We have already built various superhuman devices - superhuman diggers, lifters, throwers, calculators and so on; so my first question is what you intend to mean by superhuman intelligence? It is fairly clear that current AIs are not superhuman in anything much as independent intelligences but can only function as tools under close human supervision.

So I conclude from this that intelligence as implemented in current AI is nothing more than a tool that can be used or misused by humans but does nothing of itself. So it can be great for looking for patterns in data, astronomical or medical or whatever, but is devoid of what might be called 'common-sense'.

Are you building a better tool, or a real super-human?

If somebody finds it problematic, I am open to any solution. — Carlo Roosen

The solution is we delete your posts and ban you.

Anyhow, what happened in the the other thread was that some posters who know about Kant tried to inform you about your misunderstandings concerning his philosophy. Your response there and here is to pretend that they were wrong to do so. Plus, honestly, you've shown no evidence that I'm aware of that you know anything interesting about anything at all, including AI. If you provide that evidence and don't engage in any self-promotion whatsoever from now on, you can stay. Otherwise, you may not be contributing to the site and may not be allowed to stay.

I am sorry, but I find this comment unacceptable. Why should I show evidence of knowing something interesting? Nobody here has to prove anything. People talk about their dog, why can't I talk about my AI project? If it is not interesting what I write, people would ignore the post, but reactions keep coming.

Just a second ago I got a comment from RussellA on my posts. This comment is respectful and does resonate with what I am saying. Also the people I disagree with got a respectful reply from me. On the other hand, not all feedback I got was respectful.

Just because I have a different opinion, that is no reason to ban me. And I do not self-promote, that is what I am explaining. I want to discuss a certain topic, and for that reason I have to explain a bit more context. That this may lead to some exposure, I am aware of and I do everything to avoid that.

Also, look at this email conversation I had with Jamal earlier this week. I replace his real name with [Jamal] for privacy reasons.

On Thu, Oct 3, 2024 at 15:25 Carlo Roosen wrote:
Hi [Jamal],
This is a draft of the first post I plan to write on the forum. Please also look at the note at the end.

It’s good. It should produce a lively discussion.
[Jamal]

The "note at the end" later got objections, so I removed it. It contained a reference to my website.

I am working on a real super-human intelligence. It is really about a new kind of software architecture. The key is that it develops its own internal language, I call it Babelspeak. That language has a high bandwidth and data rate. I expect it will be able to have a broader understanding than humans. See my earlier post.

You have not really illuminated me as to what you mean by intelligence, but I infer from what you say, that you mean language manipulation. In which case, the answer to my second question is that your SAI will still be a tool that is functionless apart from the non-super human operator. Disappointing, but not unexpected.

No, your conclusion is incorrect. The plan is, it will be an autonomously operating AI that could take over the world if it had bad intentions. That is my whole point of being here, to get a grip on that aspect.

The problem is, while many of you are engaging in a positive way, I get banned if I tell too much about my project, because it would be self-promotion. So be patient.

You can talk about your dog in the shoutbox if you like, or now that I've moved this to the lounge, you can relax and be less formal here. But if you want to start a thread in a front page category in future, it should be more substantial than this one. E.g. a discussion where you get into technical details about your AI project can go in "Science and Technology".

I was not even aware of front page topics.

I get banned if I tell too much about my project, because it would be self-promotion — Carlo Roosen

I think it should be clear that you can talk about your theory as much as you want. You just can't link to outside sources [edit: or clearly self promote in your posts]. We are simply asking you to contribute something substantial without appearing to promote your website or book or etc.

And where did I do that?

Edited to be more specific. You just got to the site, You've contributed nothing yet. Maybe after you've been involved for a while and shown yourself to be a valuable member, we can be more flexible on this. But if we get the impression your only reason to be here is to advertise "Babelspeak", you will likely be considered as being of negative value to this community.

That is nonsense. If I can't use the term Babelspeak, that is what this idea is all about. Look, I am just a developer alone in my room. I don't sell Babelspeak T-shirts. I just wanted te be completely transparant. If that is not how you like it you could just say so. But your text is unacceptable by any standard of respectful discussion:

"

Plus, honestly, you've shown no evidence that I'm aware of that you know anything interesting about anything at all, including AI. If you provide that evidence and don't engage in any self-promotion whatsoever from now on, you can stay. Otherwise, you may not be contributing to the site and may not be allowed to stay. — Baden

"

I suppose it was a bit blunt.

