Introduction to Artificial Intelligence on Blockchain Part 4

AI and blockchain working together.

 

Possibilities of blockchain and artificial intelligence technologies working together

Even today there is a field in artificial intelligence that develops algorithms capable of processing data in the encrypted form. This means that a combination of artificial intelligence and blockchain could potentially work with data without making it vulnerable to attacks, making many industries much safer compared to where they are today.

In addition to this, the logic for decisions that artificial intelligence makes can be hard for people to comprehend fully because artificial intelligence can simultaneously analyze a very large number of parameters that to people may seem to be disconnected. Artificial intelligence can learn with time about which parameters are important and which are not. For humans, at least in the foreseeable future, it will be important to be able to have a track record of these decisions and to be able to analyze them. For example, even today corporations like Walmart use artificial intelligence to make decisions about which products should go into which stores.  This is also where blockchain can come to help. When decisions become a part of a blockchain, there is a record that is immutable and that has timestamps, which would people to audit the decisions knowing that there has been no tampering with the logs.

 

Management of blockchain with artificial intelligence

No matter how many predictions and useful data artificial intelligence can create and manage, its applications will be very limited until it gains the trust of decision makers and later of the general public. Trust is one of the real breakthroughs of the blockchain technology, which is why storage of records about decision-making process on a blockchain is so important. In addition to this, artificial intelligence could also manage blockchain networks of the future. Traditionally, computers and technology have been much faster than people, but often speed came at significant costs. One of the biggest costs has always been about creating instructions for the machines. Previously, if a computer did not get a code that explained how to perform a task as a series of “if/then” statements and computational steps, the computer could not get anything done. Applied to blockchain technology, this means that running blockchain networks requires a tremendous amount of processing power and computer memory.

As an example, miners on the Bitcoin network first compile transactions into blocks of Bitcoin blockchain and then seal these blocks with hashes. In a cryptography algorithm, a hash is a string of data that it shorter than the original data.

Hashes have a lot of useful properties. First, it is easy to verify data if you have a hash for the data. Second, you can’t recreate the data if all you have is a hash for it. Third, hashes are small compared to the data they describe. For example, block #522787 of the Bitcoin blockchain contains information about 2410 transactions, yet all this information is contained in the hash for the block, 00000000000000000025468b03a643471dd3b439d812d170a8d653a76dcdac17. Because the Bitcoin blockchain is fully transparent, you can see this and other information, including hashes, timestamps, transaction amounts and more, on the official Bitcoin blockchain explorer at https://blockchain.info. Finally, hashes for even very similar sets of data differ dramatically. This is why it is not possible to change even one digit in the information about transactions that have occurred on the Bitcoin network since its origination in 2009. Every block on the Bitcoin network contains the hash for the previous block. If someone were to change even one digit, the hash for the block with the new information would be completely different from the original hash and the block with the altered information would simply not “fit” into the existing blockchain.

 

Proof-of-work as an example of an algorithm that is not “smart”

However, sealing blocks with hashes also has a very big disadvantage. Creating a hash for a set of data by running it through a cryptography algorithm is a trivial task for most modern computers, including regular desktops and laptops. For this reason, the Bitcoin network has a parameter called difficulty. This parameter specifies how a resulting hash for a block should look like. Because a set of data can only have one hash, miners add a number to the data about Bitcoin transactions, generate a hash for the set of data + the number and see if this hash matches the difficulty conditions. If it does, the number that they have added (called nonce, which is short for number used once), becomes a part of the block and the blockchain. If the hash doesn’t satisfy the conditions, the miners try another number and generate a hash again. Obviously, this is all done with software and there are no humans involved trying to come up with random numbers, but this doesn’t change the fact that in essence miners on the Bitcoin network are playing a lottery by guessing random numbers.

While the system does work and does provide security for the Bitcoin network, the way it does so is hardly “smart” or “intelligent.” Another problem with mining on the Bitcoin network is that to come up with winning hashes, miners burn a lot of electricity. According to an article in the Guardian, in 2017 all Bitcoin miners globally have burned more electricity than did the country of Ireland during that same year. Other than providing security for the Bitcoin network, this electricity did not do any good and contributed nothing to the society.

An artificial intelligence algorithm could potentially tackle the issue of hash creation in a much more efficient way, without wasting electricity to play a lottery.