherman at: msherma...@comcast.net
Rick Linford at: linf...@ieee.org

For policy questions, send mail to:
Jim Bacher:  
_
To unsubscribe from the EMC-PSTC list, click the following link: 
https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1

Re: [PSES] AI & Regulatory Compliance

[Douglas Nix]

You cannot compare ChatGPT to a medical AI that has been tailored for a 
specific task like medical image analysis. That’s like saying a 1968 Mini 
Cooper and Dodge 440 HEMI Charger are comparable because they are both cars. 
Nope.

Using ChatGPT to summarize a paper, produce an abstract from uploaded text, or 
produce a set of points as a starting point is perfectly fine. You are giving 
the software the specific input material from which to generate the output. You 
cannot ask it research questions an expect a valid response because ChatGPT has 
no parameters for correctness. It only wants to give you a plausible sounding 
answer. It will give you an authoritative answer with no reference to anything 
resembling truth.

Doug Nix
d...@mac.com
http://www.dougnix.net
(519) 729-5704

"All animals except man know that the ultimate joy of life is to enjoy it."  -- 
Samuel Butler



> On Oct 10, 2023, at 14:15, Ralph McDiarmid  wrote:
> 
> Physicians have used AI (expert systems) in their offices for many years as a 
> tool to help them diagnose a problem more accurately and will greater speed.  
> It’s a tool to speed productivity, and that’s how I use it today.


-

This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to 


All emc-pstc postings are archived and searchable on the web at:
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/

Website:  https://ewh.ieee.org/soc/pses/
Instructions:  https://ewh.ieee.org/soc/pses/list.html (including how to 
unsubscribe)
List rules: https://ewh.ieee.org/soc/pses/listrules.html

For help, send mail to the list administrators:
Mike Sherman at: msherma...@comcast.net
Rick Linford at: linf...@ieee.org

For policy questions, send mail to:
Jim Bacher:  
_
To unsubscribe from the EMC-PSTC list, click the following link: 
https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1

Re: [PSES] AI & Regulatory Compliance

[Ralph McDiarmid]

Which is another way of saying that ChatGPT and other similar AI are tools used 
by professionals to assist with improving their productivity, but not something 
to be used in lieu of a professional.  (at least not in this decade).  That 
time may come, but I suspect engineering jobs are safe for the foreseeable 
future.

 

Ralph

 

From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org> 
Sent: Tuesday, October 10, 2023 10:51 AM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

Other than playing around with text queries the most useful thing I have found 
either Chat GPT or Bard (the two I have used regularly) good for is Python 
coding.  While I can never take the code as is, it does offer tips on libraries 
that I did not even know existing to more easily perform a task.  I think 
others have hit on a similar use case for presentations and other output.  The 
AI can be useful to give you some ideas that you might not have already had and 
help you think about a problem or task in a different way  As with my Python 
example, you can’t just take what it gives you, you have to do the work and 
perhaps use some clues or ideas from the AI.

 

Dan

 

 

From: Brian Gregory [mailto:brian_greg...@netzero.net] 
Sent: Tuesday, October 10, 2023 12:16 PM
To: EMC-PSTC@LISTSERV.IEEE.ORG <mailto:EMC-PSTC@LISTSERV.IEEE.ORG> 
Subject: Re: [PSES] AI & Regulatory Compliance

 

 

Chat GPT is essentially a BS generator.  A very smart friend, a very successful 
entrepreneur, finds it quite useful for writing add copy, which I think proved 
my point.

 

The one time I tried challenging GPT with a question on power factor with 
generator convention (where positive power generation is represented as 
negative), it failed miserably. 

