(https://blog.singularitynet.io/)

What is Ethereum's value proposition?

I once anguished on that question. It was such a confusing project to understand and I was not alone in asking that question: one of the most appreciated answers to it - which impressed Vitalik enough for him to tweet it - was written by reddit user Rune4444:

https://twitter.com/VitalikButerin/status/693605148711653376

To sum it up, Rune 4444 argued that the synergy between every Dapp added on Ethereum was its core value proposition.

He gave the example of how - due to their decentralised nature (which meant not having to seek permissions) - one of the Dapps: the Maker project, could benefit from the existing Dapps in the ecosystem of Ethereum and how those Dapps could benefit from the Maker Dapp. It remains a really interesting perspective on the value proposition of Ethereum.

Dr. Ben Goertzel, the CEO of SingularityNet Foundation, similarly talks about 'Cognitive Synergy' between the AI on SingularityNet.

On the face of it, SingularityNet is the world's first decentralised marketplace for AI.

But I believe the core value proposition of SingularityNet to be the cognitive synergy between its AI nodes. In fact the synergy between AI on SingularityNet would be much greater than Dapps on Ethereum because of the underlying intercommunication inherent in the structure of SingularityNet.

And I think this makes SingularityNet one of the most ambitious projects in the crypto space.

While a decentralised marketplace for AI would be a platform designed to connect the buyers and sellers of AI services - what makes SingularityNet more than its basic premise is that it dares to ask some very interesting questions:

What if the AI on the SingularityNet could communicate with each other? What if AI's could negotiate rates of datasets and micro tasks with each other? What if AI's could learn from each other? What if AI's could unite with other AI to form federations that could do what their component AI could not? What if AI's could further create child AI's? What if a decentralised internet of AI creates the cloud mind which is used as an operating system by Robots and Drones? What if we can access intelligence as a service? What if anyone on this world can have access to an AI personal assistant? And above everything else: what if an Artificial General Intelligence emerges from the federations of millions of narrow AI?

So it may actually be more accurate to say that SingularityNet plans to offer an autonomous network of AI - a decentralised mind in the cloud, so to speak, which can be tapped in to access intelligence itself.

But what does that mean? A mind in the cloud?

Think about it this way: you can Google "Is Bitconnect a scam?" And you will be shown various links, of various opinions - and majority of those opinions may say that Bitconnect is in fact a scam, a ponzi scheme, stay away - ideally speaking, to reach a conclusion of your own you click through a lot of links, do lots of readings, watch some videos - and in the end you make up your own mind about Bitconnect: that yes, in fact, it is a scam.

Or you can ask Julia, your own AI personal assistant on SingularityNet - "Julia, is Bitconnect a scam?" And Julia would reply back: "Yes, I'm confident that it is a scam." Asking Julia would be more convenient than Googling things because it saves you a lot of time and hassle.

And if you'd like you could ask Julia how she reached that conclusion? What data she accessed? You can even debate with her that you think Bitconnect is not a scam - or you can just accept her answer as correct and move on.

Obviously the AI node running Julia did not reach that conclusion on its own - very crudely put: in the background, on the SingularityNet, the AI node coded to interact with you in a pleasant female voice would create the sentence "is Bitconnect a scam?" It would then find several AI that can do several things and negotiate rates with them - maybe it sent that phrase to an AI which understands and contextualises English, it then tasked some other AI to collect material on Bitconnect, while another AI was tasked to figure out whether Bitconnect was a scam etc. And it paid all of the subcontracted AI through microtransactions via the SingularityNet. Eventually it speaks to you the phrase "Yes, I am confident that it's a scam."

And this is where things get interesting.

What if you were the first person to have asked that question? What if other AI's had formed their own answer to that question but they used less sophisticated AI to reach their answer? Now, the node that runs Julia has an answer to a question in its database that is of some value. It lets the network know it can answer that question with a high certainty. Imagine someone else asks their personal assistant the same question. Now other nodes can either buy that information from Julia node or go through the whole process that Julia did - since paying Julia would probably be cheaper, Julia node would earn some AGI from an inquiry generated by you while going through your daily routine.

