Topic: Blockchain & Machine learning (Read 380 times)

Hahaha! The shower thought of all shower thoughts. Let's have a moment of silence, and internalize that insightful question in peace.

If not blockchain then what else? what would you trust to store your consciousness on? and yes it is far out which even more exaggerates the point that AI has no business being on the blockchain in this point in time as it has no real use for it.

Would you really hypothetically speaking put your consciousness in the hands of a central authority?

If only AI or blockchains would work that way.

While I have no idea how consciousness can be stored -- no one has, humanity hasn't yet the slightest clue what consciousness even is -- I'm fairly certain it won't be on a blockchain.

Maybe one day humanity will be able to externally store memories (beyond mere audiovisual reproductions) and minds (be they human or artificial) but that technology is probably still a few generations out and won't be technology we'll be able to fathom.

Blockchain, as implemented in Bitcoin, is inefficient for "other use-cases", look at Ethereum/ICOs, because POW requires a large amount of computing resources. But it's not "messy".

There is no "technology" named blockchain. it is just an inefficient messy data structure used in bitcoin, nothing more.

This was something I was talking about in another thread also about decentralizing Github, there are benefits to a normal distributed system the real question is how long do you need the file to be maintained for? Is security of importance? If so blockchain would allow for such security to be achieved in a trustless manner. Do you need the file to live on forever store it on a blockchain with incentives, short term? go with a regular distributed system. For an AI wanting to obtain memory states and live on forever a incentivized blockchain would surely be the way to go, say if we reached a point where we could store our memory states on the blockchain such as altered carbon, would you really want to store that in a regular distributed system? No you would want to obtain the security that blockchain provides in a trustless manner, as our memory states would need to live on forever.


I think to reach true scalability we need a bit of both, while still not sacrificing security in the end;Ying and Yan.

Read on my idea here, https://medium.com/@scroda/scrodas-sub-zero-protocol-walkthrough-c1677f3d60fa

Still true I am no in agreement neither that AI should be used on the blockchain as of yet, only in extreme cases such as mentioned above.

I think the "problem" with "blockchain the technology", ie. outside the context of cryptocurrencies is two-fold:

1) Permissioned blockchains still seem rather pointless. For most use cases classical distributed databases (or distributed systems in general) are doing a much better job. Even in the iRobot takedown scenario the main properties that were missing are security and robustness. That you can do in a distributed manner without using "blockchain the technology", in a much more reliable and efficient way at a much lower cost.

2) Permissionless blockchains without PoW-based consensus algorithms so far are often either not really permissionless or not really secure or both. Most of the time you either have (a) committee-based validators or an other form of central coordination which mostly rids of permissionlessness -- which leads back to (1) above. Or (b) variants of Proof-of-Resource like PoS that -- as of now -- are still mostly work in progress in terms of security and robustness.

I guess there could be ways in which permissionless blockchains could support AI and vice versa but I think the overlap where they would complement one another are rather limited. I'm afraid in most cases you'll end up with either a shitty AI or a shitty cryptocurrency.

Where in my statement do I show to not have understood such obvious assumption? Any consensus mechanism which is highly demanding on computational resources will obviously have fee's, they go hand in hand 

No, you misunderstood. I was talking about the costliness of POW/hashing itself, which would require a large amount of computational resources.

It's inefficient for those use-cases.

This is where I would have to agree, there are many different consensus mechanism coming to light as time goes by. Most of you are assuming it would be costly because of PoW or similar mechanism in which require transactions fees in a fully decentralized network, that is a foolish assumption.  Now the real questions boils down to scalability and not about it being costly, with many start ups focusing on providing scalable solutions this should one day be out of the question also, so at this point in time the only real question to ask is how fast does it need to settle? is security more important than time? most of the time it is, as the whole process does not need to be decentralized only the end product.

Even so this does not fall under Bitcoin's Development and Technical Discussion. This is not a Bitcoin related question and Blockchain is not solely meant for use in cryptocurrency.

and as far as Blockchain and Machine Learning in relation to AI, it could allow for AI's to just become smarter and smarter overtime without them really being able to be stopped in a far out future. They could have "memory states" on the blockchain in which they would be able to access at any point in time, without anyone having the power or access to really control them centrally, most organizations have their code and are easily able to monitor or stop any activity on sight but now if we where to use blockchain they could just save their "memory states" on the blockchain and access them from anywhere with internet connectivity. If you see I-robot where Will Smith has to go to a central location to stop the bad AI well in such a case it would prevent such thing from being possible. Still something for the far out future where we are in a world where AI just keeps on advancing who knows.