 

Echoing Dan's point;  what other AI's are more suitable to electrical 
engineering concepts, including testing and reviewing lists of standards?  If I 
had to pick one question for an AI I think might qualify would be:  " does this 
requirement apply to a residential unit? "  

 

 Colorado Brian 



-- Forwarded Message --
From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org 
<mailto:0d75e04ed751-dmarc-requ...@listserv.ieee.org> >
To: EMC-PSTC@LISTSERV.IEEE.ORG <mailto:EMC-PSTC@LISTSERV.IEEE.ORG> 
Subject: Re: [PSES] AI & Regulatory Compliance
Date: Mon, 9 Oct 2023 22:21:26 -0400

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--

Dan Roman

danp...@verizon.net <mailto:danp...@verizon.net> 

On Oct 9, 2023, at 7:31 PM, Douglas Powell mailto:doug...@gmail.com> > wrote: 

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

 <https://www.linkedin.com/in/coloradocomplianceguy/> LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

 

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt mailto:reganar...@gmail.com> > wrote:

Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
brin

Re: [PSES] AI & Regulatory Compliance

[Ralph McDiarmid]

I would add that about 9 out of 10 power engineers I’ve worked with over the 
years are equally confused with the “convention” of power factor and sign 
convention of real and reactive power flow.  Microelectronics engineers also 
often get the sign convention wrong when considering current source or sink 
from a logic gate or an op-amp.   I wouldn’t beat up on ChatGPT for struggling 
with that very specific topic.

 

Ralph

 

From: Brian Gregory  
Sent: Tuesday, October 10, 2023 9:16 AM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

 

Chat GPT is essentially a BS generator.  A very smart friend, a very successful 
entrepreneur, finds it quite useful for writing add copy, which I think proved 
my point.

 

The one time I tried challenging GPT with a question on power factor with 
generator convention (where positive power generation is represented as 
negative), it failed miserably. 

 

Echoing Dan's point;  what other AI's are more suitable to electrical 
engineering concepts, including testing and reviewing lists of standards?  If I 
had to pick one question for an AI I think might qualify would be:  " does this 
requirement apply to a residential unit? "  

 

 Colorado Brian 



-- Forwarded Message --
From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org 
<mailto:0d75e04ed751-dmarc-requ...@listserv.ieee.org> >
To: EMC-PSTC@LISTSERV.IEEE.ORG <mailto:EMC-PSTC@LISTSERV.IEEE.ORG> 
Subject: Re: [PSES] AI & Regulatory Compliance
Date: Mon, 9 Oct 2023 22:21:26 -0400

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--

Dan Roman

danp...@verizon.net <mailto:danp...@verizon.net> 

On Oct 9, 2023, at 7:31 PM, Douglas Powell mailto:doug...@gmail.com> > wrote: 

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

 <https://www.linkedin.com/in/coloradocomplianceguy/> LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

 

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt mailto:reganar...@gmail.com> > wrote:

Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:

"How do you think AI will play a future role in our Regulatory compliance 
world?"

 

I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.…though I think we would have to include the usage of AI as one of the 
potential risks in the risk assessment. lol

UL touched base on this topic during their UL Innovations summit held in 
Fremont, CA yesterday.

I know there is the ISO/IEC DTR 5469 in development, but this appears, I 
believe, to only address functional safety of the AI systems in question, and 
not addressing AI in normal compliance assessments for products under the 
category of 

Re: [PSES] AI & Regulatory Compliance

[Ralph McDiarmid]

Very interesting demo of its capabilities.

 

I am wondering if the $20 per month version of ChatGPT might be worthwhile.

 

Ralph

 

From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org> 
Sent: Monday, October 9, 2023 7:21 PM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--

Dan Roman

danp...@verizon.net <mailto:danp...@verizon.net> 

On Oct 9, 2023, at 7:31 PM, Douglas Powell mailto:doug...@gmail.com> > wrote:

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

 <https://www.linkedin.com/in/coloradocomplianceguy/> LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

 

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt mailto:reganar...@gmail.com> > wrote:

Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:

"How do you think AI will play a future role in our Regulatory compliance 
world?"

 

I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.…though I think we would have to include the usage of AI as one of the 
potential risks in the risk assessment. lol

UL touched base on this topic during their UL Innovations summit held in 
Fremont, CA yesterday.

I know there is the ISO/IEC DTR 5469 in development, but this appears, I 
believe, to only address functional safety of the AI systems in question, and 
not addressing AI in normal compliance assessments for products under the 
category of laboratory, IT, audio, etc.

