Voice of the MANXers: Series #7

The Seventh MANX Ask Me Anything (AMA) was held on October 10, 2018 and focused on Blockchain and Artificial Intelligence. It was conducted in the MANX Academy of Token Economics (MATE) group, an invitation-only telegram group focused on informational and educational discussions on all aspects of blockchain technology. The members of the group are selected for their active engagement in the global MANX communities.

This session was led by Dr. Feida Zhu (Head of Artificial Intelligence), Dr. Jack Chiu (Head of Technology), Dr. Yang Tang (Head of Network Security) and moderated by Dr. Yawei Cui (co-founder) and Olivia Fan (Head of Operations). This transcript has been edited for clarity.

Yawei: Hello everyone, and welcome to our 7th AMA. Over the past few weeks, we have been trying our very best to present you with a comprehensive picture of our project along with a lot of expert insight on blockchain technology. Despite the austere situation of the crypto market, we continue to work and produce, fighting the big fight for the long term.

Yawei: A lot of people are talking about the convergence of AI and blockchain. What does it really mean? How could they help each other and create something better?

Feida: AI and blockchain are two significant technologies for all industries. Each technology has its own complexity and features and the combination of them are faced with many opportunities and challenges.

Generally speaking, AI can help improve blockchain in such areas as energy consumption, scalability, security and efficiency. Blockchain technology can help audit AI predictions, lower AI market barriers to entry and increase trust.

Olivia: How does AI play a positive role from community management to the KYC process, from onboarding investors to token sales, and then to post-ICO token value management?

Feida: In areas such as engaging with people and stakeholders, building strategy and innovation, even token value management, each part is labor intensive and needs expert experience and skills. AI can help with these specific domain applications by creating models to reduce pre-defined errors, saving a lot of labor costs.

For example, in the KYC process, there may be frauds that submit personal information and photographs downloaded from the Internet. Through techniques such as facial recognition and duplication check algorithms, AI can help detect and prevent such frauds.

Olivia: How will MANX use artificial intelligence to improve security?

Yang: The system implements a regular upgrade of the security model through self-learning and intelligent optimization techniques to adapt to external attacks and fraud prevention requirements.

Security rules are intelligently generated based on context, transaction amounts and familiarity to produce security verification methods that give customers the safest security experience.

Yawei: What intelligent authentication methods does MANX use to improve application security?

Yang: MANX uses intelligent matching security methods based on biometric authentication technology, incorporating environmental factors and transaction habits of mobile nodes into the authentication analysis.

Yawei: Many people are interested in knowing how MANX can accurately determine “who I am?” to ensure that “my money will not get lost”. Can you comment?

Yang: The security threat awareness system can automatically learn rules by clustering and classifying the access data of application systems to form a user behavior baseline to help customers identify environmental security risks, ongoing APP attacks and trace malicious transactions.

Yawei: How does the use of AI help enhance the level of privacy protection?

Jack: The privacy issue of owning personal data raises regulatory and strategic concerns. Homomorphic encryption (that performs operations directly on encrypted data), the Enigma project and the Zerocash project are all potential solutions.

The incredible progress made in machine learning over the past two years makes AI a fantastic tool for the blockchain to enhance secure application deployments, especially given the fixed structure of the system.

Olivia: What applications will MANX focus on to exploit AI technology?

Feida: MANX involves advanced technologies such as encryption, security and sub-chain that give AI a broad stage. MANX will focus on applying AI technology to assist in the design of these technologies. We believe we can improve system performance through AI.

For specific applications, such as natural language processing and computer vision, we will work with our global user community to develop algorithms based on decentralized data. We will provide instructions and templates to help users get started quickly.

Yawei: Are there any relevant projects that combine AI and Blockchain? What the strengths of MANX over other projects?

Feida: There are some projects such as SingularityNET, DeepBrain Chain and Numerai. In my opinion, projects like these combine AI and blockchain in a very rigid way. For example, one project provides a platform so developers can share their predictive models for investment strategies. In this platform, the best performance models can win rewards. In this case, the predictive model is about AI and the platform is based on the blockchain. The link between the two technologies is weak and not necessary.

MANX aims to improve system performance with AI so that we can provide efficient and cost-effective blockchain-based services for users.

Olivia: My main concern is that AI needs a lot of data to train and learn. Blockchain requires that all nodes share the same database. Is there an efficient way to spread large data to all nodes in the distributed network?

Jack: Data is distributed in a decentralized manner. When a user wants to utilize a specific category of training data, they can collect the relevant data from the decentralized nodes.

