Is it possible to build a system that first understands “right or wrong” before using probabilities?
1 Like
Hmm… When relying on neural networks, I think all current algorithms end up relying on probability.
It is possible to programmatically determine correctness and pass the results to the neural network, but I think this is limited to cases where correctness can be easily determined. For example, discarding only those that are clearly incorrect.
As a similar approach, there have been many attempts to connect multiple models and programs in series or parallel. RAG is one such example.
1 Like
I want to combine OSPF and BGP routing protocols with large models, using path optimization to replace the probabilistic thinking of large models. Currently, computer predictions are based on human values, but what if they are aligned with nature? Because I believe nature is the arbiter of truth and a hugely complex reactor—given any situation, it will produce a corresponding output. Are there any related research papers or algorithms in this direction?
1 Like
Hmm… idk but i found this:
https://www.mdpi.com/2071-1050/16/21/9239
I just joined.. First post. Excited to even be considering working with AI. Interested in Tokenization.
As to your philosophy on Nature being Aligned: I see that as a guiding star myself.
I have been thinking about Gottfried Wilhelm Leibniz and his Monad. While this philosophy may be considered a thought experiment in hindsight, Leibniz introduces Object and he also introduces Binary while being excited about the I Ching. Things I love too.
So, the obligatory introduction to a new site accomplished. So I see you have the same idea that I do! Has to be the right place for me!
So is it about ontology? Perhaps Digital Physics? Love the stuff.
As to your question: It is a huge domain such things are we consider! ChatGPT says “combining OSPF and BGP routing protocols with large models, using path optimization to replace probabilistic thinking, and aligning computer predictions with nature, this is a fascinating intersection of networking, AI, and bio-inspired computing.”
I say, I started a group here and hope to make Intelectual and Creative Friends. I too am following “Nature” and asking can we do these things differently while at the same time learning what “These Things” are.
I am following so I’d love to read more about your thoughts on this aspect. The Mind of Man is not Man-Made right?!
1 Like
I could be wrong, but I would try diving into these research areas…
- Symbolic, rule-based reasoning (DeepProbLog ( Maybe, recent paper), differentiable theorem proving, and neural-guided search)
- Subsymbolic, probabilistic inference (neural nets, embeddings, etc.)
Hope this helps you.
1 Like
Hi! Thank you so much for your response. I really appreciate that you understood what I was trying to say. I feel that we’re thinking about similar ideas from different angles — especially about aligning systems with Nature instead of pure statistics.
I’d love to talk more with you privately if you have time. Maybe we can help each other explore this direction further — philosophically and technically.
Could you send me a link to your group? Or let me know how to DM you here? 
2 Likes
ok ，thank ，i will read it
1 Like
This forum has a DM feature, but it is effectively disabled, so it is recommended to use Hugging Face Discord for active communication.
That’s just a recommendation, though; there’s no problem discussing things here.
By the way, mentions are possible.
mention test: @newone1
No one is discussing on Hugging Face Discord. May I ask which section can be used for discussion?
1 Like
Hmm… I’m not very familiar with the layout of HF Discord either…
There seems to be a channel dedicated to NLP that would be suitable for academic discussions…
The general channel is fine, but the flow is fast and we’ll probably lose track of the conversation.
1 Like
I just came to this software today, so I’m still learning the ropes.
I started a group. it looks like it functions as a kind of work-group.
To better respond to your original question: I often contrast deterministic systems—like those governed by encoders—with emergent or non-deterministic behavior, which I believe you’re approaching from the angle of probability.
Anyway, it’s your thread, Captain—lead on. :ChatGPT cleaned up my post.
1 Like
My train of thought is clear and extremely complex.Thank you for your response. I can see you’ve thought deeply about these topics over the years. At the moment, I’m mainly focused on building something a bit more structured and technical — a system that can gradually form reasoning patterns under constraints like cost, consequence, and limited energy, rather than just optimizing for correct outputs.
If you’ve ever explored systems that mix deterministic rule-based control with probabilistic exploration — especially in applied contexts — I’d be curious to hear your perspective on those.
Thanks again for sharing. Let’s see where this discussion leads.
1 Like
Ha! About Discord I have a Lifetime Ban!
