A brief introduction to Open Ledger, led by Polychain: a blockchain network for dedicated AI language model SLM

Author: ABC Alpha Researcher

Project Name: Open Ledger

Project positioning: Blockchain network for dedicated AI language model SLM

Project highlights: In the dedicated language model track, blockchain technology is cleverly used to solve problems such as AI data capture, rights confirmation, and incentives. The value of blockchain is truly brought into play.

Financing: Seed round of financing of US$8 million, led by Polychain and other institutions

Participation method: TGE has not yet been launched. Currently, you can use your computer to participate in the test network mining incentives.

The following is a detailed description of the project:

1. What is OpenLedger?

Simply put, @OpenledgerHQ is a decentralized network that combines AI and blockchain technology. Its core goal is to create a decentralized AI ecosystem where everyone can contribute data, develop AI models, and make money.

There are several key roles here: dedicated data contributors, dedicated AI model developers, Open Ledger, and dedicated AI applications.

Of these four roles, I particularly emphasized the word "dedicated", which is the core of Opne Ledger. Opne Ledger does not build a general AI language model like Open AI, but builds a dedicated AI language model, or SLM. That is, the data is dedicated, and the AI model is trained using dedicated data.

Opne Ledger absorbs, cleans and converts data into usable data based on dedicated data networks (Datanets), and then provides it to dedicated AI model developers. Opne Ledger's attribution proof mechanism built on blockchain technology can distinguish the source of data and confirm the ownership of these data to the identity of specific data providers. Therefore, Opne Ledger can clearly know that certain dedicated data is contributed by certain specific users, thereby completing ownership confirmation and incentives based on blockchain.

OpenLedger particularly emphasizes a concept called "Payable AI", which means that AI models can not only be used, but also automatically share money with data providers and model developers based on their contributions. Imagine that you are a photographer and have taken a lot of beautiful photos of beauties. You upload these photos to OpenLedger, and AI developers use your photos to train an AI model of beauties that generates art. Every time someone uses this AI to generate a portrait of a beauty, OpenLedger can track that your photo has been used, and then automatically share some of the profits with you. This is the core of Payable AI and the core operating logic of Open Ledger. This logic is not applicable to general large language models, but in SLM, that is, the track of dedicated language models, blockchain plays a huge role, and OpenLedger is the general platform for this track.

2. How OpenLedger works

To explain more clearly how OpenLedger works, we will break down several of the technical terms involved in OpenLedger and explain the role of blockchain in each sector.

The operation of OpenLedger can be divided into several key parts:

(1) Data source: Datanets

OpenLedger relies on something called Datanets to collect data.

Datanets are like specialized data markets, each targeting a certain field, such as medical data, music creation, social media content, etc. Anyone (individuals, companies, experts) can upload their data to Datanets.

For example, if you are a doctor, you can upload some anonymous case data; or if you are a programmer, you can upload code snippets. These data will be sorted and cleaned to become the "raw materials" for AI training.

The blockchain is like a super transparent ledger. Every time someone uploads data, the blockchain will record who uploaded what and how many times the data was used. In this way, no one can secretly change the data, and no one can deny that your data was not used.

For example, Bob is a game anchor. He uploaded his live clips to OpenLedger's "Game Datanet". A company used these clips to train an AI that can automatically generate game commentary videos. The blockchain records that Bob's clips have been used 100 times, and Bob can get paid according to the number of times they are used.

(2) Proof of Attribution

This is the core technology of OpenLedger, which translates to “proving who contributed what.” It can track every output of an AI model and figure out what data was used and who contributed.

When an AI model generates a result (such as an article or a picture), OpenLedger will analyze which data sources the result depends on, and then distribute the credit to the corresponding contributors. For example, if AI writes a song, the system can find out that it refers to Peter's melody and Mary's lyrics, and then automatically distribute the money to them. Blockchain ensures that this "credit distribution" process is transparent and cannot be tampered with. All contribution records are written on the chain and cannot be changed by anyone. In addition, smart contracts (an automatically executed program) will directly transfer money to the contributor's account based on the records.

For example, Jack is a writer who uploaded a bunch of short stories to OpenLedger. An AI writer generates new stories in the style of his novels. Proof of Attribution finds that 30% of the new stories are inspired by Jack’s novels, and Jack automatically gets 30% of the revenue share.

(3) Payable AI

Payable AI is the ultimate goal of OpenLedger: to turn AI models into an "asset that can make money on its own." Developers can deploy AI models on OpenLedger, and users pay when they use the models, and the fees are automatically distributed to data contributors and model developers.

For example, a company develops a medical diagnosis AI, and doctors use this AI to analyze patient data, and they have to pay a fee each time they use it. This money will be distributed through the blockchain to the hospital that provides medical data, the engineer who trained the model, and so on.

