Re: [agi] Iteratively Tweak and Test (e.g. MLP => KAN)

[immortal . discoveries]

Ye!! It sounds like John is saying a learning algorithm for learning 
algorithms? With an open-minded attitude always. He seems to highlight that it 
is about math and efficiency and accuracy of an AI/ its code, and that not only 
about thinking about them but also running them - perhaps streaming it using an 
AI brain. ? But doesn't it have to run the code to find out no?
--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T1af6c40307437a26-M4ed402f041c9840a00323496
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[James Bowery]

“Of all tyrannies, a tyranny sincerely exercised for the good of its
victims may be the most oppressive. It would be better to live under robber
barons than under omnipotent moral busybodies. The robber baron's cruelty
may sometimes sleep, his cupidity may at some point be satiated; but those
who torment us for our own good will torment us without end for they do so
with the approval of their own conscience.”
― C. S. Lewis

"c'est pire qu'un crime; c'est une faute" (it's worse than a crime; it's a
mistake).
― Charles Maurice de Talleyrand-Périgord

The Sam Altmans of the world are bound and determined to exercise tyranny
for the good of their victims -- which amplifies any mistakes in choosing a
world model selection criterion (ie: loss function).

Now, I'm not saying it is preferable that they exercise tyranny (as
opposed to, say, taking down civilization and starting over again); I'm
just being realistic.

PS:  Where's Ilya ?


On Sat, May 11, 2024 at 3:37 PM Keyvan M. Sadeghi <
keyvan.m.sade...@gmail.com> wrote:

> Anything other than lossless compression as Turing Test V2 is best called
>> a "Rutting Test" since it is all about suitors of capital displaying one's
>> prowess in a contest of bullshit.
>>
>
> If an email list on AGI that’s been going on for 20 years can’t devise a
> benchmark for AGI, wouldn’t history call them useless wankers? Do you
> want Altman to achieve it without you having a say?
> *Artificial General Intelligence List *
> / AGI / see discussions  +
> participants  +
> delivery options 
> Permalink
> 
>

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-Mdda4bbab8fcc9d8e55f5d587
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[Keyvan M. Sadeghi]

>
> Anything other than lossless compression as Turing Test V2 is best called
> a "Rutting Test" since it is all about suitors of capital displaying one's
> prowess in a contest of bullshit.
>

If an email list on AGI that’s been going on for 20 years can’t devise a
benchmark for AGI, wouldn’t history call them useless wankers? Do you want
Altman to achieve it without you having a say?

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-Mf6489f9bf28785f036297bd2
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[Keyvan M. Sadeghi]

>
> Your test is the opposite of objective and measurable. What if two high IQ
> people disagree if a robot acts like a human or not?
>
> Which IQ test? There are plenty of high IQ societies that will tell you
> your IQ is 180 as long as you pay the membership fee.
>
> What if I upload the same software to a Boston Dynamics robot dog or robot
> humanoid like Atlas, do you really think you will get the same answer?
>

Valid criticisms 

I wanted to start the conversation on a true benchmark, mission
accomplished! 

If a consensus is formed in this community, the results can be published in
AGI25?

Here’s some ideas for addressing the points Matt raised:

- Add a code postfix to ground the conditions
- E.g. Ruting Binet100_humanoid_SH
- Above example can mean:
  - The IQ test taken by the observing person is Stanford-Binet 100
questions in 24 minutes
  - The robot is in humanoid form, quality of the parts not important
  - The robot has the “Sight” and “Hearing” of the five human senses,
quality of the sensors not important
- The necessary and sufficient condition for passing the test is if only
one person validated by the IQ test confirms that the robot has human-like
behavior
- Someone can take a bribe and confirm a robot, that’s a fraudulent case of
passing the test, and can be contested by the scientific community
- A committee of trusted test takers can exist to take the test annually on
a live stage!