Also, is there someone out there that will be speaking to this during the 2024 
ISPCE in Chicago next May? Perhaps there should be an IEEE PSES technical 
committee created for this?

Side note: Nordcloud puts out a good summary blog on this:  
<https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/> 
https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/

There are other AI blogs out there in the ether but more tuned for the 
financial sector and other industries….

Looking forward to the discussion. 

 

P.S.

I am now the Chair for the IEEE PSES Risk assessment technical committee and am 
looking for any volunteers to join our group. This topic will be one to discuss 
soon within our group. Please email me at  <mailto:regan.ar...@ieee.org> 
regan.ar...@ieee.org or  <mailto:reganar...@gmail.com> reganar...@gmail.com if 
you are interested in joining.

Cheers!

Regan Arndt


  _  


This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to All 
emc-pstc postings are archived and searchable on the web at: 
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/ 

Website: htt

Re: [PSES] AI & Regulatory Compliance

[Ralph McDiarmid]

“Even so, I do find them useful as a starting point for writing White Papers 
and PowerPoint slides or procedural documents at the paragraph-level, for 
internal distribution. “

 

I do the same.  It’s useful now, even though not perfect, and AI is only going 
to get better and that will happen rapidly.

 

Physicians have used AI (expert systems) in their offices for many years as a 
tool to help them diagnose a problem more accurately and will greater speed.  
It’s a tool to speed productivity, and that’s how I use it today.

 

Ralph

 

From: Douglas Powell  
Sent: Monday, October 9, 2023 4:30 PM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

 <https://www.linkedin.com/in/coloradocomplianceguy/> LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

 

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt mailto:reganar...@gmail.com> > wrote:

Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:

"How do you think AI will play a future role in our Regulatory compliance 
world?"

 

I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.…though I think we would have to include the usage of AI as one of the 
potential risks in the risk assessment. lol

UL touched base on this topic during their UL Innovations summit held in 
Fremont, CA yesterday.

I know there is the ISO/IEC DTR 5469 in development, but this appears, I 
believe, to only address functional safety of the AI systems in question, and 
not addressing AI in normal compliance assessments for products under the 
category of laboratory, IT, audio, etc.

Also, is there someone out there that will be speaking to this during the 2024 
ISPCE in Chicago next May? Perhaps there should be an IEEE PSES technical 
committee created for this?

Side note: Nordcloud puts out a good summary blog on this:  
<https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/> 
https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/

There are other AI blogs out there in the ether but more tuned for the 
financial sector and other industries….

Looking forward to the discussion. 

 

P.S.

I am now the Chair for the IEEE PSES Risk assessment technical committee and am 
looking for any volunteers to join our group. This topic will be one to discuss 
soon within our group. Please email me at  <mailto:regan.ar...@ieee.org> 
regan.ar...@ieee.org or  <mailto:reganar...@gmail.com> reganar...@gmail.com if 
you are interested in joining.

Cheers!

Regan Arndt


  _  


This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to All 
emc-pstc postings are archived and searchable on the web at: 
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/ 

Website: https://ewh.ieee.org/soc/pses/  <https://ewh.ieee.org/soc/pses/> 
Instructions: https://ewh.ieee.org/soc/pses/list.html (including how to 
unsubscribe) <https://ewh.

Re: [PSES] AI & Regulatory Compliance

[Dan Roman]

Other than playing around with text queries the most useful thing I have found 
either Chat GPT or Bard (the two I have used regularly) good for is Python 
coding.  While I can never take the code as is, it does offer tips on libraries 
that I did not even know existing to more easily perform a task.  I think 
others have hit on a similar use case for presentations and other output.  The 
AI can be useful to give you some ideas that you might not have already had and 
help you think about a problem or task in a different way  As with my Python 
example, you can’t just take what it gives you, you have to do the work and 
perhaps use some clues or ideas from the AI.

 

Dan

 

 

From: Brian Gregory [mailto:brian_greg...@netzero.net] 
Sent: Tuesday, October 10, 2023 12:16 PM
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance

 

 

Chat GPT is essentially a BS generator.  A very smart friend, a very successful 
entrepreneur, finds it quite useful for writing add copy, which I think proved 
my point.

