Abstract

This is a prediction on AI and how it relates to the AutonomI network (previously known as the SAFE network). Some people will hate this, but I hope everyone at least takes notice.

Observation

AI/LLMs are here

AI or LLM or Neural nets are here. What is means, how close to AGI/SGI or whatever is irrelevant for this discussion.

We can talk to computers and they can talk back! We have a non biological partner, no matter how smart or dumb, that we can speak with. It’s here and getting better at communicating, better than any animal or other biological creature, it is communicating.

So with that behind us and no proclamation of advancement or doom, we are where we are. Things change from here, and change in dramatic ways. This we know.

Still we wish to define intelligence and consciousness etc. Not me. I just work with what I have and can see and interact with. It’s here.

This is coming in several forms, agentic, multi-modal, fine-tuned, RAG (Retrieval Augmented Generation), as well as differing model sizes and capabilities. The world is changing and that will not slow down.

Prediction

What we end up with, nobody knows, but what I strongly believe is:

• Direct hardware access will be their API, no OS level, not IOS not Android, but direct hardware access.
• Apps and browsers as we know them will be bypassed
• Possibly all, but at least most, human written code will be replaced by direct human to machine capabilities
• Files and directories or folders become antiques that only a select few refer to. Who wants to search manually through millions of files for knowledge when you can just ask
• Humanoid robots will enable the physical world to be an extension of our computer and likewise, controlled by us via our AI. Essentially, we all have not only computer and knowledge skills, but these extend to the physical world.

Ok, that’s actually a lot and each item could be a massive post on its own, here we need speed and clarity, not pontification and back rubbing.

Effect on AutonomI

These changes to our computer interaction move the goalposts for sure. Rather than storing just any data and creating file structures, web sites, browsers and so on, Autonomi, must now protect humanity directly by protecting our own personal AI.

Centrally controlled AI is obviously evil, same with robotic control. The intimate relationships we will all have with our data/knowledge/AI must be protected and that means secure private and personal to each human, regardless of where you are and how you connect, using whatever device. This part is vital and without it society will struggle to maintain individuality. That individuality is how we evolve and innovate, in fact it is how we live. Losing that could lose all of us.

This is Autonomi’s greatest offering. The ability to secure and make available your AI conversations.

Which models will Autonomi provide?

The network should make any locally runnable open source model available to every person on the planet. It can take the human conversations and store them securely, ensuring only you have access to them. These are your property and private thoughts. They must travel with you and be available from anywhere, only to you.

Agents

Agents are perhaps the most simple and most powerful of all todays software tools. They are not understood my most, but represent a massive opportunity for us all.

An agent is basically a set of configurations we apply to an LLM to focus it on a specialised task. For instance we can have agents that represent

• CEO
• Software Developer
• Software Q/A
• Software tester
• Web site builder
• Customer support

What we do is simply supply basic information to make this happen. This info can be as simple as

{
Name: CEO
Skill: Build teams and have final say on deliverables
BackStory: You are a CEO of a sucessful sofware company ...
LLM: <any LLM we wish goes here>
}

This small piece of text is all we need to create an agent. Some LLMs may be tuned for such actions, but even a generic one can be guided as an agent as above.

The power of agents

The strength of this approach is we no longer simply ask a question from a single AI and get an answer, right or wrong. Instead, we ask our team of agents. They delegate tasks, use tools, such as web search, stock checker, spreadsheets and so on. They check each other’s output and can demand it’s not good enough and their task has to be redone.

So now we have planning, recursion, long term workloads and much more. This allows us to take manly LLMs, combine them in clever ways and have them perform and check tasks for us. This basically allows humans to create virtual companies, project teams and much more. Even with their own customer support, feedback and upgrade mechanism to continuously develop their commercial offering. All without human intervention.

Or we have teams to control our house, workshop, education and more. Really, there is no bound.

Autonomi and agents

This is a match made in heaven. Autonomi allows us to publish an agent and update that. Others can use it and lock to a version of our CEO (for instance) or always use the latest (Autonomi puts you in control at all times).

