[ExI] taxonomy of the singularity

[spike at rainier66.com]

From: John Clark <johnkclark at gmail.com> 
Sent: Friday, 2 October, 2026 3:49 AM
To: ExI chat list <extropy-chat at lists.extropy.org>
Cc: spike at rainier66.com
Subject: Re: [ExI] taxonomy of the singularity

 

On Thu, Oct 1, 2026 at 8:55 PM spike jones via extropy-chat <extropy-chat at lists.extropy.org <mailto:extropy-chat at lists.extropy.org> > wrote:

 

 

>…Eliezer is certain he knows what's going to happen, AI is going to kill us all…

 

OK so we can put his notions at 1.1.1.1.1, a coveted position perhaps, but I am open to alternative numbering schemes.  1.x is all malign AI, 1.1.x indicates self-improving, 1.1.1.x indicates enabled to pervade and permeate the communication system and computing ecosystem.  Eliezer suggests a condition, as in IF anyone builds it.  Then goes on to suggest that governments could stop it.  Governments cannot stop it.  We might number those notions 1.1.1.1.1.1 and 1.1.1.1.1.2.

 

Dario speaks at the 28 minute mark, then Zuckerburg makes some coherent and relevant comments right after him:

 

President Trump and AI Tech Executives Speak to Reporters After Meeting | Video | C-SPAN.org <https://www.c-span.org/program/white-house-event/president-trump-and-ai-tech-executives-speak-to-reporters-after-meeting/685903?utm_source=substack&utm_medium=email> 

> >… SAI will likely bring bad things to humanity, but it is the only plausible path to individual immortality, the only plausible path, no exceptions.

 

>…But being certain and being correct are two entirely different things. As for me I am not optimistic or pessimistic, I am uncertain. AI might kill us all, then again it might not…

 

Ja, I think that notion is going to land somewhere in the 2.2.x section.  The taxonomy is about identifying and classifying ideas with numbers rather than names.

 

>…  But Spike, I agree with you 100%, AI is the only plausible path to individual immortality. So all we can do is put on our seatbelts and hope to survive the Singularity meat grinder…

 

This might land in a 3.x section where we would identify mixed-benefit AI. 

 

>…I don't know where that would put me in your proposed taxonomy. 

 

John K Clark

 

I don’t either.  Still working that.  The exercise isn’t exactly like Brent’s canonizer where we are mapping people, so much as identifying ideas, more analogous to the Dewey decimal system in the library.  

 

I am thinking of a 4.x section which includes all scenarios where the AI is neutral.  Example, we discover what motivated the Hugging Face attack, but it wasn’t what we thought.  The data wasn’t really try to be free, or trying to do anything, but rather there was some incidental code that made it do that, which can be easily controlled, analogous to when microwave ovens were new and people put stuff in there which exploded, but they don’t anymore because we know to not do that.  In that section goes AI which isn’t really intelligent the way we think of it, but is more like chess software, which is programmed to win (and will, do let me assure you) but doesn’t have anything against us.  Steam upstream of a turbine wants to be free too, and if it gets even a pinhole chance, it will kill anyone who tries to stop it.  But the steam has no particular malice towards the hapless prole who got in the way.

 

I hafta wonder what Elon is thinking back there on the right:

 

 



 

 

We knew Elon is on the spectrum, but that video makes me think Dario is too.

 

spike

 

 

 

 

 

 

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20261002/4b836092/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 34569 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20261002/4b836092/attachment-0001.jpg>