 

The one time I tried challenging GPT with a question on power factor with 
generator convention (where positive power generation is represented as 
negative), it failed miserably. 

 

Echoing Dan's point;  what other AI's are more suitable to electrical 
engineering concepts, including testing and reviewing lists of standards?  If I 
had to pick one question for an AI I think might qualify would be:  " does this 
requirement apply to a residential unit? "  

 

 Colorado Brian 



-- Forwarded Message --
From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org 
<mailto:0d75e04ed751-dmarc-requ...@listserv.ieee.org> >
To: EMC-PSTC@LISTSERV.IEEE.ORG <mailto:EMC-PSTC@LISTSERV.IEEE.ORG> 
Subject: Re: [PSES] AI & Regulatory Compliance
Date: Mon, 9 Oct 2023 22:21:26 -0400

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--

Dan Roman

danp...@verizon.net <mailto:danp...@verizon.net> 

On Oct 9, 2023, at 7:31 PM, Douglas Powell mailto:doug...@gmail.com> > wrote: 

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng 
Chatbot, and a few others.  

 

My first test was to see if these Large Language Models (LLM) could accurately 
answer questions where I feel I have good expertise, such as "How do I select a 
complete list of safety-critical components using UL XYZ?" or "Write a 
comprehensive Test Protocol for  a  kitchen appliance using IEC 
60335-*X-X." For the most part they failed in several points when the questions 
got just a little technical.  Sometimes, I would challenge the AI on answers it 
gave, and occasionally it would backpedal. 

 

Even so, I do find them useful as a starting point for writing White Papers and 
PowerPoint slides or procedural documents at the paragraph-level, for internal 
distribution.  Try prompting an AI to generate 20 PowerPoint slides for basic 
Risk Assessment Training using ISO 31000.  In every case however, I only use AI 
to do the initial grunt work and then personally edit everything for accuracy. 
So far, my favorite is Perplexity.AI because it gives attribution of sources 
for all its answers, second place is possibly Claude.AI.  And been using 
QuillBot.AI to check grammar and paraphrase complex paragraphs.  But when it 
comes to life-safety issues, I will always be very careful and check the work.

 

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll 
prompt the AI with something like: "Entropy simply isn’t what it used to be."  
or "Schrödinger’s cat walks into a bar. And doesn’t." And of course the AI 
never gets the joke.  

 

-Doug

 

Douglas E Powell

Laporte, Colorado, USA

 <https://www.linkedin.com/in/coloradocomplianceguy/> LinkedIn

 

(UTC-06:00, US-MDT)

 

 

 

 

 

On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt mailto:reganar...@gmail.com> > wrote:

Greetings fellow members,

Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:

"How do you think AI will play a future role in our Regulatory compliance 
world?"

 

I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.…though I think we would have to include the usage of AI as one of the 
potential risks in the risk assessment. lol

UL touched base on this topic during their UL Innovation

Re: [PSES] AI & Regulatory Compliance

[Brian Gregory]

 Chat GPT is essentially a BS generator.  A very smart friend, a very 
successful entrepreneur, finds it quite useful for writing add copy, which I 
think proved my point. The one time I tried challenging GPT with a question on 
power factor with generator convention (where positive power generation is 
represented as negative), it failed miserably.  Echoing Dan's point;  what 
other AI's are more suitable to electrical engineering concepts, including 
testing and reviewing lists of standards?  If I had to pick one question for an 
AI I think might qualify would be:  " does this requirement apply to a 
residential unit? "Colorado Brian 

-- Forwarded Message --
From: Dan Roman <0d75e04ed751-dmarc-requ...@listserv.ieee.org>
To: EMC-PSTC@LISTSERV.IEEE.ORG
Subject: Re: [PSES] AI & Regulatory Compliance
Date: Mon, 9 Oct 2023 22:21:26 -0400


Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

--Dan Romandanpses@verizon.netOn Oct 9, 2023, at 7:31 PM, Douglas Powell 
 wrote:I've done some playing around with ChatGPT, 
Claude.AI, Perplexity.AI, BIng Chatbot, and a few others.   My first test was 
to see if these Large Language Models (LLM) could accurately answer questions 
where I feel I have good expertise, such as "How do I select a complete list of 
safety-critical components using UL XYZ?" or "Write a comprehensive Test 
Protocol for  a  kitchen appliance using IEC 60335-*X-X." For the most part 
they failed in several points when the questions got just a little technical.  
Sometimes, I would challenge the AI on answers it gave, and occasionally it 
would backpedal.  Even so, I do find them useful as a starting point for 
writing White Papers and PowerPoint slides or procedural documents at the 
paragraph-level, for internal distribution.  Try prompting an AI to generate 20 
PowerPoint slides for basic Risk Assessment Training using ISO 31000.  In every 
case however, I only use AI to do the initial grunt work and then personally 
edit everything for accuracy. So far, my favorite is Perplexity.AI because it 
gives attribution of sources for all its answers, second place is possibly 
Claude.AI.  And been using QuillBot.AI to check grammar and paraphrase complex 
paragraphs.  But when it comes to life-safety issues, I will always be very 
careful and check the work. Incidentally, I've tried a few fun things as a 
lunchtime diversion.  I'll prompt the AI with something like: "Entropy simply 
isnt what it used to be."  or "Schrödingers cat walks into a bar. 
And doesnt." And of course the AI never gets the joke.   -Doug
 Douglas E PowellLaporte, Colorado, USALinkedIn (UTC-06:00, US-MDT)
On Fri, Oct 6, 2023 at 3:59PM Regan Arndt  
wrote:Greetings fellow members,
Our industry will not be immune to this new era of AI technology and want to 
bring up this topic again to obtain some more insight from the forum into:
"How do you think AI will play a future role in our Regulatory compliance 
world?"
 
I can see some real benefits to this when applying it to a risk assessment, EMC 
redesign including change out of critical components, & possibly to help 
support engineering judgments in lieu of testing (i.e. temperature modeling), 
etc.though I think we would have to include the usage of AI as one of 
the potential risks in the risk assessment. lol
UL touched base on this topic during their UL Innovations summit held in 
Fremont, CA yesterday.
I know there is the ISO/IEC DTR 5469 in development, but this appears, I 
believe, to only address functional safety of the AI systems in question, and 
not addressing AI in normal compliance assessments for products under the 
category of laboratory, IT, audio, etc.
Also, is there someone out there that will be speaking to this during the 2024 
ISPCE in Chicago next May? Perhaps there should be an IEEE PSES technical 
committee created for this?
Side note: Nordcloud puts out a good summary blog on this: 
https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/
There are other AI blogs out there in the ether but more tuned for the 
financial sector and other industries.
Looking forward to the discussion.
 
P.S.
I am now the Chair for the IEEE PSES Risk assessment technical committee and am 
looking for any volunteers to join our group. This topic will be one to discuss 
soon within our group. Please email me at regan.ar...@ieee.org or 
reganar...@gmail.com if you are interested in joining.
Cheers!
Regan Arndt
This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send yo

Re: [PSES] AI & Regulatory Compliance

[Dan Roman]

Has anyone tried feeding a standard or set of standards, into Chat GPT for 
example, and ask it product construction or testing questions? I've seen demos 
of Chat GPT digesting complex equipment manuals and being able to generate what 
amounts to a quick start guide and answer questions about operating or 
servicing the equipment. I can't try this with the free version.

