https://www.superintelligence.io/
Читать полностью…As an end user, I’ve encountered several challenges with ASI1. The system's outputs often appear unreliable — including hallucinations and frequent inaccuracies in mathematical reasoning. So far, I haven’t observed performance that inspires confidence across any specific domain.
From the outside, this creates a rather unfavorable impression of the system’s maturity and capability. As an end user, I also lack visibility into ongoing research efforts or promising developments that may be underway behind the scenes.
Based on my personal testing, I’m currently not convinced that the existing architecture is robust or future-proof. My impression, after multiple trials, is that the system appears overconfident in its responses while demonstrating limited adaptability or learning capability. If this trajectory continues, I worry it may lead to increasingly chaotic results rather than sustainable progress.
Do you have a group in Spanish?
Читать полностью…On that note, while some might consider it meaningless in the grand scheme, there is zero honesty about the burn. Nobody knows who is charge of it, why it has been delayed (or canceled, impossible to believe anything from admins at this point). The community is not listened across many issues and this has damaged the credibility
Читать полностью…@DTaylor this is an example of unclear communication when hundreds of millions of tokens are moved from treasury wallets in an on-chain world with trackable transparency.
In this case unfortunately it seems it had to be decoded by a single user while it could have received direct clarity from the team as its hard to mask such massive transactions, they dont just go unnoticed and cant be answered (When questioned) that they are just normal transfers for risk management or liquidity provision etc.
Would love to hear more about your full setup, as I am trying to work on my own, but every time I do so, it just feels like a total time sink with little to no results. You seem to have found a great setup using neo4js and I hope this would be it!
Читать полностью…First of all, I’d like to clarify that I’m a physicist, not a data scientist or AI engineer. I’ve been exploring AI tools on my own to try and build innovative projects.
As you mentioned, I was initially very excited by the Knowledge Graph (KG) feature of ASI:One. My original plan was to transform structured information from technical PDFs into a proper KG. I thought ASI:One could handle this automatically. However, when I asked relatively simple questions about the PDF I had uploaded, the answers I got were often irrelevant or incomplete.
Eventually, I had to manually construct triples — (subject, predicate, object) — with the help of ChatGPT, and then load them into a Neo4j graph database using Cypher queries for efficient querying and exploration. Honestly, after months of work, the results are very satisfying, and I now have a functional, domain-specific knowledge graph.
That said, I would have much preferred to build all of this within ASI:One, but in my experience, it wasn’t the most effective tool for handling large volumes of structured technical content — especially ASCII files (Python, HTML, Java, etc.), which other LLMs currently process better.
For context, I’ve been a strong supporter of the ecosystem: I’ve held AGIX and FET tokens since 2018 and I genuinely love the vision behind the project. I’m committed to porting a working agentic version of my project (and more to come) into the Agentverse, but for non-specialists like me, the learning curve and technical hurdles are still significant.
Again, this isn’t a criticism of the many great features and products developed by the alliance — just an honest reflection on the challenges I’ve faced given the specific nature of my work
It's been over half a year (long long time in crypto) since my last infographics and I'm planning to update it soon. The only AI project on my list that I can't find complete financial information about is ASI.
Читать полностью…No wonder the price is lagging compared to the rest of the market. Credibility has been lost and it keeps getting worse.
Good to have you here, hopefully you can bring solutions to all the issues that are holding ASI back.
I've personally tested ASI:one and while the KG and features are ok, I did not like uploading personal data to their hosted servers, whether those cloud servers be decentralized or not. I want control of my data and I don't want my data to be used for training. Thus, I did not see a reason to use it over my local RAG setup and I will wait to see how ASI:Create will treat data ownership. I'm a long LONG time follower of FET (since 2019?) because I always believed that we are headed towards AI, and FET felt like they were ahead of the time back then. Bought and sold some over the years, I came back post merge and when AI became THE narrative but was on the fence knowing the Fetch company was in the brink of bankruptcy. Also not a fan of merging and supply dilution, so I just recently came back with a bag when all dusts have settled, and hoping to see some actual use coming with ASI:Create, which is supposed to use a decentralized knowledge graph that I have been following since 2018.