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-M534a366eeb945fdb092a6a13
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[Matt Mahoney]

Your test is the opposite of objective and measurable. What if two high IQ
people disagree if a robot acts like a human or not?

Which IQ test? There are plenty of high IQ societies that will tell you
your IQ is 180 as long as you pay the membership fee.

What if I upload the same software to a Boston Dynamics robot dog or robot
humanoid like Atlas, do you really think you will get the same answer?


On Sat, May 11, 2024, 7:59 AM Keyvan M. Sadeghi 
wrote:

> It’s different than Turing Test in that it’s measurable and not subject to
> interpretation. But it follows the same principle, that an agent’s behavior
> is ultimately what matters. It’s Turing Test V2.
> *Artificial General Intelligence List *
> / AGI / see discussions  +
> participants  +
> delivery options 
> Permalink
> 
>

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-M6a15dcd8d68f096880f8c3c8
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Iteratively Tweak and Test (e.g. MLP => KAN)

[Quan Tesla]

Seems like you have a plan John.

On Sat, May 11, 2024 at 7:56 PM Nanograte Knowledge Technologies <
nano...@live.com> wrote:

> Just a programmable thought...
>
> https://www.youtube.com/watch?v=k6H27YAKbDU
> 
> Explaining the Inner Product in Quantum Mechanics (Using Bras and Kets)
> 
> In this video I will explain the inner product in quantum mechanics using
> braket notation. Link to my braket video (Dirac Notation):
> https://www.youtube.com/watch?v=4XZBwVE8PuA_channel=NickHeumann If you
> enjoy my content, please consider checking out my Patreon!
> www.patreon.com/nickheumann Also, consider subscribing and following me
> on my ...
> www.youtube.com
>
> --
> *From:* John Rose 
> *Sent:* Saturday, 11 May 2024 17:00
> *To:* AGI 
> *Subject:* [agi] Iteratively Tweak and Test (e.g. MLP => KAN)
>
> Seems that software and more generalized mathematics should be discovering
> these new structures. If a system projects candidates into a test domain,
> abstracted, and wires them up for testing in a software host how would you
> narrow the search space of potential candidates? You’d need a more general
> mathematical model that has insight into efficiency projections. And the
> abstracted software may require a somewhat open-ended generalization
> capability for testing since the candidates would take on unknow forms.
>
> https://arxiv.org/abs/2404.19756
>
> Ballmer: "Developers, developers, developers, developers!"
> *Artificial General Intelligence List *
> / AGI / see discussions  +
> participants  +
> delivery options 
> Permalink
> 
>

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T1af6c40307437a26-M5d08914b5afc6e160bf47e1a
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Iteratively Tweak and Test (e.g. MLP => KAN)

[Nanograte Knowledge Technologies]

Just a programmable thought...

https://www.youtube.com/watch?v=k6H27YAKbDU
[https://www.bing.com/th?id=OVP._s1oBZ5dyEqQ3_Q-wZZNkAHgFo=Api]
Explaining the Inner Product in Quantum Mechanics (Using Bras and 
Kets)
In this video I will explain the inner product in quantum mechanics using 
braket notation. Link to my braket video (Dirac Notation): 
https://www.youtube.com/watch?v=4XZBwVE8PuA_channel=NickHeumann If you enjoy 
my content, please consider checking out my Patreon! 
www.patreon.com/nickheumann Also, consider subscribing and following me on my 
...
www.youtube.com


From: John Rose 
Sent: Saturday, 11 May 2024 17:00
To: AGI 
Subject: [agi] Iteratively Tweak and Test (e.g. MLP => KAN)

Seems that software and more generalized mathematics should be discovering 
these new structures. If a system projects candidates into a test domain, 
abstracted, and wires them up for testing in a software host how would you 
narrow the search space of potential candidates? You’d need a more general 
mathematical model that has insight into efficiency projections. And the 
abstracted software may require a somewhat open-ended generalization capability 
for testing since the candidates would take on unknow forms.

https://arxiv.org/abs/2404.19756

Ballmer: "Developers, developers, developers, developers!"
Artificial General Intelligence List / AGI / 
see discussions + 
participants + delivery 
options 
Permalink

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T1af6c40307437a26-Meb6c401b5b8880ae4c598084
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Hey, looks like the goertzel is hiring...