This is a totally new way to share and decentralise community created content on a massive scale. As people gravitate towards the person who creates the best CEOs or Software dev agents, then folk will use this and pay the creator. It’s quite magical.

## Introduction

In recent years I have studied to help create the SAFE network, along with brilliant Engineers. It’s been a journey and one of a typical creative endeavour. I have always figured AI or at least evolutionary programming would be a big part of that. From simple perceptrons through genetic programming and neuroevolution (NEAT in particular). Recent LLM success has certainly added an intriguing angle to these studies, mainly the notion that copying a brain may not be the best route to “intelligence”. This is fascinating, but more pleasing is the simplicity of the LLM models.

## Simplicity

Recently the tech world was stunned by generative AI models, in particular GTP, then the rest of the world followed as it became accessible via CHAT GTP. What is remarkable to me though is the simplicity of “guessing the next token (word)” when we introduce context (attention is all you need). This shocked many of the “fathers” of AI to their very core. Not in how capable it was per se, but how simply it managed to get to human level capabilities.

We always thought creativity and our intelligence separated us from machines, it turns out that may be the thing we have most in common, at least for now. That should shake everyone!

## Creativity

Creativity is an interesting concept. It involves many steps, imagining things, testing things that might work, but probably don’t and then forging a path to make the previously impossible possible. What is imagining, if not hallucinating (As the AI world has come to know it). So hallucination is good, it’s required for creativity. Stop any creative process mid way and you will have a hallucination like outcome.

Creativity also requires research, crazy ideas and something rather interesting, unbiased and limitless patience from partners. Innovation tends to come from individuals, big stuff like electricity, magnetism, relativity, QED and more stemmed from individuals. Partners are difficult as they introduce ego (their idea must be used) and they are not ready and willing when the lightbulb moments strike (late night, very early morning etc.). Having a near human level LLM with broad expertise in many areas opens doors that just did not even exist last year.

## Complexity

Professionals love complexity. It seems to boost the ego of practitioners in many areas, but nature hates it. When I say complexity I mean very narrow areas of knowledge presented in large and confusing language with many graphs and weird Greek lettered formulae or work that is so complex it is only understandable by a tiny percentage of experts. I see this as detrimental to the evolution of our species/knowledge.

Nature is complex, our brains were thought to be complex, but that complexity may be much less so in my opinion.

I think many “experts” will take modern AI and make even more complex papers and formulae from it. That will boost ego and do little for humanity in most cases. Complexity is really something we use when we cannot understand the fundamentals of life and the interaction of those fundamentals. Back to “attention is all you need” and really think about that, it’s fascinating.

The complexity of understanding the coordination of many simple artefacts to produce something as powerful as a human or as successful as an ant colony is incredibly difficult, as it involves extreme deep thinking. It also requires, as we also see in nature, many iterations with tiny changes each time. These systems will not work right until the last tiny change, then they do. At this point though, another tiny change may break them. Interestingly, when they work it’s unlikely they can be described in any detail which is where Engineers get nervous, but they shouldn’t as they are themselves actually built in this way.

This, to me, is the real area of complexity. It’s very much like a neural network that works, we cannot describe it in detail, we cannot describe and ant’s make up in detail, at least in relation to it’s function as part of the colony. No amount of Greek letters and advanced mathematics seems to be appropriate. Many people may disagree, but this is where I feel real advances can be made, should we be able to teach this kind of thinking at a deep level.

## Context with respect to LLMs

A slight deviation here. Context is important when guessing the next thing. This is what gave modern generative AI (LLMs in particular) the jump we now see. However, it’s more than context in relation to the training, i.e. the model itself. That training data is many hundreds or more exabytes of raw data that is encoded into some vector representation of context or distance between tokens (attention).

## Why are the current AI’s not that smart in some ways

This is again context. The training data may be perfect (it’s not, but assume it is), but the context from us to the AI is incredibly limited. It’s a small window in which we have this conversation. That window may allow only a few hundred messages in a conversation, and even then it’s vague in the middle of the conversation (technical issue). The context on our side, the human, is limited.