I believe Niko is genuinely a great dev and I have only received great support him. I also like to think that ASI will follow through with most of their promises despite some setbacks. What I do hope to see looking forward is some more transparency on their business model and the tokenomics of the FET (soon ASI) token. I am a pure fundamentalist and will only hold tokens that generate positive earnings, or those that I believe eventually will. I'm glad to know Niko is working on making network numbers transparent, and I look forward to updating those numbers on my infographics when they become available.
Sidenote: you should consider reporting those network numbers on tokenterminal, as this is where me and many whales do their research.
Right, his tasks can be accomplished outside of ASI. And that's an issue for him because it sounds like he was hoping to be able to accomplish his tasks within the ASI suite, just as he would with Chatgpt.
I feel like you are basically implying that ASI One is trash for anything other than entering text prompts and receiving a generated text reply.
If so, why is the ASI project implying they are so ahead of openai?
It's like the communication and reality are so conflicting...
More or less. He just seems to be pushed back by lack of features on ASI:1, but for his use case (scientific project + LLM + KG) he can very easily test on a local set up with any graphdb, like neo4js and train his AI locally with ASCII files.
Читать полностью…I'm perplexed. If your project involves using a knowledge graph, what stops you from simply deploying a local RAG and use neo4js and uploading your graph in a hosted server? What is the requirement of using ASI:1 there specifically?
Читать полностью…If you want your marketing to success 1st you have to do is look for a quality CM who can create a good community. Our community is trash right now.
Читать полностью…Because again, credibility starts at that base-level.
Читать полностью…I think i already did send you information about OpenCog Hyperon and the MeTTa Programming lenguage. You could easily get deeply involved as it will push Neurosybolic and Metahypergraph AI, which will highly compliment the ASI:One development. This will also highly reduce hallucinations and accelerate the AI to AI communication.
I'm just slightly confused despite being so actively looking for solutions and get involved early on in what you are looking for, I just doubt you will find an answer or a solution, unless you directly go to the source i offered you.
ASI:One and it's ecosystem is growing updating and upgrading. It depends what you are looking for. If you are a customer, you likely will have to wait a few upgrades cycles. If you are a developer, you can get directly involved with it's creation. You can choose.
I also found it hard to get some answers to some questions as I want to promote ASI and it's ecosystem to the company I work for. It just seems .ost resources goes to hands on development and delivery.
Nonetheless on each partners Website Cudos; Fetch; SingularityNet and Oceanprotocol there is lot's of information aswell. They also have a Roadmap which you can take reference on what's coming.
Now the question is, which of the ones are you. A Developer or Programmer or a Customer. You have for the first two a link in your DM since 2 or 3 weeks, is it customer, then i recommend to look up their Roadmap
I’m pretty sure qwello will solve your problem. Question is… when is it coming out…
Читать полностью…Natuarally like all of you I also was concerned if not worried about Chat GPT agent release,
After doing quite a bit of research it turns out even it is not that good,
The same constraints apply to open ai if not more since ASI 1 is very agentic focused,
They clearly have plugin advantage in that they can speak directly to companies to enable payment or integration,
However this brings us back to the ASI USP,
They could potentially gain momentum once some hurdles are fixed,
However I believe the decentralised nature of ASI is the reason we will win,
Once llm operation and execution functions are solved,
ASI will still have more agents and a network of adapting interconnect agents,
FUD can come in different ways, but look a little deeper and you’ll realise though open ai is paying 100million,
They still haven’t solved some problems that ASI has,
I believe the intergratjon of ASI create, hyperon and the other tech stacks will bring problems of their own,
But if you look clearly,
You’ll see that we have a front facing product, and there is a team working on the other intergratjons in the background,
ASI remains revolutionary as long as Open Ai do not also decide to build a team to create their own Ai programming language and all the other features Singularity Net is working on,
The blind spot is not nice,
And I want to see asi create and its roadmap also but we can’t forget the effort the team is putting in ❤️
TRNR deal same applies, unclear media coms (SEC Reg FD) with respect to Humayuns fudder response statement about the deal not being OTC. Hope things improve moving forward and some follow up clarity is given to investors who deserved it
Читать полностью…A major de-risking move in case the required SG license does not materialize or becomes delayed, thus ensuring access to token liquidity throughout this process.