Fun story.. I love an online game based on the royal game of Ur, I loved and played it so much that, I, having studied random data for three decades, saw that the difference between their Bot Levels of {easy, medium, hard and zoo animal } had their AI tuned partially by skewing probabilities of the dice. I thought hey, they don’t know what I know so I wrote them a dice function which is really fast and flat across the outcomes of {0,1,2,3,4} I mean the only way to get a tighter distribution would be to use a quantum Rand(n) function.
So I was really stoked to get over to their Discord HQ and give it to them so the game world be fair dice for all.
Well I was told by the guy in chat that he didn’t like my attitude for pointing out their dice Sucked! Naturally after the ban I realized I had exposed probability-corruption and I was banned for life for doing a good deed. As far as Discus goes; I don’t know what I am missing . But don’t be sad.. It is hard some days but I keep busy and keep on living 
So sorry folks, Discus is out for me. I do have my own forum anyone can join if we are violating the spirit of the Beginners forum here. You see as a layman who just can’t stop with the dynamical systems thinking, I am a bit of the proverbial Odd Duck. Then again bleeding edge AI, in what is obviously a Technology War AI centric combat, needs all it’s Odd Ducks it can recruit! Why look at @newone1 they are dedicated to a trajectory! I am dedicated to one as well.
Full disclosure here: I have never programmed on a GFX Card before last weekend and CUDA looks really good but for the simple work of developing things, OpenCL will run on my AMD card, Laptop and the NVIDIA all at the same time so, I can get the experience of not only AI designing but learning MPI programming (message passing interface) while having a good excuse to stretch the old fingers again and write OpenMP code too. Never done that before but I get it!
@newone1 Dude! just jump in on the coding. It develops the brain. I vowed to avoid interpreted languages many many moons ago but Python isn’t going anywhere and is employed in AI BIG-TIME time Everyone starts someplace. We have AI now to hold our hand and even let do the heavy lifting of code layout.
Anyway, I have pulled my last workstation out of storage and am making it my dedicated AI lab. Also ChatGPT pointed out that my Laptop has a builtin GFX processor with 6 compute units that can run OpenCL just fine. That was a surprise as I had thought built in GFX would be wimpy. Well, that will allow some development on Laptop and a remote unit to run some home AI to learn with!
So to be fair to @newone1 and to you @john6666 I’m just now exposing myself (legally) to Hugging Face because I know what is coming in this my proverbial time of life on my “Golden Pond” for this old “Odd Duck.”
So friends I am in learning mode and for those who read posts more then write posts this is some background on the basic idea of the question according to ChatGPT this morning.
Good morning, Friend Ernst03.
The terms OSPF and BGP refer to two major types of network routing protocols—tools computers use to find the best path for data to travel across networks. Here’s a breakdown in plain terms:
OSPF (Open Shortest Path First)
- Used within a single organization or system.
- Think of it like an internal GPS for a company’s network: it finds the shortest path through routers within that network.
- It’s deterministic: all routers have a shared map and calculate the best route the same way.
BGP (Border Gateway Protocol)
- Used between different organizations (e.g., between internet service providers or between large company networks).
- It’s like diplomacy between nations: routers decide how to route based on agreements, policies, and metrics.
- It’s more policy-driven and flexible, not just shortest path.
Why might someone bring up OSPF and BGP in the context of AI?
In the thread you’re referring to, the poster might be suggesting:
“What if we could use routing protocols—not just for network traffic—but to guide AI reasoning or decision-making?”
They’re possibly imagining a network of AI decisions where instead of probabilistic jumps (like in neural nets), the system chooses structured, optimized paths—like how OSPF or BGP find efficient routes in networks.
They may want routing logic to replace probabilistic logic in large models—treating ideas like nodes in a network, and choosing how to traverse them using deterministic or policy-based paths.
If this sounds a bit like trying to give a large language model a GPS of meaning, you’re probably right. It’s not a common idea—but it’s intriguing.
1 Like
Hi! I really appreciate your thoughtful reply. I’d be interested in continuing the conversation with you more in-depth — where can I find your forum or how can I reach you? I’d love to discuss more about AI, reasoning systems, and the routing-based approaches. Looking forward to learning together.
1 Like
I am seriously a beginner. Thought to come here and check things out.
I will be exploring how language is converted to tokens in time.
As for projects I have a paper I’m writing so I’m needing to finish that before I start other things.
Your idea is advanced for my study. I’m much more boolean at this time then high level language.
I am curious what you do @newone1 . Is this your work or are you exploring?