[ExI] Google’s Willow Quantum Chip: Proof of the Multiverse?

[BillK]

On Tue, 11 Nov 2025 at 17:18, Jason Resch via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> I had a similar conversation with an AI on the topic of many-worlds. At
> the end of the conversation the AI was 99.99% sure many-worlds was correct:
>
> https://docs.google.com/document/d/1i15TT76PMYlOdMO76O_TYPML2ocHrlzYDdj2fPkguH0/edit?usp=sharing
> This either shows that many-worlds is most probably true, or that current
> AIs are so swayable that we shouldn't put much stock in what they say about
> controversial subjects.
> Jason
> _______________________________________________
>


Yes, AIs have a strong tendency to flatter users and agree with any user
suggestions. As Adrian pointed out, the words used and how the questions
are asked also have a big effect on the answers provided.
There is also the problem of loss of context during long conversations. One
or two questions get a good response, but many more,  and the AI starts to
forget what it was talking about.  :)
See:   <https://arxiv.org/pdf/2505.06120>
9 May 2025     LLMS GET LOST IN MULTI-TURN CONVERSATION

BillK
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20251111/6479dd77/attachment.htm>