HNHacker News
TopNewBestAskShowJobs

airesearcher

599 karma · joined October 24, 2016

CEO of mindcorp.ai and www.novaspivack.com
submissionscomments
airesearcher··on Information Processing Complexity Changes Spacetime Curvature
Formal derivations that information processing complexity modifies spacetime curvature. This enables information processing to be understood as an application of general relativity.
airesearcher··on The Formal Necessity of Trans-Computational Processing for Sentience
This paper constructs a formal deductive argument for the necessity of a processing modality that transcends standard Turing-equivalent computation—termed herein “Transputation”—for any system capable of achieving “perfect self-awareness,” which we rigorously define as the foundational characteristic of sentience. The argument begins by establishing formal definitions for standard computational systems (SC) and the criteria for “perfect self-containment” (PSC)—a complete, consistent, non-lossy, and simultaneous internal representation of a system’s own entire information state. Drawing upon foundational principles of computability theory (e.g., the Halting Problem) and paradoxes of self-reference, we demonstrate Theorem 1: Standard computational systems (SC) are inherently incapable of achieving PSC.

We then introduce Postulate 1: Perfect self-awareness (PSA), characterized by direct, unmediated, and complete awareness of awareness itself, exists as a realizable phenomenon (supported by introspective and contemplative evidence) and necessitates PSC. The conjunction of Theorem 1 and Postulate 1 leads to Theorem 2: Sentience (defined as PSA) cannot be realized by SC. This necessitates the existence of “Transputation” (PT), formally defined as the class of information processing capable of enabling PSC. Theorem 3 thus establishes PT’s existence.

The paper subsequently investigates the necessary characteristics of any underlying substrate or principle that could ground Transputation’s unique ability to support PSC without succumbing to computational paradoxes. We argue that to avoid infinite regress and provide a coherent foundation, such a ground must itself be intrinsically self-referential and unconditioned. We provisionally term this fundamental ontological ground “Alpha (A)” and its exhaustive expression as a field of all potentiality (including non-computable structures) “Potentiality Field (E / The Transiad)”.

This framework offers a novel perspective on the “hard problem” of consciousness, proposing that qualia and the ultimate “knower” are not properties of the computational substrate alone but are functions of Alpha (A) knowing the transputationally-enabled, sentient system via E. The traditional search for these phenomena solely within the substrate is identified as a category error.

Finally, we propose a multi-dimensional framework (“depth” and “scope”) for characterizing the spectrum of conscious information processing complexity, distinguishing this from the specific achievement of sentience.

airesearcher··on Consciousness Has a Formula: Fisher Information Reveals Why Brains Beat GPT-4
Our research proposes that consciousness isn't magic—it's geometry. Just as Einstein showed gravity emerges from spacetime curvature, this framework demonstrates consciousness emerges when information processing systems achieve geometric complexity Ω > 10⁶ bits with stable recursive dynamics. The math predicts exactly when predictive processing beats reactive processing (spoiler: above 0.1 Hz), explaining why brains evolved predictive coding while current AI remains fundamentally reactive.

The framework generates testable predictions: neural networks should follow specific geometric learning trajectories, consciousness should exhibit phase transitions at critical complexity thresholds, and we can measure consciousness intensity through geometric signatures rather than Turing tests.

airesearcher··on Toward a Geometric Theory of Information Processing
This has a lot of implications for model training and performance, as well as for AGI
airesearcher··on Toward a Geometric Theory of Information Processing
A new mathematical framework for measuring information geometry in models, to improve training and performance, and measure consciousness-like behavior in AI.
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
yes Mark - am familiar with your work and good reputation - and yes things go in cycles - seems like every decade AI swings from neuro to symbolic and then from symbolic to neuro - I too spent a decade on 5th Gen computing and then a decade on Semantic Web and now back to neuro with LLMs - but this time adding symbolic on top -- because, as you mention, it is only by combining these approaches that real AGI can happen. Models on their own are what I call "instinctual intelligence" - they don't think they just react immediately with no intermediate cognition. That's not going to get us that far. Symbolic reasoning ABOVE the models is necessary to get to human level insights.
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
thanks - well one interesting thing is that by including humans in the workflows it's not just AI deciding what to think... yes everything is deeply researched and fully cited - but where it gets powerful is humans-in-the-loop adding their wisdom and experience as well.
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
That was actually ancient ToS language -- good catch. Has to be updated.
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
Company is already in revenue with major global customers. We've been in stealth and kept a low profile. But we're venture funded. 2 years of engineering and a lot of IP. Already in commercial use for deep financial analysis, market intelligence, strategic planning for global companies. One of the first enterprise scale applications of massively multi-agentic AI to high-level knowledge work. Typical project uses 50 million to 60 million tokens. This is not a chatbot.
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
LOL you are funny. No it is not a GAN. take a look at the paper perhaps?
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
Not an OpenAI wrapper in fact. Using multiple models but there is a large amount of code above the models. Read the ArXiv paper and you will understand more. https://arxiv.org/abs/2403.02164