⁣--
Dan Roman
danp...@verizon.net​

On Oct 9, 2023, 7:31 PM, at 7:31 PM, Douglas Powell  wrote:
>I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI,
>BIng
>Chatbot, and a few others.
>
>My first test was to see if these Large Language Models (LLM) could
>accurately answer questions where I feel I have good expertise, such as
>"*How
>do I select a complete list of safety-critical components using UL
>XYZ?*"
>or *"Write a comprehensive Test Protocol for  a  kitchen
>appliance using IEC 60335-*X-X.*" For the most part they failed in
>several
>points when the questions got just a little technical.  Sometimes, I
>would challenge the AI on answers it gave, and occasionally it would
>backpedal.
>
>Even so, I do find them useful as a starting point for writing White
>Papers
>and PowerPoint slides or procedural documents at the paragraph-level,
>for
>internal distribution.  Try prompting an AI to generate 20 PowerPoint
>slides for basic Risk Assessment Training using ISO 31000.  In every
>case
>however, I only use AI to do the initial grunt work and then
>personally edit everything for accuracy. So far, my favorite is
>Perplexity.AI because it gives attribution of sources for all its
>answers,
>second place is possibly Claude.AI.  And been using QuillBot.AI to
>check
>grammar and paraphrase complex paragraphs.  But when it comes to
>life-safety issues, I will always be very careful and check the work.
>
>Incidentally, I've tried a few fun things as a lunchtime diversion.
>I'll
>prompt the AI with something like: "*Entropy simply isn’t what it used
>to
>be.*"  or "*Schrödinger’s cat walks into a bar. And doesn’t.*" And of
>course the AI never gets the joke.
>
>-Doug
>
>Douglas E Powell
>Laporte, Colorado, USA
>LinkedIn 
>
>(UTC-06:00, US-MDT)
>
>
>
>
>
>On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt 
>wrote:
>
>> Greetings fellow members,
>>
>> Our industry will not be immune to this new era of AI technology and
>want
>> to bring up this topic again to obtain some more insight from the
>forum
>> into:
>>
>> *"How do you think AI will play a future role in our Regulatory
>compliance
>> world?"*
>>
>>
>> I can see some real benefits to this when applying it to a risk
>> assessment, EMC redesign including change out of critical components,
>&
>> possibly to help support engineering judgments in lieu of testing
>(i.e.
>> temperature modeling), etc.…though I think we would have to include
>the
>> usage of AI as one of the potential risks in the risk assessment. lol
>>
>> UL touched base on this topic during their UL Innovations summit held
>in
>> Fremont, CA yesterday.
>>
>> I know there is the ISO/IEC DTR 5469 in development, but this
>appears, I
>> believe, to only address functional safety of the AI systems in
>question,
>> and not addressing AI in normal compliance assessments for products
>under
>> the category of laboratory, IT, audio, etc.
>>
>> Also, is there someone out there that will be speaking to this during
>the
>> 2024 ISPCE in Chicago next May? Perhaps there should be an IEEE PSES
>> technical committee created for this?
>>
>> Side note: Nordcloud puts out a good summary blog on this:
>>
>https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/
>>
>> There are other AI blogs out there in the ether but more tuned for
>the
>> financial sector and other industries….
>>
>> Looking forward to the discussion.
>>
>>
>>
>> P.S.
>>
>> I am now the Chair for the IEEE PSES Risk assessment technical
>committee
>> and am looking for any volunteers to join our group. This topic will
>be one
>> to discuss soon within our group. Please email me at
>regan.ar...@ieee.org
>> or reganar...@gmail.com if you are interested in joining.
>>
>> Cheers!
>>
>> Regan Arndt
>> --
>>
>> This message is from the IEEE Product Safety Engineering Society
>emc-pstc
>> discussion list. To post a message to the list, send your e-mail to
>All
>> emc-pstc postings are archived and searchable on the web at:
>> https://www.mail-archive.com/emc-pstc@listserv.ieee.org/
>>
>> Website: https://ewh.ieee.org/soc/pses/
>> Instructions: https://ewh.ieee.org/soc/pses/list.html (including how
>to
>> unsubscribe) 
>> List rules: https://ewh.ieee.org/soc/pses/listrules.html
>>
>> For help, send mail to the list administrators:
>> Mike Sherman at: msherma...@comcast.net
>> Rick Linford at: linf...@ieee.org
>>
>> For policy questions, send mail to:
>> Jim Bacher at: j.bac...@ieee.org
>> --
>>
>> To 

Re: [PSES] AI & Regulatory Compliance

[Douglas Powell]

I've done some playing around with ChatGPT, Claude.AI, Perplexity.AI, BIng
Chatbot, and a few others.