What do you mean with 'cryptography technology' ?

Cryptography is a mandatory component of the blockchain.




While you are right with some use cases, especially the patient chart, i disagree with all of these being unsuitable.

Digital Identity, would for example be used to 'sign' documents or contracts using a private key.
You could also implement something a viewing key to read the information (like with monero). So it wouldn't be publicly available for everyone and you'd have a proper handy system.

I believe it would depend on what your definition of a blockchain is. Because if your definition includes arriving into consensus through POW, then I believe all those "use-cases" would NOT need a blockchain, but a database.

These applications would not be particularly well suited for blockchain technology:

• Digital Identity
• Data management between organizations
• Patient chart
• IoT

The above would be better suited to use cryptography technology. There is little advantage to using blockchain technology for any of the above, because once an entity has access to the information, they can trivially replicate it in unencrypted format.

Supply chain and real estate might be two good useful applications for blockchain technology, especially the later. Real estate becomes tricky when liens are involved because a lien can result in ownership being unilaterally being transferred, and liens can sometimes be put in via judicial action. Parcels can also be split up, which would be difficult to account for with blockchain technology. It would also be a problem when private keys are lost and/or stolen.

Supply chains using blockchain technology would rely on the trust that private keys are not being sold/transferred among market participants. A good use for this would be to trace the history of diamonds to ensure a particular diamond is not a "blood diamond", however if private keys are being transferred, those trafficking blood diamonds might be able to launder diamonds into the supply chain.

Plus point out how costly it is for the "blockchain" in each node in the system to come to consensus that its data is what was agreed upon by the whole network as "the truth", DECENTRALLY.

@DaveITN,

Classic problems better get their classic solutions rather than being twisted with anything that shows up as a buzz word in the yellow tech media. Permissioned blockchain is centralized and depends on the honesty of participants, but if there is no adversary why should we care about immutability? And when we don't want immutability what would be the reason behind using an inefficient data structure such as blockchain?

Even immutability alone is achievable ways cheaper and easier with classic database systems which resist against the unfamous single point of failure problem nowadays through clustering features.

supply chains, are better off using permissioned chains with fewer nodes. there is no need for such huge decentralization.  this also allows there to be no transaction fees which is vital.  3- 5 nodes is quite enough for a supply chain blockchain.

private chain

no fees

just spin up 3 nodes on AWS.

1 controlled by devs
2 controlled by farmers association
3 controlled by distributor
maybe another one controlled by retailer.

then periodically write the state on ethereum or BTC blockchain for notarisation

there it is.

Yes but you can use the blockchain, wether the bitcoin blockchain or another one as a pretty powerful record keeping and transactional machine.  Cryptocurrency act as the financial security that ensures this record remains secure (nobody wants to lose money) it is two sides of the same coin. no, the only application of blockchain is not just currency transactions. but also secure information.

Look, decentralization and/or trustlessness does not mean replication or resisting single physical point of failure. To keep a system truly decentralized you need to let it be open and permissionless.

Now, for our open/permissionless/trustless/decentralized system either there is some value produced or not. If there is no value it would be rather a dummy absurd system with no purpose and no utility and if we are dealing with some kind of value we need to keep our system secure against adversaries and spammers.

But adversaries are complex entities that follow their goals with very sophisticated patterns of behavior. So we need game theory and incentive system: a cryptocurrency token is born, congrats.

It is possible to have smart contracts and/hence an unlimited number of applications supported by cryptocurrencies but blockchain alone? Of course not! It would be just a nightmare! Just forget about it.

IMO those applications can indeed be used decentralized.
The supply chain, for example.

You would be able to track the harvesting of the cottage, the transport via ships, the creation of yarn and finally the manufacturing of a shirt completely transparent.
There are multiple entities involved in such a process. And with a blockchain tracking all of these steps, there would basically be no central authority.


Yet, there is not a single project which does that.
Anything one can find at the moment, indeed just uses blockchain as a buzzword. But i am confident that we can have several applications backed by a blockchain in the future which have absolutely nothing to do with cryptocurrencies.

I'd say not a single of the mentioned applications needs bitcoin at all.
That actually was the point - to have applications for a blockchain excluding cryptocurrencies.


Just because centralized databases are better at some points, this doesn't mean that they should be used for all of those applications.
One could say the same about bitcoin. Why use a blockchain if you can use a bank ?

The trustless and decentralized aspect is what makes it special. And this can also be desired in a supply chain, identity management, IoT, etc..