Going back to creativity, it’s a back and forth conversation, hallucinating, testing and changing. With limited context windows, current LLM’s are crippled. When we add to that the “one shot” type conversation where the AI starts the answer almost immediately, it means we get a step-by-step look at the creative process. When we see a hallucination we stop and say “hey this is broke” but I don’t believe it is. I think we stopped a creative brain before it was able to even take step 2 of a multi-step process.

There are projects to increase the human side context and perhaps even make that unlimited. There are also projects to allow the AI to iterate, but they are in early stages. This problem, though, is not complex at all and projects like memory-gtp, multi stage prompting and agents are already making headway, if not already there in alpha form.

However, I believe the AI will be able to take our individual (and hopefully private) life experience, but importantly it will not be complex models and algorithms that achieve this. It will be as simple as we have today. So simplicity is key here, it cannot be stressed enough the simplicity of these systems is what is startling us all.

## Conclusion

The SAFE network has taken a massive turn recently and gone back to its roots. These are founded in simplicity, a natural system based on the functions of an ant colony. That sophistication through what looks like a complex series of actions we cannot understand, much like the neural networks in these simple LLMs. Each neural network node is super simple too, it’s just adding them together that gives us complexity that leads to sophistication. However, we cannot understand the simple reactions when they are added together, and neither should we.

It is the simplicity of the individual components communicating that gives us not only natural capabilities, but the ability to evolve systems. Complex mechanisms themselves are not scalable, understandable or natural.

The SAFE network, likewise, is based on simplicity and recently the team have made this even simpler, but with dramatically powerful consequences. Hard to explain, but easy to see the outcomes.

This is my excitement in AI, not the ability to create complex algorithms and be applauded as some kind of super genius, but to dive deeper into creating systems that comprise super simple steps to give us even more capable systems. Then, for us all to really understand the connectivity of all things, at a fundamental level. It’s not about genius through complexity, it’s about genius through simplicity.

AI will allow us to create and evaluate simple interconnected systems in a way we never could. While others create complex components, I am ecstatic to look deeper and deeper into the cellular automata approach (again) ,enhancing creativity and advancing our species. So more ant investigation and much more fun!

Shaping the Future: From Data to Knowledge in a Privacy-Centric World

In the rapidly evolving digital landscape, the traditional paradigms of data management are being challenged. As the digital footprint of individuals and organizations continues to expand, there arises a critical need to shift from data-centric to knowledge-centric models. This post delves into key areas of this transformation: Digital Advertising, Personal Data, Shared Data, Secure Knowledge Management, and Nano Payments.

1. Digital Advertising Reimagined

The current digital advertising model primarily focuses on capturing consumer attention. However, as Artificial Intelligence (AI) advances, a paradigm shift towards value-based advertising is imminent. Consumers, assisted by AI, will seek the best value, pushing advertisers to offer competitive pricing and quality assurance. The shift to value proposition could foster long-term relationships with consumers, promoting a culture of transparency, fairness, and consumer satisfaction. This shift reflects a mature understanding of consumer needs, moving from mere attention capture to delivering real value.

2. The Evolution from Personal Data to Personal Knowledge

Transitioning from data to knowledge represents a novel approach to personal data management. Technologies like machine learning and vector databases can extract and compactly represent knowledge, minimizing data storage requirements and enhancing privacy. This transition has the potential to impact various industries significantly, offering a granular level of control over the information shared and fostering a sense of ownership and control among individuals.

3. Collective Wisdom: The Future of Shared Data

The concept of sharing knowledge instead of raw data opens new vistas for collaborative research and innovation, especially in sensitive fields like healthcare and finance. Cryptographic techniques like Homomorphic Encryption and Zero-Knowledge Proofs ensure privacy and authenticity in this knowledge sharing paradigm. A novel proposal further enhances privacy by introducing an anonymized response mechanism, where entities deposit their answers to questions in a designated “dropbox”, decoupling the identities of questioners and responders, thereby fostering a more privacy-centric knowledge sharing ecosystem.

4. Secure Knowledge Management: Harnessing the SAFE Network

The SAFE Network, with its decentralized, privacy-focused approach, provides a solid foundation for managing and sharing knowledge securely. Its principles align well with the idea of giving users control over their data and knowledge. The integration of secure data management on the SAFE Network with knowledge extraction techniques could create a powerful platform for managing and sharing knowledge securely, paving the way for a new era of secure, user-centric digital ecosystems.

