RVCrypto
RVCrypto

@RvCrypto

30 Tweets 6 reads Mar 02, 2023
One of the biggest narratives, not only in crypto, is AI/ML.
I think I found the next Unicorn in the making, combining AI with blockchain to create the largest neural network: $TAO.
Although already a midcap, $TAO could still do a #x100 from here.
A๐Ÿงต
1) Problem
Today AI is inefficient, non-transparent and incredibly expensive.
The AI models are developed in centralized data siloโ€™s by the likes of Google, IBM, OpenAI and Microsoft.
Google alone spends 75% of their electrical cost on machine learning. The usage of a machine learning model like GPT3 costs $12 Million per hour!๐Ÿ”ฅ๐Ÿ’ต
Small players donโ€™t have access to, nor the money for these operations, putting the power in the hands of a few, not the many.
Furthermore, a new model (GPT3) has to relearn everything that the old model (GPT2) already learned.๐Ÿ‘€
This relearning process occurs very often because AI research doubles every year. This makes the AI sector extremely inefficient.
2) Use case
@bittensor_ is an open-source protocol that powers a scalable, globally-distributed, decentralised neural network focused on AI/ML.
The interesting thing here is that the goal is to create a network on which information compounds, which means a new model doesnโ€™t have to relearn everything a previous version of the same model already learned.
Furthermore, when using the bittensor network there is also a value transfer from other models that donโ€™t exist in a centralized AI silo.
The new model is basically learning from all the models and their data available in the neural network.
One of the Co-founders, Jacob Steeves, called it โ€˜a continues machine learning libraryโ€™.
This library is accessible for everyone in a truly decentralized manner and will accelerate the adoption of AI.
3) Consensus mechanism
The consensus mechanism of $TAO can be best explained by following the same path that the data follows when added to the network. These 4โƒฃ steps involve different parties with different roles in the process.
Lets dive in๐ŸŠโ€โ™‚๏ธ
1โƒฃ A company, that producing knowledge in the form of AI/ML models, contributes to the network by uploading this data.
If the data is valuable, they will be rewarded in $TAO.
2โƒฃ Various AI companies will validate the quality of this uploaded AI/ML models.
They validate it by using the models to solve specific problems. Once they reach consensus, they will get rewarded with TAO tokens.
Who gets rewarded in these first two steps is based on information significance measurement, it is like a matrix with contributors and validators getting a share of the rewards based on their "usefullness" ranking.
This first part of coming to consensus happens like smart contracts on a L2 and only when consensus is reached the information is added to the L1 PoS blockchain of $TAO.
3โƒฃ The third step is executed by the validators of the PoS blockchain, validating the blocks of the "L1" PoS blockchain.
This party doesnโ€™t need any knowledge about AI. But to be eligible to become a validator, you need 1024 $TAO and computing power to validate the blocks.
4โƒฃ After the AI models with value are added to the blockchain, this data is ready to be utilized by the end consumers, mainly AI companies running new models.
The capacity on the network is owned by the $TAO holders and they can utilize it to fulfill their needs.
So, accessing the @bittensor_ network can be done in two ways; The company has to hold $TAO to gain access or the company has to pay someone with $TAO, that is not using its full capacity on the network, to gain access.
If the network reaches its full potential, I can see $TAO wars happening just like what happened with $CRV in the curve wars ๐Ÿ”ฅ
The beautifull thing with BitTensor is that not only big companies, like IBM, Google or Microsoft, but also smaller ones can use this sophisticated library full of interesting AI/ML models and data for their own projects.
4) Team
Jacob Steeves and Ala Shaabana CO-founded BitTensor.
They both have great academic backgrounds and working experience in the sector, working for Google and the likes.
They have assembled a team of 18 at the moment. You can have a look here:
linkedin.com
5) Tokenomics
$TAO officially launched November 2021, bittensor did a fair launch, with no ico, no pre-mine, no team allocation.
The tokenomics are like those of $BTC, with 21 million max supply and a halving every 4 years, the first one happening in 2025.
So, looking at how early you are, this is like buying bitcoin in 2010.
As of now, the total issuance of $TAO is a little less then 3.5M, creating a MC of around $100M.
6) Energy
Energy consumption is a hot topic, not only in crypto, but also with AI and ML.
With central organizations running the same AI/ML models a lot of energy is wasted. With $TAO this can be done collaboratively and all at once.
To be clear, yes, bittensor is still going to use energy, like Google does for AI/ML, but in a way more effective way without the waste of energy.
To be fair, energy is always going to be consumed but it shouldnโ€™t be wasted ๐Ÿ™
7) Personal view
This one is hard to grasph, but once you see the potential you can not unsee it ๐Ÿคฏ
$TAO is like $BTC but not for digitally trusting value/money but for the neural network of AI/ML.
In other words, #Bitcoin is the biggest supercomputer in the world for digital trust and bittensor could be that for digital intelligence.
The only thing bigger than the likes of Google and OpenAI is all of them combined.
All permissionless open and decentralized.
And $TAO has the potential and incentive to make these centralized companies use the network, right @Dreadbong0?
Let me know if you found this thread helpful and what you think of $TAO.
Follow me @RVCrypto for more ๐Ÿงตs on high potential coins.
Like and retweet my first tweet for the bird algo ๐Ÿ™
8) How to buy?
$TAO is not on a CEX yet, but it is still very easy to buy $TAO if you want just follow the exact steps in the guide made by @dreadbong0.
Also check the official OTC discord: discord.gg
youtube.com
9) Further research
If you want to dive in deeper, check the whitpaper and docs on bittensor.com
Also check this awesome interview by @jeremiecharris
towardsdatascience.com
Or go directly to youtube via youtube.com

Loading suggestions...