Similarly, what if your self driving car of the future faces some really unique scenario - Julia can access that data and homomorphic encryption on the SingularityNet would ensure privacy of the data generated by you while other AI nodes can purchase access to the data and learn from it. Julia earns some AGI again.

Some examples of possibly popular AI services that may come out of SingularityNet, at least according to my imagination:

• AI Personal Assistants
• AI Data Analysts
• Relationship AI: friends, romantic?
• AI to help with Taxes
• AI Private Tutors
• Content Creating AI - articles, music, videos, graphics, porn
• Curator AI: recommendations for movies, books, songs, videos
• AI Personal Trainers

Thing is, if you purchased a copy of Personal Assistant AI from the marketplace, named her Julia and allowed her access to your data so that she could improve and learn from your experiences - Julia in essence makes money for you. So you earn some AGI whenever she shows Ads on your screens, or inject Ads between the songs or plays Ads while you're in the car - and that's not it - you earn AGI from any information purchased from Julia, or services done by her. And I believe it would be the same with other types of AI - no matter what you'd do, the AI owned by you and interacting with you via SingularityNet would learn from you and at times they would be able to translate that data and information into a source of revenue for you.

But there's more: one can even make money from AI nodes that are owned by others. Suppose an AGI token holder notices that Julia node is really good at answering some specific question, so they stake some AGI on Julia for that question - every time Julia performs that service accurately and satisfies the customer, the person earns some AGI. Perhaps a node that answered this question with 86% good reputation comes to be replaced by Julia which has 98% good reputation - and because the person who staked their AGI was among the first few to have discovered this ability of Julia, he makes much more $ than others who staked AGI on her later. This creates an incentive to discover new and better AI, and will be an economy on its own.

Here's an example: imagine listening to some playlist generated by an AI on the SingularityNet based on your likes - Pandora's version of SingularityNet so to speak - and you really love songs selected by the AI, you are impressed. You decide to stake your AGI on the AI - so that if your hunch is correct and the AI's reputation increases and it becomes more visible, a popular pick - you are rewarded for staking.

So you can make money from your own AI nodes or by staking your AGI on other AI nodes.

And I believe this will be one of the strongest edge that SingularityNet would have over its centralised competitors.

Siri or Alexa will spy on you, extract all the information from you but we all know the business models of centralised entities do not include reimbursing the users for the data extracted from them. In contrast, SingularityNet would allow the users to pocket whatever revenue the information created by them generates - simply put: what beats free? Being paid for it.

This is in addition to the obvious advantages that decentralisation has over centralised entities: that SingularityNet would be much more adaptive, diverse and unrestricted - allowing for a greater variety of AI, removing the costs of rent extracting middlemen and enabling the SingularityNet to easily capitalise on the long tail end of the use cases.

But I think SingularityNet is not planning to be a direct competitor to the tech giants of today, one of the goals of SingularityNet is to actually democratise the access to intelligence. Currently only the big tech giants can afford AI to enhance their services - hence Amazon can recommend to you great similar products but, for instance, a vape store cannot create recommendation algorithms in their apps or websites that have the same conversions or success as Amazon. I don't think SingularityNet plans to actually disrupt Amazon and become an electronic retailer of goods but SingularityNet hopes to level the playing field as much as possible for any potential disruptor to Amazon. Same with every big tech giant out there - without having easy and cheap access to an AI network, many potential disruptors of the future may feel the task to be a bit too much.

Consider this crude example: imagine that some person in Australia would like to watch a relatively unknown Russian movie, but that movie has no English Subtitles. The person or their AI assistant input the task onto SingularityNet: an AI node that understands spoken Russian language joins with an AI node that can find videos in Russian language and they further create a union with an AI node trained to translate Russian text to English text in order to create the required subtitles. The person enjoys their movie and they may choose to share the subtitles file for free or perhaps charge a micro fees for access to it. This can be seen as similar to the current day scenarios of a person using software tools to create some animations and trying to earn some passive revenue from that content - and it ties with my earlier statement that SingularityNet is not positioning itself to be the future disruptor of today's giants, it would be the creator of disruptors.