What was blunt? And I still don't understand what I am allowed to say or not. What I wrote is that, in trying to be transparant, there might be a time that anything I write on this topic could be interpreted as self-promotion, because there might come a time that I do need promotion. And I asked anyone if they feel it this way, to let me know so I could adapt. Why is that self-promotion?

Saying the name of your website "Babelspeak" gives no information about your ideas, it just advertises your website. It doesn't matter what your Super Human AI is called, you can simply converse here about its design, background philosophy and so on for now, i.e. the substance not the name. That's the last word on it. And best of luck with your project.

Edit: The name and link can remain on your profile though.

You are wrong. Babelspeak is what I call the internal language this AI is going to develop. So it is the core of my idea.

I'd rather unpublish my website.

I wrote my ideas here https://thephilosophyforum.com/discussion/15491/i-am-building-an-ai-with-super-human-intelligence.

Just because I have a different opinion, that is no reason to ban me. — Carlo Roosen

This is not the reason banning is being considered.

The reason is from the Site Guidelines:

Advertisers, spammers, self-promoters: No links to personal websites. Instant deletion of post followed by a potential ban.

That is why I contacted the admin before I sent my first post. And I made agreements with him about what was allowed. Now it seems that I cannot mention "Babelspeak". That is impossible, it is the core of what I want to talk about.

And Baden literally said that

The solution is we delete your posts and ban you.
What happened in the the other thread was that some posters who know about Kant tried to inform you about your misunderstandings concerning his philosophy. Your response there and here is to pretend that they were wrong to do so. — Baden

So it was - and still is - not clear what I did wrong and how I should change my behavior.

Is it ok if I refer to "Bubblespeak" in the future? I don't mind.
I can unpublish my website, I don't mind.

But people are asking for more information about my project. Even in this thread, unenlightened was asking more. How should I handle that? Earlier I said, look at my profile, you'll find a link. Mentioning that was allowed, according to jamal. I said a few times Google "Babelspeak", maybe that I shouldn't do?

I am reasonable. But please give me a workable solution.

Is it ok if I refer to "Bubblespeak" in the future? I don't mind.
I can unpublish my website, I don't mind.

But people are asking for more information about my project. Even in this thread, unenlightened was asking more. How should I handle that? Earlier I said, look at my profile, you'll find a link. Mentioning that was allowed, according to jamal. I said a few times Google "Babelspeak", maybe that I shouldn't do?

I am reasonable. But please give me a workable solution. — Carlo Roosen

I think you came up with some workable solutions here. It would be a lot harder to make the argument that you're here for self-promotion if you didn't promote your website, and instead stuck to things like this encyclopedia page, or other sources aside from your website.

Basically if you use philosophy articles to explore the questions you're good, but if you use your website then it seems like you're here to promote your website rather than discuss the philosophical implications of your ideas.

The plan is, it will be an autonomously operating AI that could take over the world if it had bad intentions. — Carlo Roosen

Or good intentions, presumably. Isn't that what humans do as best they can anyway - take control of the world to adapt it to their requirements? And in building said device, I assume you in particular want to take over the world with good intentions?

So do you think that intention can arise from language processing, or do you have another system in mind that will fire up autonomous action and the language aspect will again be the tool of this autonomous intentional system?

Incidentally, just don't put links to your work, don't talk about fund-raising, and talk sensibly about what you are interested in, and there will be no problems. This is an ad free site paid for by the owner and members' donations, and people are sensitive about it. If you are not exploiting us or disrupting us, there will be no bother. Move on from the rules and warnings, and have the interesting discussion, and things will be fine, apart from being called an idiot now and then ... that happens to everyone.

Are you a seeker of truth? — Carlo Roosen

I seek to understand the limits of the uses, misuses & abuses of 'truth'.

[M]y challenge is going to be, to get answers that are applicable to reality.

Philosophy begns by reflecting on foundational questions which we do not know how to answer (yet), how to validly raise such questions without begging them and how to reformulate them ad infinitum or until they become empirically decidable (i.e. "applicable to reality"). As I understand it, the existential challenge is to adaptively make sense of reality in general and thereby strive to live accordingly. Seeking "to get answers" leads merely to (dogmatic, sophistical) religion, Carlo, and is not, imo, philosophy. You've come to the wrong site for "answers".

