[SystemSafety] New IET report

[Peter Bernard Ladkin]

On 2021-08-05 17:08 , Derek M Jones wrote:
> Peter,
> 
>> (4) develop a new standard for functional safety of programmable safety-related systems; and
> 
> Why is a new standard needed?  The conclusion states the reason
> very clearly (i.e., the data is needed to apply the existing
> standards is not available).

Aha. "AI changes everything" -> "we need new standards". A very unimpressive argument.

NASA was flying aircraft -- very large aircraft, an MD-11, as well as military aircraft, the NF-15B 
-- with real people on board, under FBW control of *dynamic* DLNNs, a quarter century ago.

The V&V for those flights did not say "AI changes everything; therefore our V&V doesn't apply; we 
need new V&V". To the contrary, the analysis was carefully done to demonstrate V&V for these radical 
systems according to NASA's then-current V&V standards. This is all well documented. You can 
download the papers, and buy the books on Amazon.

There is no mention of any of this pioneering NASA work in the report.

A view more in line with the views of people I know working intensively on this matter (many or most 
of whom were also ignorant of the NASA work) is that V&V criteria for "AI systems" can be built onto 
existing functional safety standards. They may or may not be right, but in any case this approach 
contradicts the IET-report claim.

PBL

Prof. Peter Bernard Ladkin, Bielefeld, Germany
ClaireTheWhiteRabbit RIP
Tel+msg +49 (0)521 880 7319  www.rvs-bi.de





-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature
Type: application/pgp-signature
Size: 840 bytes
Desc: OpenPGP digital signature
URL: <https://lists.techfak.uni-bielefeld.de/pipermail/systemsafety/attachments/20210805/c6052b97/attachment.sig>