The challenge is to ensure that data integrity can be maintained with lightweight methods. We don’t require all nodes to store the database.

Yawei: How will AI technology change the blockchain? Or the reverse question, how will blockchain change AI? What’s new that can be expected from their combination?

Feida: That’s actually several big questions! Let me take them one at a time.

For the first question (how will AI can change blockchain), we generally focus on energy consumption, scalability, security, efficiency and data.

For the second question, AI can also benefit from blockchain. One example is to increase trust — once part of our tasks is managed by autonomous agents, having a clear audit trail will help bots to trust each other. In another example — catastrophic risk scenarios — an AI coded in a DAO with specific smart contracts will only be able to perform those limited actions and nothing more.

MANXer V: Will AI be more effective with blockchain?

Feida: Blockchain will help AI to explain how it arrives at decisions. Having a clear audit trail can improve the trustworthiness of the data and the models and it can also provide a clear route to trace back the machine decision process.

Secure data sharing means more data, and then better models, better actions, better results and more new data. Network effect is all that matters at the end of the day.

Olivia: There are several popular machine learning frameworks such as PyTorch and Tensorflow. How will you let developers develop and deploy their algorithms?

Jack: MANX will focus on applying AI technology to encryption, security and sub-chain techniques. Frameworks like those you mentioned are suitable for developing NLP or computer vision algorithms.

MANX also provides a platform where users can share data and AI algorithms. We allow users to choose whichever framework they prefer to develop their algorithms.

MANXer R: How will using artificial intelligence be beneficial for blockchain DAPPS? Can it make smart contracts even smarter using machine learning concepts, for instance identifying bugs in smart contracts or using natural language processing to define smart contracts?

Feida: Great question! Using AI to automatically detect bugs in algorithms is a hot topic. Thanks to the recent rapid advances of AI in this research area, I think that’s a very promising direction.

MANXer Z: I think the assumptions behind the convergence of AI and Blockchain are somewhat “optimistic” in nature. What are possible problems of this convergence?

Jack: In MANX, the user experience is always our most important objective. AI and blockchain are both meant to enhance the user experience. If AI can help improve performance and reduce costs, we will embrace AI tech.

In practice, AI tech is applied in different areas such as encryption, security, sub chain and consensus. As long as the AI shows value in these areas, we will use it!

Yawei: How can AI help from a regulatory point of view and will the convergence of AI and Blockchain benefit the development of RegTech?

Jack: RegTech operates over the financial and regulatory spaces. RegTech projects include employee surveillance, compliance data management, fraud prevention and audit trail capabilities.

AI will benefit the deployment of RegTech. For example, AI can be trained to detect fraud so that it can detect and automatically alert when suspicious operations occur.

A blockchain is a secure immutable database shared by all parties in a distributed network. The immutable history is perfect for audit purposes. The convergence of AI and blockchain will surely benefit RegTech.

MANXer V: AI needs a lot of resources, what do you think about this?

Feida: Computing resources and data are controlled by major tech companies. MANX provides a platform so that users can share their data and algorithms motivated by incentives, bridging the gap between algorithm developers and data.

Olivia: Currently, major tech companies dominate AI since they have the computing resources and enormous data pools necessary to train AI algorithms. I think the key to democratizing access to AI and letting SMEs/everyday consumers have the ability to create their own AI applications is decentralizing data and lowering computing costs for training. Will blockchain help level the AI playing field?

Feida: Computing resources and data are the real obstacle to developing algorithms. Cloud computing resources such as the Amazon AWS is not cheap. The collection of large-scale training data is even harder.

It’s one of our goals that SMEs can access decentralized data in our platform. Incentive mechanisms will encourage data and machine vendors to offer their services in exchange for cryptocurrency.

The exchanges between different parties in the marketplace are facilitated by smart contracts.

Developers can easily share the AI services that they develop so that they can access more data to finetune their model.

Olivia: Speaking of machine learning, could you please give us a brief comparison between AI and Machine Learning (ML)?

Jack: We think of AI as an interdisciplinary field, which covers the study of many sub-disciplines, such as natural language processing, computer vision, the Internet of things and robotics.

If we compare AI to the human body, it has to possess a brain to carries out a variety of tasks and is in charge of specific functions such the language (NLP) and sight (computer vision). Machine learning can be seen as specific movements, action or thoughts we develop. There is no widely accepted definition to distinguish these two concepts.

Olivia: How can we incentivize consumers to share their data on the blockchain so AI can have access to sufficient raw data?