Should humanity even want to build this SHAI?

This pseudo-question reminds me of silly maxims like "if man was meant to fly, he would have wings". Besides, "humanity" is a concept, sir, and not an individual who can "want". Instead of idling here like a child pondering on whether or not s/he "should want" to learn from (an) adult(s), build a "superhuman artificial intelligence" first so that then you/we can discuss with it what questions it/we should by asking.. :brow:

Bruh, just call it "my idea", I get you have pride in your work, I used to :lol:

but this is just all so trivial. you seem like a very intelligent person with much to contribute. things can be pretty strict around here (outside of the Lounge or Shoutbox), as it should be, so just bear that in mind and use terminology instead of familiarity. please. do it for me. As a programmer myself, I can't wait to discuss programming philosophy with you and to share my (granted much more base) insights with you. :up:

Granted it's nothing to do with AI, indirect or otherwise. (I think the only thing I'm working on remotely close is say, searching a database for the user's past inputs or whatever the project is currently and suggesting "Would you like to add (another) XYZ template here?", etc. But there are fundamental methodologies and object models/principles I feel will coincide with yours in at least some minor way.

Let me try to simplify this as best as I can:

Your idea (see, not so hard) attempts to not improve but fundamentally change how AI functions, due to it being an entirely new method utilizing several processing "avenues" as opposed to a single one, or so you say. Do you have experience with behind the scenes LLM coding? I for one, and I'm sure many, even as a programmer, find the idea a bit daunting and hard to grasp. I, as a simple programmer, would imagine, OK, there's a database with every single word, sentence, or entry that has ever been fed into it, which crosschecks a secondary database (if not several) to ensure appropriate and above all accurate responses are met. I imagine there's something of a percentage number for many, many "qualities" of a given phrase or answer about to be output, including, "tone", "aggression", "easygoingness/casualness", "accuracy", "friendliness", etc. That's even before it begins to compile and understand phrases such as "How are you doing?". I would image the first word "how" would indicate a desire to explain something, there are simple non-AI scripts that analyze sentences to determine: subject, predicate, intent, verbage, etc. That isn't difficult. So from there, a search is performed. And said results are compiled in human expression and terms such as "The answer to that is", "therefore", "but also", etc.

So, how is it you are improving on that base (crude if not outright poor explanation on my part) methodolgy? What do these alternate "avenues" provide? You mentioned something about it taking in sensory information to perhaps remind the user "don't forget your keys". How does it know the person has keys? How does it know they are in need of a reminder? How will it determine who the other people are around the environment (you mentioned the idea of telling it to "don't do that when my mother is around, etc")?

The hard to grasp part I find difficult is assigning multiple complex variables and if statements (requirements) for advanced queries. A simple query, say, "If there are 12 rooms on every floor of a building, and there are 15 floors, how many rooms are in the building?" is something I could probably even manage myself. From what I understand, you must already have this base functionality nailed down to a tee. So, where do we go from there and to what end?

Your idea (see, not so hard) — Outlander

He thanks for your interest. For me it is perfectly fine to talk about 'my idea'. It is not about pride, believe me, I passed that station 5 years ago.

I am an experienced programmer in many different fields. My current job is in robotics, food industry. I have experience with neural nets and with the math behind it, but it is the first time I work from a core developers perspective. Generally such a transition is not a problem for me.

"My idea" is about trying to make everything inside the AI independent of human concepts. Current chess programs use their neural net to rate board positions, but their minimax tree search is still written by humans. As you said correctly, LLM's use all kinds of human written preprocessing. My belief is that this is the limiting factor for true intelligence. I could be right, I could be wrong, I don't care. It is the direction I decided to go.

The solution I have in mind, and one that I am currently writing at "tic tac toe" level, is that halfway the neural net I will have an abstraction layer with a smaller number of nodes. Then there is a decision algorithm that is fully governed by other neural nets that decides whether the information continues the normal route or goes to another module.

Other modules can store the data, or modify it, or keep it circulating. All governed by neural nets. This is where I can experiment whatever I like.

Is this a crazy idea? Yes it is. Will it work? In my experience, I have invented many crazy things and all of them worked. Not all of them were successful in the marketplace. Three software patents on my name are underway, if that counts.

