latest AMA with Prof. Deng! Feb 13, 2021 1- What is the current status of AI implementation on the Matrix mainnet?
The mining function already uses an AI algorithm for image recognition.
We are building AutoML on top of the Matrix mainnet.
We are also building a NFT mechanism to ensure the right of AI models and data.
2- Is AI computation capabilities implemented so all it takes is to upload an AI model and data ot the network?
The target of AutoML is that you can automatically derive an AI model after uploadign data.
Of course, this would be incremental. Complex datasets and tasks (e.g. NLP) would still need human experts to build a model, but relatively tasks would be OK.
3- Please elaborate of the current capabilities and status. Aside from the AI marketplace and non functional APOCs, is the mainnet ready to start serving a client with live AI computation?
Generally, yes. The mainnet is ready for AI computations.
4- Does the bio wallet aim to be they key to a safe and easy way to use crypto wallets?
This would be a great step towards mass adoption.
Yes. It supports more user-friendly and safer access to the wallet.
Of course, there're difficulties. One problem is the the hardware module is power-hungry and thus requires relatively large batteries.
5- Is this bio wallet only for matrix network usage?
The technology is general and can be used for any crypto-currencies.
6- Could you explain more in detail about the autoML search function on the matrix network?
Is it supposed to make users find apps on the network? Why is it so complicated to find apps on a blockchain network?
No, it's not to find an application. It's designed to automatically build an AI model for a give task (e.g., image recognition) for a given set of data.
Previously, the model has to be created and tuned by AI experts.
With the AutoML app, people can automatically get an AI model by submitting a dataset and defining the task.
7- When will we have a matrix network that can go deep into reality?
It depends on how you mean by "go deep into reality". Generally, it's a marketing problem. I think everything we do is target this goal.
8- Could you explain why matrix should have a chance against AI clouds from google, tencant etc?
And why should matrix be the choice for AI scientists to use ahead of google etc?
A blockchain based AI cloud will be less competitive in terms of performance (i.e. throughput), but can be more cost-efficient.
Another advantage is that it supports better privacy. When you upload your data to google and tencent clouds, it's potentially risky because they are also data users and app developers.
On the blockchain based platform, all data are decentralized and thus less prone to data abuse.
https://t.me/matrixainetwork/788689 https://twitter.com/matrixai_news/status/1360625180289163268?s=28