Blockchain ensures that transactions are open and transparent. It is clear how the money is divided and who gets how much. Smart contracts can also automatically execute the profit sharing without the need for a middleman, saving money and worry. For example, Alice is an AI engineer who developed a translation AI and deployed it on OpenLedger. Every time someone uses this AI to translate a file, Alice and the contributor who provides language data (such as Ruby who uploaded English teaching materials) can share the money. The blockchain automatically calculates the ratio and transfers the money directly.

(4) OpenLedger decentralized network

Instead of putting all data and computing on a single company server, OpenLedger uses a decentralized blockchain network, which means data and AI models are distributed across nodes around the world, so no one company can dominate.

Users around the world can run OpenLedger nodes (like a small server) to help store data and run AI calculations. In return, they can also earn some rewards.

The blockchain is responsible for coordinating these nodes to ensure data security and credible computing results. The blockchain will record who runs the node and how much work has been done, so that no one will be missed when issuing rewards.

For example, Black runs an OpenLedger node at home and stores some music data for the platform. Someone uses this data to train AI, Black's node records the transaction, and the blockchain sends him some tokens as compensation.

3. What problems that were difficult to solve before have been solved by OpenLedger using blockchain technology?

There are several major problems in the AI field now. OpenLedger cleverly uses blockchain technology to provide its own solutions:

(1) Data sources are not transparent

Most AI companies randomly grab data from the Internet to train models. You don’t even know your data is being used, let alone get paid. OpenLedger uses blockchain to record data sources and usage, and attribution proof ensures that contributors are treated fairly. For example, many AI drawing tools are now trained with artists’ paintings, but the artists don’t get a penny. OpenLedger allows artists to upload their works, and AI automatically pays them when they use them.

(2) Centralization Risk

Currently, AI data and models are controlled by big companies. They can shut them down or change them at will, and users have no say. For example, if an AI platform suddenly says "you are not allowed to use it", you will be dumbfounded. OpenLedger's blockchain allows data and models to not be monopolized by a single company, and users can also participate in running the network, making power more decentralized. Just like in the early days of YouTube, video creators were completely dependent on the platform's mood to share money. OpenLedger is like building a "decentralized YouTube" for AI, where creators have more control.

(3) Contributors do not receive any rewards

The current situation is that AI models make a lot of money, but ordinary people who provide data and engineers who develop models often get nothing. OpenLedger's Payable AI allows every contributor to share the benefits, motivating more people to participate, and the data quality will be higher. Just like YouTube later gave UP hosts advertising revenue, the video quality increased rapidly. OpenLedger also wants to make AI data richer through revenue sharing.

(4) AI models are not diverse enough

Currently, most AI models are developed by large companies, which are general-purpose and difficult to meet niche needs (such as local language translation). OpenLedger focuses on the field of specialized language models. Datanets make it easier to collect data in professional fields, so that developers can create more accurate "specialized AI". For example, a language AI for a minority ethnic group may not be developed because of the lack of data. OpenLedger's Datanet allows ethnic groups to upload language data, so that developers can create exclusive AI.

The cleverness of OpenLedger lies in that it did not directly cut into the large language model track, but cut into the dedicated language model SLM track, which can give full play to the characteristics of blockchain technology.

Let’s list the roles that blockchain technology plays in OpenLedger:

• Bookkeeping: Record every data upload, usage, and revenue sharing, which is transparent and cannot be tampered with.
• Money sharing: Use smart contracts to automatically distribute profits to contributors in proportion, without the need for human intervention.
• Decentralization: Distribute data and computing tasks to global nodes to prevent a single company from dominating the market.
• Security: Data is stored encrypted, and blockchain ensures that no one can secretly change the data.
• Incentive: Give token rewards to those who run nodes and contribute data to encourage everyone to participate.

Summarize

The core of OpenLedger is to use blockchain to create a fair and transparent ecosystem for dedicated AI models. It collects data through Datanets, tracks contributions through Proof of Attribution, and automatically distributes money through Payable AI, so that everyone can get a share of the development of AI. Blockchain here ensures data security, transaction transparency, and fair distribution of money, and also prevents the entire system from being controlled by large companies.

The OpenLedger team keenly discovered that the general large language model track is monopolized by super companies such as OpenAI, Google, and Microsoft. They do not need blockchain and will not use blockchain to equalize themselves. The dedicated language model track (SLM) is an AI track that relies more on decentralized communities and decentralized forces to drive it. In this sector, AI and blockchain have achieved a clever combination.

OpenLedger also shows us a possibility, that is, the combination of blockchain and AI should not only start from technical needs, but also from the market. First, the business should be established, and then we can see what value blockchain and AI can play here. OpenLedger shows us this possibility.

Open Ledger testnet mining link: https://testnet.openledger.xyz/ The official has detailed guidance documents, just follow the steps to participate. Risk warning: The above content is for discussion and exchange only, DYOR.