Feida: The algorithms developed for different application scenarios need different training data. New data may stimulate new applications. They reinforce each other.

On the data provider side, we incentivize users to share their data with tokens. The more frequently their data is used, the more rewards they will receive.

On the data consumer side, they can request specific data in specific structures. Providers may share, construct or label such data in exchange for cryptocurrency

Olivia: What will be your first application that combines AI and blockchain? When will it be launched?

Jack: In the short term, MANX will provide AI templates for image recognition, chatbots, etc. to stimulate users to share data and test new algorithms. In long-term, MANX will apply AI technology to encryption, security and sub-chain techniques to improve system efficiency.

Olivia: How can you incentivize the continued development of AI algorithms?

Feida: Developers can share AI services they develop to receive cryptocurrency. In addition, MANX will create an account for each developer.

The performance history of previously developed models is recorded on the blockchain. Service buyers want to use services provided by high-level developers. This forms a closed loop system. Developers provide better algorithms to the public forming an immutable performance history. Buyers buy services according to performance history, which in turn encourages developers to develop more and better AI algorithms.

Yawei: Unlike the traditional capital market, there are so many token exchanges in the world. Will the convergence of AI and blockchain help rationalize the exchange space?

Jack: The main role of an exchange is to provide a matchmaking system to facilitate the purchase and sale of cryptocurrency. I don’t see AI having an impact on this.

MANXer D: The more valuable data is, the more it costs to use it. What will provide balance between data and prices?

Feida: That’s determined between the data provider and the data consumer. Data is a traditional product where the product price obeys market principles of supply and demand.

MANXer D: Is it possible that AI will help set data pricing based on the category of data and other available information?

Feida: Yes, that’s possible. Smart pricing schemes have already been applied in big companies such as Walmart and Amazon. We will also suggest a price for data providers.

MANXer D: AI is closely tied with self-study and a powerful analysis on which everything is built. Does this mean that self-promotion is possible in improving the security of MANX?

Feida: AI can be classified in different ways. For now, most AI is narrow AI, which means it is nothing more than a specific domain application or task that gets better by ingesting further data and “learns” how to reduce the output error. You probably mean General Artificial Intelligence — that’s the next stage.

Yawei: What sectors, functions or products and services will be disrupted by AI the most and the soonest?

Jack: The adoption of AI and the speed of its deployment vary according to industry. Many places have already felt the impact from AI such as agriculture, customer support, healthcare, software development and retail.

Yawei: In many situations, AI is not that “intelligent” because the underlying data is too dirty or the algorithm is not stable. How will you help prevent such cases?

Jack: Data cleaning is a hot topic in AI. Many efforts from the research and industry communities have been made to enhance data quality, which is beyond the scope of today’s discussion. Different applications need different tricks.

Olivia: I don’t want my personal information to be used by AI. Even though all the information on the blockchain is public, is there an option that users can choose to keep their information private?

Jack: MANX attach great importance to protecting user data. First of all, the information is encrypted. Only your private key can unlock the encrypted data. Second, you can choose whether to share the data.

MANXer Z: How do you see yourself, a trained expert in AI, helping MANX provide SMEs with SLAS? Could you please give us some examples?

Feida: SLAS, security, longevity, accessibility, scalability:

Security: The system implements a regular upgrade of the security model through self-learning and intelligent optimization techniques to adapt to external attacks and fraud prevention requirements. Security rules and security levels are intelligently generated based on context information, transaction amounts and familiarity to produce the best security verification methods to give customers the best security experience.

Longevity: MANX will apply AI technology to encryption, security and sub-chain techniques to improve system efficiency. The adoption of AI is these areas are topics for long-term research.

Accessibility: MANX provides a platform so that users can share their data and algorithms motivated by incentives, bridging the gap between algorithm developers and data. Also, MANX supports users in developing and sharing AI algorithms developed by current machine learning frameworks.

Scalability: Blockchain is growing at a steady pace of 1MB every 10 minutes. AI can introduce decentralized learning systems such as federated learning or new data sharding techniques to make the system more efficient.

Stay tuned for our 8th AMA about our MANX Academy of Token Economics (MATE). We’ll cover such topics as Why did we set up MATE? Why is MATE important? What can MATE do?

We’ll answer these questions as well as share how we believe that the setup of MATE is indicative of the ambitious nature of MANX project.

To learn more about MANX, please visit https://www.manx.network/. Follow us on telegram: https://t.me/macrochain. Subscribe to our official announcement channel: https://t.me/manxannouncements