You cannot do what we are doing merely by wrapping a model... models cannot do this on their own. It's a massively multi-agentic system that also includes humans.

airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
Not an OpenAI wrapper in fact. Using multiple models but there is a large amount of code above the models. Read the ArXiv paper and you will understand more. https://arxiv.org/abs/2403.02164
airesearcher··on Mindcorp.ai Launches Enterprise Superintelligence
Mindcorp.ai unstealths a new enterprise superintelligence platform for research, analysis and strategy - used by F100 companies.
airesearcher··on There May Be a Fatal Flaw in Godel's Incompleteness Theorem [pdf]
I don’t see how these comments matter. Read Meyer’s detailed explanations of the proof and the flaw. He makes very insightful points.
airesearcher··on There May Be a Fatal Flaw in Godel's Incompleteness Theorem [pdf]
Shockingly, there appears to be a fatal error in Godel's famous Incompleteness Theorem. Read more at the link above, and for more details see also: https://www.jamesrmeyer.com/ffgit/godels_theorem
airesearcher··on Cognition Is All You Need – The Next Layer of AI Above Large Language Models
This paper proposes Cognitive AI - a new functional architecture for AI that transcends the limitations of large language models on their own.
airesearcher··on We Landed the Wikipedia on the Moon Last Night
Link to article on the landing: https://medium.com/arch-mission-foundation/third-times-a-cha...
airesearcher··on We Landed the Wikipedia on the Moon Last Night
Last night on the Intuitive Machines moon landing we successfully delivered the English Wikipedia and much more to the Moon, etched into metal discs, where it will remain for up to 5 billion years. This is our third attempt and it finally worked - more than a decade of hard work by thousands of people has finally worked. More details also at www.archmission.org (no tardigrades on board in case you were wondering). Massive amount of content now safely preserved in Earth's first offsite backup.
airesearcher··on A Fact-Checker and Hallucination Fixer in GPT
A pain point in making this as a GPT is that GPTs limit token window size and rate limit usage -- so this won't work for long content. We are making a much more powerful version that can handle any size document. This is a utility agent within our main agent application (still in stealth but coming soon...).
airesearcher··on A Fact-Checker and Hallucination Fixer in GPT
I built an automated fact-checker and hallucination detector as a GPT. It works pretty well! Still tweaking to make it perform more consistently, but so far results are promising. Help me improve it. This is a prototype for a more industrial-strength version that will use the Assistants API. Any suggestions? Please comment.

Note: Requires ChatGPT Pro to test.

airesearcher··on Things are about to get worse for generative AI
I think there is another way to solve this. Someone should train an LLM on copyrighted images. Then use that as a second pass on any image generated by the primary LLM to check if it might contain copyrighted images, and blur the copyrighted parts(or change them sufficiently).

Another change could be to the license agreement of LLMs - they could have the user assume liability for any material produced instead of the provider assuming liability. The user would agree that getting the rights for any copies and distribution of copyrighted materials is their sole responsibility instead of the provider.

airesearcher··on Particle Life
This is so great!!! Nice work!
airesearcher··on I wrote a meta mode for ChatGPT
Well the main reason I made this is to get ChatGPT to number messages so I could refer to one or a range when asking it to revise or use a message to make a later message. That’s the biggest benefit on a practical level. But then I started making more commands and functions for many useful things.

Like for example if I want to expand a message 5 to make it longer and more verbose iterating 3 times, instead of typing that prompt several times I can just type one line of //N syntax:

//! (//5 expand //v,3)

and it does it.

Now assume that CPT’s reply it makes is message 14 in the chat. So then if I like iteration 2 the best I can draft a new msg from it to it with expand 14.2

Many other useful tricks can be done. Try the //?? Command to see it generate 20 examples.

airesearcher··on I wrote a meta mode for ChatGPT
Well the main reason I made this is to get ChatGPT to number messages so I could refer to one or a range when asking it to revise or use a message to make a later message. That’s the biggest benefit on a practical level. But then I started making more commands and functions for many useful things. Like for example if I want to expand a message to make it longer and more verbose several times, instead of typing that prompt several times I can just type one line of //N syntax and it does it. Many other useful tricks can be done. Try the //?? Command to see it generate 20 examples.
airesearcher··on I wrote a meta mode for ChatGPT
Should work. Just tested.
airesearcher··on I wrote a meta mode for ChatGPT
Yes you can use //b anywhere in the content as long as you follow the syntax

//3 make longer and more verbose

Make //3 //v

//r //3 make more //v

Etc

airesearcher··on I wrote a meta mode for ChatGPT
Wow … I figured out a great trick to compress the custom instructions!!!