But all this is not what I want to discuss here, on a philosophy forum. The whole thing has many other aspects I do want to understand better. One of them is: will this AI contribute to human self-destruction, or will it help avoid it? To understand this better, I think it important to see the information flow between modules as a language, as an analogy to human thinking. Only here it is that my term "bubblespeak" comes in. (name altered in harmony with the new guidelines. I also deactivated the website after my own proposal).

Maybe you do not see the need for this philosophical discussion. It is not totally clear for me either. I need time to find that out. That is part of the journey I am here for.

In the meantime it seems a lot of people are enjoying the discussions, and several do agree with me on a deep level. It is only the moderators that keep changing the rules. I did not include links to my website. I mentioned the name of the language, but without url. I suggested a few times to google the name, maybe I shouldn't have done that. On the other hand I was allowed to point them to my personal page that had a link to my website - to me that is the same thing. I went through many discussions with several people who were not cool.

I am good. I feel happy here, I like talking about my idea after several years of just thinking about it. I want to obey any rule that they put in place to keep this place healthy for everybody.

Incidentally, just don't put links to your work, don't talk about fund-raising, and talk sensibly about what you are interested in, and there will be no problems. — unenlightened

For the record, I did NOT put links to my website. I mentioned fundraising only to tell people that if they thought I was self-promoting, I was happy to adjust anything they didn't like.

For philosophers I would expect people to read better. Never mind, I am good. My website is offline. I will use the term bubblespeak whenever I want to address the internal language that my AI will develop.

And for your and other peoples information, if somebody wants to read more about the technical side of my plans, read above post.

Seeking "to get answers" leads merely to (dogmatic, sophistical) religion, Carlo, and is not, imo, philosophy. — 180 Proof

This viewpoint has changed over different time periods. Right now I have the feeling that philosophic viewpoints have dogmatic aspects. People on the forum said it is not allowed to talk about fundamental reality...

Your website can be online, it's the promotion of it that's forbidden on the front page. You can keep the website in your profile too! If you look at my profile I have my twitter linked.

One of the things here is -- you're not the first person in the world to have thought about a Super Artificial Intelligence. Especially amongst philosopher-enthusiasts. We are challenging your ideas, in the spirit of philosophy, and it seems you want to double down more than discuss.

No worries. I have my own odd obsessions that I try to avoid in talking to others, too, because I've found that hearing others' viewpoints helps me more than "preaching the word"

And, if you want to "preach the word", at least on a philosophy forum, you ought use philosophy resources. It's not like philosophers decided to stop thinking this whole time, so you could learn something from them. As I noted, they've even been thinking about AI. The thing you're supposedly interested in.

People on the forum said it is not allowed to talk about fundamental reality... — Carlo Roosen

It's allowed, though I'm at least skeptical of the notion.

Others have pointed out how your book has expressed ideas that philosophers have alread thought over.

"What is Fundamental Reality?" would at least be a question to explore.

Talking about "fundamental reality" like you're the one who knows the AI-Human system will SEE THE REAL -- well, philosophers have thought about this before, and this is part of the resistance you are receiving.

Did you read the SEP article I linked?

But there was NO mention of my website in my post, I am telling that for the 100th time. The name bubblespeak is a name for the language that my AI will develop. That is all I wrote. (the name changed conform the new rules) [edit] a website must have .com or .org , I didn't add that.

Some persons are able to put 2 and 2 together in a way that doesn't make sense to others. I'd call this sense-making: Your post had NO mention of your website, but we took a peek at your website before responding and saw how much of it was like your posts.

Changing the name is a good sign, IMO. It means you're not here to simply have the name repeated to make it more popular on the 'net.

Did you read the SEP article I linked?

The problem is, I discussed it with jamal first, he approved. Then a moderator deleted my post, I had the discussion again. Then a second moderator changed another post and called me things I wouldn't call my dog.

From my first email to jamal I have been very clear in explaining why I am here, that it might be controversial and maybe not entirely in line with what this website is meant for. I agreed to every rule I was given, gave jamal the content of the website and my proposal for the first post. I did everything I could.

As for the content of my (now offline) website, see it as sketches I made for things I want to discuss here. Because that's what they are. I came here well prepared.

Just stop changing the rules and don't accuse me of things I didn't.

No, I did not read your SEP article yet. I will, a bit later.