5. Nano Payments: Enabling Autonomous Transactions

The convergence of the SAFE Network’s nano payment processing capabilities with personal AI technologies could usher in a new era of financial transactions. This scenario could enable real-time payments for real-time services, micro-subscriptions, and decentralized finance (DeFi) applications, ensuring a seamless user experience while adhering to financial regulations and standards.

Conclusion

The technological and conceptual shifts discussed in this post paint a picture of a future where individuals have more control over their data and knowledge, where new business models become viable, and where privacy is a fundamental pillar, not an afterthought. As we transition from a data-centric to a knowledge-centric world, the opportunities to reshape the digital landscape in a privacy-centric manner are boundless.

What?

---

A chain is an image we all know. It represents strength and reliability. Bitcoin has made use of such imagery in the implementation of a blockchain. A blockchain is a chain of links where each link couples transactions (blocks of transactions) in a reliable and cryptographically secured manner.

Data chains represent, instead, links that couples together data identifiers. A data identifier is not the data. As the blockchain does not hold bitcoins a data chain does not hold data.

This post discusses an overview of data chains, there is also an RFC covering some basic use of DataChains and also a codebase to implement DataChains, as well as Data handling (recognition) and long term storage.

Why?

---

What use is that then? If these things hold no data then what use are they?

This is a critical question to answer and resist writing off as irrelevant, just yet. If we can be assured a transaction happened in a blockchain, then it allows us to know where a bitcoin should exist as if it were a real thing. It is the same with a data chain, if we can be sure a piece of data should exist and where it should exist. However with data chains, identified data is real (documents, videos, etc). That is, if we know these files should exist and can identify them, then we just need to get the data and validate it.With a network that both stores and validates data and their IDs, we gain a lot in efficiency and simplicity as compared to blockchains which cannot store significant amounts of data such as files. Data chains would additionally allow cross network/blockchain patterns, but one must ask why do that, duplication is not efficient?

Data handling 

Therefore if a block identifies indisputable information about a file, such as (naively) a hash of the content, then we can be presented with data and compare to a valid block. If the block says the data is recognised, then it was created by whatever mechanism the network agreed to assume responsibility for the data.

Now we can historically validate that data belongs on the network and it was paid for or otherwise accepted, i.e. it is valid network data. This is a great step, some will think, oh I can do that with XYZ, but hang on for a few minutes.

Network handling

Here, we separate from a standard one truth chain and look at chain parts, or a decentralised chain on a decentralised network. A chain where only a very small number of nodes will know of the validity of some small sets of data, but the network as a whole knows everything and all data. Ok, we can see now this is a bit different.

We need now to look again at the chain a little closer, here is another picture to help. We can see here that there seems to be links and blocks. This is an important issue to cement in our thinking.  In a blockchain for instance the link is simple (simple is good, in security the simpler the stronger in many cases).

Here though, a link is another agreement block. These links are actually a collection of names and signatures of the group who agreed on the chain at this time. Each link must contain at least a majority of the previous link members. With every change to the group then a new link is created and added to the chain. Please see the consensus overview blog for an overview of these groups and their importance.

Between these links, lie the data blocks, these represent data identifiers (hash, name, type etc.) of real data. Again these are signed by a majority of the previous link (in actual fact they can be slightly out of order as required in an out of order network, or planet 🙂 ).

Now the picture of a DataChain should be building. But how does it split into a decentralised chain?

Splitting up

The easy way here is to consider just the above chain picture, but also remember back to a xor network, or plain binary tree, here is one as a reminder.

Remember also we have a notion of groups of nodes. So lets take a group size of 2 as an example.

The process would be:

• Node 1 starts the network, there is a link with one member.
• Node 2 starts and gets the chain so far (it’s small with only node1 listed). Node 1 then sends a link to node 2 signed by node 1.
• node 2 then sends a link to node 1 signed by node 2.