My first test was to see if these Large Language Models (LLM) could
accurately answer questions where I feel I have good expertise, such as "*How
do I select a complete list of safety-critical components using UL XYZ?*"
or *"Write a comprehensive Test Protocol for  a  kitchen
appliance using IEC 60335-*X-X.*" For the most part they failed in several
points when the questions got just a little technical.  Sometimes, I
would challenge the AI on answers it gave, and occasionally it would
backpedal.

Even so, I do find them useful as a starting point for writing White Papers
and PowerPoint slides or procedural documents at the paragraph-level, for
internal distribution.  Try prompting an AI to generate 20 PowerPoint
slides for basic Risk Assessment Training using ISO 31000.  In every case
however, I only use AI to do the initial grunt work and then
personally edit everything for accuracy. So far, my favorite is
Perplexity.AI because it gives attribution of sources for all its answers,
second place is possibly Claude.AI.  And been using QuillBot.AI to check
grammar and paraphrase complex paragraphs.  But when it comes to
life-safety issues, I will always be very careful and check the work.

Incidentally, I've tried a few fun things as a lunchtime diversion.  I'll
prompt the AI with something like: "*Entropy simply isn’t what it used to
be.*"  or "*Schrödinger’s cat walks into a bar. And doesn’t.*" And of
course the AI never gets the joke.

-Doug

Douglas E Powell
Laporte, Colorado, USA
LinkedIn 

(UTC-06:00, US-MDT)





On Fri, Oct 6, 2023 at 3:59 PM Regan Arndt  wrote:

> Greetings fellow members,
>
> Our industry will not be immune to this new era of AI technology and want
> to bring up this topic again to obtain some more insight from the forum
> into:
>
> *"How do you think AI will play a future role in our Regulatory compliance
> world?"*
>
>
> I can see some real benefits to this when applying it to a risk
> assessment, EMC redesign including change out of critical components, &
> possibly to help support engineering judgments in lieu of testing (i.e.
> temperature modeling), etc.…though I think we would have to include the
> usage of AI as one of the potential risks in the risk assessment. lol
>
> UL touched base on this topic during their UL Innovations summit held in
> Fremont, CA yesterday.
>
> I know there is the ISO/IEC DTR 5469 in development, but this appears, I
> believe, to only address functional safety of the AI systems in question,
> and not addressing AI in normal compliance assessments for products under
> the category of laboratory, IT, audio, etc.
>
> Also, is there someone out there that will be speaking to this during the
> 2024 ISPCE in Chicago next May? Perhaps there should be an IEEE PSES
> technical committee created for this?
>
> Side note: Nordcloud puts out a good summary blog on this:
> https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/
>
> There are other AI blogs out there in the ether but more tuned for the
> financial sector and other industries….
>
> Looking forward to the discussion.
>
>
>
> P.S.
>
> I am now the Chair for the IEEE PSES Risk assessment technical committee
> and am looking for any volunteers to join our group. This topic will be one
> to discuss soon within our group. Please email me at regan.ar...@ieee.org
> or reganar...@gmail.com if you are interested in joining.
>
> Cheers!
>
> Regan Arndt
> --
>
> This message is from the IEEE Product Safety Engineering Society emc-pstc
> discussion list. To post a message to the list, send your e-mail to All
> emc-pstc postings are archived and searchable on the web at:
> https://www.mail-archive.com/emc-pstc@listserv.ieee.org/
>
> Website: https://ewh.ieee.org/soc/pses/
> Instructions: https://ewh.ieee.org/soc/pses/list.html (including how to
> unsubscribe) 
> List rules: https://ewh.ieee.org/soc/pses/listrules.html
>
> For help, send mail to the list administrators:
> Mike Sherman at: msherma...@comcast.net
> Rick Linford at: linf...@ieee.org
>
> For policy questions, send mail to:
> Jim Bacher at: j.bac...@ieee.org
> --
>
> To unsubscribe from the EMC-PSTC list, click the following link:
> https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1
>