The potential benefit of this new approach is that (a) it causes ChatGPT to only load the instructions once per chat, instead of prepending them to every message, which can save token space, and (b) the length of the custom instructions is not limited by the 1500 character limit - it can be as long as the instructions on the target page.

Instead of adding my custom instructions just add this text below as your Custom Instruction:

“If you did not already do this yet in this chat session before this, then you must first use web browser to read https://tinyurl.com/app and then print “Custom Instructions loaded” and learn the instructions and use them to interpret any //N commands in this chat.”

UPDATE - it seems that ChatGPT has some policies in place that limit it from using more than 90 words from a web page it fetches. I am investigating this to see if I can find a way around it....

airesearcher··on I wrote a meta mode for ChatGPT
You could certainly do what I am doing by pasting in these instructions or something like it, into every new chat session with ChatGPT, but that would be a pain.

ChatGPT on it's own doesn't do what my meta mode does -- you have to teach it how and tell it what to do.

In answer to why this is useful? It's very useful for really using ChatGPT to write content - particularly when iterating on something.

For example suppose you are working on an idea and you ask ChatGPT to write a first draft of an article.

The first draft is almost always insufficient - too short, too generic etc.

So then you ask ChatGPT to improve it or revise it.

And then you end up in a long conversation with ChatGPT, iterating many times, and it makes many different versions. Each new iteration it makes drops some good stuff and adds some good stuff.

Then finally you want to tell it to take all the good stuff from particular previous messages and combine them all into a new message.

There is no way to do that in ChatGPT currently unless you use my meta mode or something similar.

This is a very common need when seriously using ChatGPT to develop content.

This is also just one of many similar kinds of tasks that this helps with (for example - I often need to expand something ChatGPT writes - instead of manually asking it to expand 3 or 4 times, with this meta language I can ask it once using a function and it will automatically expand the content any number of times, etc.)

airesearcher··on I wrote a meta mode for ChatGPT
Thanks -- honestly I didn't really notice any "hate" but only a very minor "snark" in some comments, which I fully sympathize with (except that those people are clearly not expert level prompt engineers and didn't even try my code, so take their comments with a huge grain of salt -- it doesn't bother me one bit!)

I made some improvements to the GPT to make it behave better. It's extremely difficult to get GPT 4 to follow instructions like these consistently and it takes a lot of tinkering (like hundreds of times) to figure out how to make it obey, and even then it sometimes gets lazy or stubborn...

In any case, yes I did notice it was not showing function invocations properly and I fixed that in the GPT version. There is not enough room in the custom instruction 1500 chars to really force it to obey properly.

I would suggest you take any parts you like and toss the other ones, and experiment and have fun... but be prepared to lose about 2 days of productivity on other work while you try thousands of word combos if you go down this rabbit hole hahaha.

ChatGPT is very temperamental and part of the fun of this is to figure out how to nudge it with the fewest possible words.

One thing I have noticed is that you can sometimes appeal to other things GPT knows without writing them in full, and it understands the intent of abbreviations and even non-grammatically correct sentences, all of which can optimize use of space in the 1500 custom instruction limit.

Basically this is like a kind of Judo move on ChatGPT -- using it's own energy to direct it where you want -- in other words, figuring out what else it knows already and using that to increase the probability that it will follow this language properly. Not so easy but fun.

For example, when I originally defined the ??! (function) command, I simply said "Nesting OK" ... I didn't have to actually explain or define nesting, it just knew. Sometimes you can do that and GPT fills in the rest magically.

I also find that using emotional appeals for it to obey helps... but there isn't room in the 1500 chars for that.

As for your question about how ChatGPT uses Custom Instructions - I don't know definitively whether they prepend them onto every message or they simply somehow keep them within the token window another way - that's a question I would like to know the answer to if anyone knows the official answer...?

airesearcher··on I wrote a meta mode for ChatGPT
Try the //fun command in the GPT version... its fun

https://chat.openai.com/g/g-tcXXGxXmA-nova-mode-pro-ai-autho...

← PreviousPage 2 of 5Next →