So now the chain has two links, one with node 1 alone and the next with nodes 1 and 2, both signed by each other. This just continues, so no need for me to bore you here,

However, then node 4 joins and assuming all nodes have a purely even distribution (meaning 2 nodes address starts with 0 and the other two nodes address start with 1) the chain splits! two nodes go down the 0 branch and two go down the 1 branch. The chain has split, but maintains the same “genesis” block. The link with node 1 only. Between these links the data blocks are inserted as data is added (Put), edited (POST) or Deleted from the network, each data block again is signed by the majority of the previous link block (to be valid).

So as more nodes join then this process continues, with the chain splitting as the network grows (for detail see RFC and code). This allows some very powerful features, which we will not dive too deeply in but to name a few as example:

• Nodes can be shown to exist on the network (i.e. not fake nodes).
• Nodes can prove group membership.
• Nodes can have a provable history on the network.
• Nodes can restart and republish data securely that the network can agree should exist.
• Network can react to massive churn and still recover.
• Network can recover from complete outage.
• Network can handle Open ledgers for publicly verifiable data.
• Network can individually remember a transaction (so for instance a single safecoin transaction can be remembered as a receipt, at a cost though).
• Network can handle versioning of any data type (each version is payed for though).
• Fake nodes cannot forge themselves onto a chain without actually taking over a group (meaning they cannot be fake nodes 🙂 ).
• As data blocks are held between verifiable links then data validity is guaranteed.
• As data is guaranteed some nodes can hold only identifiers and not the data, or a small subset of data.
• As not all nodes are required to store data, churn events will produce significantly less traffic as not all nodes need all data all the time.
• The network can handle nodes of varying capabilities and use the strongest node in a group to handle the action it is strongest with (i.e. nodes with large storage, store lots, nodes with high bandwidth transfer a lot etc.).
• Archive nodes become a simple state of a capable node that will be rewarded well in safecoin, but have to fight for its place in the group, too many reboots or missed calls and others will take its place.
• Nodes can be measured and ranked easily by the network (important security capability)

There is a lot more this pattern allows, an awful lot more, such as massive growth or an immediate contraction in the number of nodes. It is doubtful the full capability of such a system can be quantified easily and certainly not in a single blog post like this, but now it’s time to imagine what we can do?

How? 

Moving forward we need many things to happen, these include, but are not limited to:

1. Design in detail must be complete for phase 1 (data security and persistence)
2. Open debates, presentations and discussion must take place.
3. Code must be written.
4. Code must be tested.
5. Integration into existing system.
6. End to End testing
7. Move to public use.

Of these points 1, 2, 3 & 4 are ongoing right now. Point 5 requires changes to the existing SAFE routing table starting with an RFC. Point 4 will be enhanced as point 2 gets more attention and input. Point 6 is covered by community testing and point 7 is the wider community tests (i.e. Alpha stages).

Conclusion

Data chains would appear to be something that is a natural progression for decentralised systems. They allow data of any type, size or format to be looked after and maintained in a secure and decentralised manner. Not only the physical data itself (very important), but the validity of such data on the network.

Working on the data chains component has been pretty tough with all the other commitments a typical bleeding edge project has to face, but it looks like it has been very much worth the effort and determination.  The pattern is so simple (remember simple is good) and when I tried to tweak it for different problems I thought may happen, the chain fixed itself. So it is a very simple bit of code really, but provides incredible amount of security to a system as well as historical view of a network, republish capability etc. and all with ease. The single biggest mistake in this library/design would be to try any specialisation for any special situation, almost every time, it takes care of itself, at least so far.

This needs a large scale simulation, which we will do in MaidSafe for sure prior to user testing, but seems to provide a significant missing link from many of today’s networks and decentralised efforts. I hope others find this as useful, intriguing and interesting as I do.

We think there will be many posts depicting various aspects of this design pattern over the next while as we get the ability to explain what it offers in many areas. (i.e. when we understand it well enough to explain it easily)

 

 

The features included in decentralized networks can be quite varied based on the proposed goals of the technology. From the sharing capabilities offered by Bittorrent to user privacy enabled via To…

Source: Introduction & Technical Overview of SAFE Consensus