Whats up crypto people!
Artificial intelligence is a buzz word used in many industries, to either get investors or to grab the attention of potential clients/users. I see day in and day out whitepapers and business plans which discuss AI and sell it as part of their project but really its just a bot or a series of commands in a program which really does nothing.
So how do you tell what a real AI is and what is a fake AI?
This is a question I’ve seen a lot of Blockchain industry professionals discuss but honestly I have never heard somebody tell you exactly what it means for an AI to be an AI and this is due to the individual not being an AI engineer… Simple. BUT lucky for you, I am actually working with an AI engineer on the BEAST platform which will be available on this website.
Now usually only large hedge fund s are able to use or create such an AI.
So let’s get to the nitty gritty, I had a discussion with Alex and asked him the following questions and this is what he had to say:
1- What makes a true AI ?
Well there are numerous versions of AI and Machine Learning.
There are also many names and misconceptions about AI.
One simple distinction can be made between AI and a Bot.
A bot will operate after a pre programmed code, for example if the price goes above 100 buy, if price then falls below 99 sell.
People could debate the terminology forever, but the short answer is that what we call AI will absorb all the information and learn from it, meaning it will create new data to make better predictions.
2- How is your AI different to others?
Most AI’s run on a local computer/machine and most often they are not Neural network algorithms (since neural networks take longer to train and therefore cost more to run). Ours is run in the cloud enabling us to scale the processing capacity up easily, and ours is a Neural Network = tries to learn as similar as possible to a human brain.
But the most unique feature about our AI is the data we put into it, data is what feeds an AI. Data for an AI is the same as nutrition is for a human. Garbage in = garbage out
3- What are the biggest difficulties in building an AI in the Blockchain world?
- AI needs a lot of high quality structured data (for example thousands of rows is not a lot of data).
- Working with all the complex trading data. All old time traders will tell you the same thing … strategies must change, what worked in the past will not work forever. As soon as someone finds a technical pattern to make huge sums of money, it soon becomes obsolete as more traders start to copy the strategy.
My point is that trading is extremely complex and probably the most competitive “game” a human can do, that’s why it is not an easy task for an AI that is built by humans (with a human way of thinking) to find a different way of making money.
PS. Many Neural Network AI’s are so complex that even the ones who constructed them can not understand what and how it makes its predictions.
I also asked Alex to write a little about the Artificial intelligence he is designing for the BEAST platform I think it will really help you the next time you are looking at a project or whitepaper which uses the Buzz word “Artificial intelligence”
We are using a cloud based Machine Learning provider, we chose cloud based because we crunch so much data and are continuously adding more and more data that pretty fast it became unsustainable to run it on a local machine.
This solution enables us to simply pay a bit more and get more crunching power out, really fast to you. No server downtime maintenance or bad performance. It is all done so that our users will get the best possible predictions at the most convenient way.
We have built a Neural network, one for Bitcoin and one for Ethereum.
“A neural network is a set of interconnected layers. The inputs are the first layer, and are connected to an output layer by an acyclic graph comprised of weighted edges and nodes.
Between the input and output layers you can insert multiple hidden layers. Most predictive tasks can be accomplished easily with only one or a few hidden layers. “
The task of predicting the price movements is no easy thing to do, that’s why we use 1000 hidden nodes with 1000 iterations (the number of times the Neural network tries to improve).
That is a huge number.
Regardless of the time you are reading this our AI is improving.
As I mentioned before we are always adding more and more data to our Neural Network, just look at the classes of data we put in:
Bitcoin data starting from 2009
Bitcoin data starting from 2009
- Transaction Volume in USD
- Numbers of Transactions
- Total market cap in USD
- Price in USD
- Exchange Volume
- Generated Coins
- Number of active Addresses
- Average difficulty to open a block
- Median value in USD
- Median Fee
- Block size
- Block count
Not done yet
We are also giving you the latest Buzz from Twitter and Reddit, and not just in any format as a simple bot does. No, we use AI and Machine Learning to classify every single post about the coin, if they are positive or not.
You will then be able to see what others are thinking or saying in a summarized way (you don’t want to look over thousands of posts).
And you will be able to see the change from 24 hours ago.
So to conclude and my personal opinion is that a true AI, in the Blockchain world, is something which needs to continuously be worked on, it’s something which needs to be updated and fed information by the creator. hahaha sounding real futuristic as I write this but it’s happening, the future is coming and the Blockchain industry is pushing this development quicker then anyone else. Quiet possibly our platform will be the first to crack the dream of providing a true AI to the general public and maybe someone else will.
I would recommend you join our chat, to find out more, the platform will be on this website and its development has already begun, find out more here:
Alex and I will be happy to answer any further questions you may have.