-

This message is from the IEEE Product Safety Engineering Society emc-pstc 
discussion list. To post a message to the list, send your e-mail to 


All emc-pstc postings are archived and searchable on the web at:
https://www.mail-archive.com/emc-pstc@listserv.ieee.org/

Website:  https://ewh.ieee.org/soc/pses/
Instructions:  https://ewh.ieee.org/soc/pses/list.html (including how to 
unsubscribe)
List rules: 

Re: [PSES] AI & Regulatory Compliance

[MIKE SHERMAN]

 
 
  
   Regan —
   
  
    
   
  
   Good topic. Maybe we should also consider some techniques for sniffing out the BS from AI generated text, which appears to source whatever is on the web, right or wrong. 
   
   
   
  
   Note also that the new Machinery Regulation also addresses AI. See its Whereas clause 12 for some background, as I recall. Would LOVE to see presentations on this at ISPCE. 
   
  
    
   
  
   Mike Sherman 
   
  
   Sherman PSC LLC
   
   
   
On 10/06/2023 2:59 PM MST Regan Arndt  wrote:

   
 

   
 


Greetings fellow members, 
Our industry will not be immune to this new era of AI technology and want to bring up this topic again to obtain some more insight from the forum into: 
"How do you think AI will play a future role in our Regulatory compliance world?" 
  
I can see some real benefits to this when applying it to a risk assessment, EMC redesign including change out of critical components, & possibly to help support engineering judgments in lieu of testing (i.e. temperature modeling), etc.…though I think we would have to include the usage of AI as one of the potential risks in the risk assessment. lol 
UL touched base on this topic during their UL Innovations summit held in Fremont, CA yesterday. 
I know there is the ISO/IEC DTR 5469 in development, but this appears, I believe, to only address functional safety of the AI systems in question, and not addressing AI in normal compliance assessments for products under the category of laboratory, IT, audio, etc. 
Also, is there someone out there that will be speaking to this during the 2024 ISPCE in Chicago next May? Perhaps there should be an IEEE PSES technical committee created for this? 
Side note: Nordcloud puts out a good summary blog on this: https://nordcloud.com/blog/how-ai-can-help-you-obtain-regulatory-compliance/ 
There are other AI blogs out there in the ether but more tuned for the financial sector and other industries…. 
Looking forward to the discussion. 
  
P.S. 
I am now the Chair for the IEEE PSES Risk assessment technical committee and am looking for any volunteers to join our group. This topic will be one to discuss soon within our group. Please email me at regan.ar...@ieee.org or reganar...@gmail.com if you are interested in joining. 
Cheers! 
Regan Arndt 


   This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to All emc-pstc postings are archived and searchable on the web at: https://www.mail-archive.com/emc-pstc@listserv.ieee.org/ 
   Website:  https://ewh.ieee.org/soc/pses/ Instructions: https://ewh.ieee.org/soc/pses/list.html (including how to unsubscribe)List rules: https://ewh.ieee.org/soc/pses/listrules.html 
   For help, send mail to the list administrators:Mike Sherman at: msherma...@comcast.netRick Linford at: linf...@ieee.org 
   For policy questions, send mail to:Jim Bacher at:  j.bac...@ieee.org 

   To unsubscribe from the EMC-PSTC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1 
  
 


This message is from the IEEE Product Safety Engineering Society emc-pstc discussion list. To post a message to the list, send your e-mail to 
All emc-pstc postings are archived and searchable on the web at: https://www.mail-archive.com/emc-pstc@listserv.ieee.org/

Website:  https://ewh.ieee.org/soc/pses/ 
Instructions: https://ewh.ieee.org/soc/pses/list.html (including how to unsubscribe)
List rules: https://ewh.ieee.org/soc/pses/listrules.html

For help, send mail to the list administrators:
Mike Sherman at: msherma...@comcast.net
Rick Linford  at: linf...@ieee.org

For policy questions, send mail to:
Jim Bacher at:  j.bac...@ieee.org


 To unsubscribe from the EMC-PSTC list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=EMC-PSTC=1