[John Rose]

On Wednesday, May 08, 2024, at 6:24 PM, Keyvan M. Sadeghi wrote:
>> Perhaps we need to sort out human condition issues that stem from human 
>> consciousness?
> 
> Exactly what we should do and what needs funding, but shitheads of the world 
> be funding wars. And Altman :))

If Jeremy Griffith’s explanation is correct it would invalidate some literature 
on the subject. I would like to see rebuttals.
And if he is right then models of the development of human intelligence may be 
affected.

I do see potential issues but it's worth entertaining to see if it "fixes" or 
alters some models. Though many individuals may not take notice since it could 
require a deep refactoring of their worldview But some may find comfort in 
this explanation regarding their own personal behavior and an understanding in 
observations of such behavior 
--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/Tb63883dd9d6b59cc-M6fae07260db6c30dcb0a97c0
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[James Bowery]

Anything other than lossless compression as Turing Test V2 is best called a
"Rutting Test" since it is all about suitors of capital displaying one's
prowess in a contest of bullshit.

On Sat, May 11, 2024 at 6:59 AM Keyvan M. Sadeghi <
keyvan.m.sade...@gmail.com> wrote:

> It’s different than Turing Test in that it’s measurable and not subject to
> interpretation. But it follows the same principle, that an agent’s behavior
> is ultimately what matters. It’s Turing Test V2.
> *Artificial General Intelligence List *
> / AGI / see discussions  +
> participants  +
> delivery options 
> Permalink
> 
>

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-M8377f8b3f36a06f85afc3716
Delivery options: https://agi.topicbox.com/groups/agi/subscription

[agi] Iteratively Tweak and Test (e.g. MLP => KAN)

[John Rose]

Seems that software and more generalized mathematics should be discovering 
these new structures. If a system projects candidates into a test domain, 
abstracted, and wires them up for testing in a software host how would you 
narrow the search space of potential candidates? You’d need a more general 
mathematical model that has insight into efficiency projections. And the 
abstracted software may require a somewhat open-ended generalization capability 
for testing since the candidates would take on unknow forms.

https://arxiv.org/abs/2404.19756

Ballmer: "Developers, developers, developers, developers!"
--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T1af6c40307437a26-M8b051de20c2a71345de3edf1
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[Keyvan M. Sadeghi]

It’s different than Turing Test in that it’s measurable and not subject to
interpretation. But it follows the same principle, that an agent’s behavior
is ultimately what matters. It’s Turing Test V2.

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-M0841709f213990d0960f613c
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] Ruting Test of AGI

[Keyvan M. Sadeghi]

>
> An LLM has human like behavior.  Does it pass the Ruting test? How is this
> different from the Turing test?
>

The instructions are clear, one should upload the code in a robot body, and
let it act in the real world. Then a high IQ human observer can confirm
whether the behavior is human-like or not.

--
Artificial General Intelligence List: AGI
Permalink: 
https://agi.topicbox.com/groups/agi/T244a8630dc835f49-Mb3f256d5dcb7288784e8b408
Delivery options: https://agi.topicbox.com/groups/agi/subscription

Re: [agi] α, αGproton, Combinatorial Hierarchy, Computational Irreducibility and other things that just don't matter to reaching AGI

[Quan Tesla]

Thanks for the references Rob. I'll be sure to pay the links a proper
visit. Yes, De Bono was on every consultant's lips for a while.

Not corporate, but in specialist operational training for the military.
This included doctrine, drills, deployment and R in counter-insurgency
warfare.