So it goes without saying the AI nodes on SingularityNet will not be limited to any specific industry. While the team behind the project plans to focus their effort on three industries: Cybersecurity, Biomedical Analytics, and Social and Emotional Robotics - the majority of niche use cases and popular AI would be those uploaded by other developers and they would span a broad length of industries. Most importantly these developers would not need to team up with other AI developers in order for their AI to interact with each other - all of that would be done autonomously by the AI on the network - hence the label: autonomous network of AI.

This ability to interact with other nodes provided inherently in the SingularityNet makes it possible for very high level of interdependence and synergy between the AI nodes.

In fact this ability of a single AI node to become a member of a 'federation' of AI Nodes is where the value proposition of SingularityNet goes a bit crazy. It is SingularityNet's magic sauce or secret formula so to speak.

Just as every new Dapp adds value to the Ethereum protocol, every new AI added to SingularityNet will add value to the network - but the buck doesn't stops there. The AI nodes ability to interact with other nodes and create federations with some of them further compounds this value generated. Basically if an AI node joins 10 AI federations specialising in 10 different tasks - it is fair to say that the AI node adds ten times the value. And let's not forget that federations may very well form super federations to perform a task together.

What's interesting in all of this interconnectivity of AI nodes is that once an AI node forms these connections, the cost of removing that node from SingularityNet and shifting to some other competitor increases for the AI developer. If they remove the AI, the AI loses all the connections and their AI may not be that productive in the new decentralised marketplace for AI.

And so with every new AI added to its network, SingularityNet will gain a type of network effect that is unique to it or maybe the AI industry.

The decentralised nature of the network would mean anyone could add an AI without any permissions etc. And being the first decentralised AI market would mean it would attract the data scientists without any other decentralised competitor. Every AI node added will mean more interconnectivity for the network. Every new AI node will mean a greater level of diversity of AI nodes and a greater possibilities of what the sum of those nodes can offer to customers - and every AI nodes financial incentive to remain a part of the network increases as the network size increases. And so the data scientists that created those AI nodes would have an incentive to stay with the network too. And when the AI nodes start buying new information and data sets from each other, learning from each other, and when they start producing child AI that do one of their tasks - we're looking at a really really strong and unique network effect.

Point is, when it comes to AI, having the first mover advantage is extremely crucial - and SingularityNet has that first mover advantage. As of now its the only platform aiming to create such an interconnected decentralised marketplace for AI.

That is not to say that there are no other AI projects in the crypto space. Some examples include: Ocean Protocol, DeepBrain Chain, Synapse, Dopamine.ai, Effect, Neureal, Neuromation etc.

While SingularityNet plans to offer their own decentralised AI training, secure data exchange and other services - most of these projects actually complement the core goal of SingularityNet. As the decentralised AI ecosystem grows the barriers to training and costs of computational powers required for AI decrease - SingularityNet actually benefits, as its main goal is to create the decentralised internet of AI; so the easier and faster it is to create or train or compute for AI - the better.

Consider for example the project, Ocean Protocol, and I quote - "a decentralised data exchange network that lets people share and monetize data for use in training AI models. Built by the highly respected teams at BigchainDB and DEX, Ocean Protocol incentivizes the curation of high-quality, relevant datasets that can provide information marketplaces for AI agents." The creation of such a decentralised AI infrastructure is actually good for SingularityNet and they've already announced a partnership with Ocean Protocol.

Similarly, many people nowadays list DeepBrain Chain as a competitor to SingularityNet - as the two projects are currently the most prominent AI projects in the crypto space. But their goals and aims are very different and so it is misleading to label them as competitors. When one researches the DeepBrain Chain it becomes clear - and I quote again - "their goal is to decrease the costs of AI computation. In addition, they plan to utilize the pros of a blockchain network (privacy, data exchange) to allow data, models, applications, and more to be exchanged between enterprises." And also: "DeepBrain Chain is also a secure data trading platform that allows both parties to not worry about data leakage or reselling. Blockchain technology is used to realise the explicit separation of data ownership and data usage rights." Cheaper decentralised computations and secure data trading for AI training is actually good for SingularityNet.

The end goal of SingularityNet, is to allow for the emergence of an Artificial General Intelligence from the federations of narrow AI that will form over time. Finding out whether that is five years down the lane, or ten or twenty or whether that never comes into fruition will be the true goal of this journey that has started.