Читать полностью…Thanks for the clarification, really puts us in the context of a user trying to find the best currently available solution on the market for our needs. I happen to be a pharmacist who would also like to develop an AI chatbot based on verifiable knowledge in order to help me and my colleague assess patient drug profiles and day-to-day drug analysis. Currently, I'm on the same boat as you and started working on a local graphdb (blazegraph in my case) and trying to find a suitable solution to convert PDFs and online databases to json ld. I've tested unstructured but they no longer offer a free API, tried other solutions that I don't remember but the PDFs were just too complex.
Now I'm just getting started with neo4js and it seems like the msot complete solution.
Some projects have negative earnings in the$200M+ yet they still report it to tokenterminal. As a token holder, net earnings generated by the network and redistributed to the holder (through burns for instance) is the most important.
Читать полностью…This model is very similar to the top / mid / bot funnel strat. But I agree the pyramid collapses when infra & cred isn't there. And the key to cred is transparency.
Читать полностью…I am not involved with building ASI chat, and I do not think ASI should compete with openai, grok or any of those centralized LLMs. It's a losing battle and only part of the solution.
Instead, ASI should focus on developing decentralized neuro-symbolic AI, so using blockchain infra, decentralized knowledge graph to train agents and what not, which is what they seem to be doing right now.
Some of these solutions already exist so no need to reinvent the wheel, but to integrate those to make truly decentralized agents that share collective memory and reasoning.
So rather than just looking at the ASI LLM model and see whether it can compete with openai is missing the bigger picture here.
P.s. by the way, respectfully do you happen to be involved with building the ASI chat product?
Either way, what do you think the ASI project should do in order to better compete with openai with regard to the chat product?
He can plug in ASI-mini LLM through llamma if he wish, or just pretty much use any other LLM api out there
Читать полностью…I can't speak for Quentin, but it sounds like he's saying he can do it all with chatgpt, but can do none of it with ASI. So you are right, what's the point of ASI if you can't do anything with it. Is that the point you are making?
Читать полностью…Various social media refer back to the telegram group for further general communications, which are real-time thus a popular avenue to pose questions. But not always as popular in terms of receiving responses or answers. Sometimes the discord channel is advertised for somewhat deeper level of engagement, or for submitting official tickets for suggestions, technical support, etc, and, the discord also has sections for participating in various governance activities. The discord is much more organized than the telegram, but perhaps much slower paced as well. And the github channel(s) would be for the more technical participants who are sharing directly in the build effort.
So if a person wants to express a sentiment or is seeking a response to something that is not too overly software-technical, they'll usually find telegram is the most useful place. There are great moderators here. Like, excellent moderators that do a great job communicating on behalf of the company, but it is almost non-existent for a technical feature related question, like the one above, to actually be addressed in the chat, unless ben goertzel himself actually drops by to stir the pot of excitement and share some gems of info once in a blue moon. But nobody from the project teams actually interact on the telegram channel. Perhaps too much content not specific to their domain to sift through just to find concerns that they can address? Perhaps there are some fundamental barriers between people who are solely concerned with building AI and those who are more concerned with the financial aspect of the project to communicate effectively in the telegram chat style. Two different worlds. Communication is probably not very compatible.
Anyways, back to the point. The question I highlighted (that was posted just a few hours before my post) is not a financial related or token related question. It is a specific product usability question. With all the mixture of various topics being expressed in telegram, i would expect a question like his to just fall into the ether to be forever lost.
Tldr: perhaps discord would be a good place for his question to be addressed if it is not being addressed here in telegram?
Maybe the best answer is simply: soon?
I have almost no idea. I'm just a token holder watching the show 🙏
I so much agree with this. A good mobile dev team needs to be put together to make the mobile apps great while the web team can continue with the already great work.
Читать полностью…Understood. 🫡
Is there a formal feedback loop anywhere? If not, that could / should be a nice easy one to address.