I appreciate your views on quantum bastardization. In my case, I
continually test my work against comparative, industry-standard frameworks
and "methodologies". Not many SSMs around though.

I'm also taking your point about technical AGI specifics on board. My
contention is that via my method the MMI for knowledge engineering has been
completed and extensively tested in the field for more than 10 years, on
commercisl projects in the public and private sectors. It think it's ready
for automation. This is where the gifted developers would feature.

This state of completion should then settle ongoing disputes around
ambiguity, nestedness, hierarchy, and so forth. I'm not claiming
perfection, but the work's been done, and well done.

 I've been extracting heuristics and axioms from the resultant BOK. One
such being a 6x6 matrix for probability-based, holistic-systems
specification. I think it's Cox that'll tell us that this feature would
satisfy the definition for the method being a quantum-enabled system.

Why carry on reinventing the wheel because it wasn't invented in one's
backyard? In general, I just find such reasoning suboptimal.

My SSM's approach is dedicated to any system specification being mostly
driven by core systems, as an inside-out (atomic) focus. That's the closest
to the standard model we can probably get.

The quest for including functionality for entanglement and quantum gravity
is now on. My hypothesis is that the Po1 equation would hold the key to
evolutionary functionality. I refer to this mechanism as the triple-alpha1
process.

Could such a system generate its own light energy? Theoretically, yes. AGI
would be energy self generative.

The fractal specification method embraces quantum coherence. Thus,
normalizing components as pure, polymorphic objects. Further, being
inherently driven by meaningfulness, in the sense of emergence (outcomes
management).

Last, satisfying a clinical requirement in providing 1-step mutation
functionality. This translates into tracable knowledge mutation (as
evolutionary systems mimicking NDA).

I've been investigating if the matured  diagrams could be converted into
rich knowledge graphs. Given the method and output in adherence to IEEE
compliance, I see no reason why this cannot be done. I see a fit.

The resultant structure of systems information is standardized in a common,
symbolic language, context dependent, content independent, robust, and
scalable.

I'm more purist epistomologist today than bandwagoneer. Hence, I still
integrate the new with the existing. I imagine my methodology as a proper
reasoning and decision-making engine within a version of AGI. Perhaps, in
the role of a co-enabler of human2machine consciousness. Those algorithms
are doable (credit to Penrose and Hamaroff).

The prof mentor/friend and I have been integrating (theoretically) my
method with KIM, their statistical knowledge engine. That introduces the
4x4 PDCA (Plan, Do, Check, Act) matrix.

These are but components for functional AGI. We have no plan to actually
bring an AGI version to "life" yet, which places me outside the general
competitor ring and fully independent. No doubt though, we're busy
designing an AGI version, independently of each other.

A few white papers (with definitions and references), and research results
of mine (using the diagramming "blocks, and arrows" method - aka "Essence")
and supporting industry-integrated architectural frameworks can be
viewed/downloaded on Researchgate. Might be worth a quick browse?  Happy to
engage in further discussions, without divulging deeper algorithms.

The main search string would be: Robert Benjamin and tacit knowledge
engineering

Good chat!

On Sat, May 11, 2024, 09:39 Rob Freeman  wrote:

> In the corporate training domain, you must have come across Edward de
> Bono? I recall he also focuses on discontinuous change and novelty.
>
> Certainly I would say there is broad scope for the application of,
> broadly quantum flavoured, AI based insights about meaning in broader
> society. Not just project management. But not knowing how your
> "Essence" works, I can't comment how much that coincides with what I
> see.
>
> There's a lot of woo woo which surrounds quantum, so I try to use
> analogies sparingly. But for ways to present it, you might look at Bob
> Coecke's books. I believe he has invented a whole visual,
> diagrammatic, system for talking about quantum systems. He is proud of
> having used it to teach high school students. The best reference for
> that might be his book "Picturing Quantum Processes".
>
> Thanks for your interest in reading more about the solutions I see. I
> guess I've been lazy in not putting out more formal