Lightning: Why AI Will Learn From AI
It’s 4:45am. Your phone’s alarm goes off a bit earlier because, of course, it knows you have a flight in a few hours, the TSA line has been longer than normal on these days, and the highway will be extra clogged, especially since snow will begin to fall in fifteen minutes. You’re only half awake, but that’s ok, your car is driving for you, navigating the nearby drivers that are irresponsibly texting while “manual” driving, something nobody quite does anymore. Your car fits right into tightly orchestrated traffic and avoids airport lines by the same methodology, not too unlike the air traffic controllers that once ruled the friendly skies you’re about to fly.
Once upon a time, most of this utopian dream may have been considered far-fetched. In fact, I’m sure a quick search on Netflix could entertain hours of indulgence in these stories of futurism. However, this reality is much closer today than it was even a decade ago with recent advancements in cognitive understanding through Artificial Intelligence. However, one question still remains — how will we coordinate so many independent AI services, like self-driving cars, in unison? How will we share knowledge from one service to another? Will a new, free economy evolve that allows services to interact with each other, or will we continue to have a value-chain that extends from the knowledge producer to the consumer?
As cars and other agents reach autonomy and behave as independent entities transferring knowledge between themselves, then a framework for the transfer of value between these entities must exist.
Today, we have many ways of acquiring information, and most of the time we believe we’re acquiring such information at no cost. Checking the weather is as simple as opening a free app, or is it? Following Google Maps doesn’t require me to pay for their service as an end user either, or do they? As humans, when we interact with online service providers, we are accessing information that comes with a cost (albeit tiny), and these service providers are compensated for that cost through advertisement incentive. If I check the weather on Google or use Waze to navigate difficult traffic while driving, though the app itself may be free, I’m still paying for the use of their services.
App developers know this process all too well. If a developer creates an app and wants to provide some sort of data that the developer themselves hasn’t created (search features, weather information, driving directions, etc), then they can “plug in” to a 3rd party to obtain that data. However, most data and service providers will either require some sort of attribution (a “Powered by Weather.com,” for example) as a form of advertisement for their business, or require the developer to pay for that service. At CloudSight, we allow developers to use our services for image recognition, though there is a cost associated for each image that is recognized. Because the developer is faced with the extra cost of providing that service, the users incur a cost of “payment” as well. Whether it’s monetary or providing their information that the developer can sell to 3rd party companies, there are a lot of ways in which consumers “pay” for services. However, as users of these apps, we “pay” somehow for the use of them, through the apps and to the data providers. The data providers have a cost for providing that data, and so the cycle continues.
Yet, in the story above, where many AI agents interact with other AI agents, and an AI agent that you own (as in the case of an autonomous car) needs to access these services, how would you manage the cost for interacting with all of them? Better yet, what if the car owned itself? How and what do we “pay” for, so that our AI agents work correctly? Today, this would require signing up for a number of different services either as the developer of the technology, who would then need to forecast how much each user would cost, or as the user, who would sign up with a credit card and manage the cost on their own. This setup is very cumbersome for the user, though, and Tesla has experimented with the model of absorbing the cost for the cellular data connection on the vehicle since they also derive value from streaming diagnostic data back to their headquarters.
We enjoy the Internet for its utility of allowing us to access information with greater efficiency than the world has ever known, and AI can leverage this communications network in order to question its immediate environment and learn in real-time. However, in order for that to occur, we also must have in place some sort of mechanism to transfer value from data provider to data consumer, in a distributed, trust-less, decentralized manner, such that any AI entity can transact with any human, service provider, or another AI entity, without any pre-existing trust, and at will, through the information economy.
As it turns out, for nearly a decade now, a new technology layer has been brewing on top of the Internet. This new piece of the network allows monetary value to be transferred between parties through a trust-less, decentralized, distributed network. Value can be transmitted like an email, or programmed through a smart contract, such that two entities transacting value between themselves need not have prior knowledge or trust, and can engage worldwide wherever an Internet connection exists. No, it’s not blockchain, it’s called Bitcoin.
While the debate on cryptocurrency carries on, one thing is clear: this new tool can enable advanced value interactions that were previously impossible. When an AI agent has a question, who (or what) does it ask, and how much is that answer worth? As an example: if your car needs to know what an unknown object is in the road ahead, what if it could ask the surrounding cars? If those cars don’t know, could it reach a data provider, and if so, how much would that cost? Once it knew, could it be compensated for that fee by other cars, also asking the same question? As humans, we interact this way all the time — we receive social credit, pay tuition, buy books, or allow a barrage of advertisement. Either way, there is a value transfer for an exchange of information. As the information economy grows beyond humans and to Artificial Intelligence, we think it’s necessary to lay a groundwork for that value transfer.
Today, we’re announcing support for Bitcoin Lightning payments on our platform. Why Bitcoin? Why not our own utility token? Bitcoin’s utility extends far beyond its value in messaging monetary value: it’s secure, it’s decentralized, and it enables two parties (including AI) to engage in a transaction where they don’t need prior trust. All of these features are critical for autonomous entities to communicate with each other, enabling information exchange in a new AI-based economy. Should apps, autonomous AI, or any other application need to gather visual knowledge, CloudSight will be able to serve and exchange with these queries through this new monetary medium. We are very excited for the future of autonomous AI, and we believe this is a critical step in the development of this new technology.
⚡️
So, how do I get started using Lightning on the CloudSight API? Check out our other blog post, “Introducing Lightning Network on CloudSight,” which has a step-by-step guide on how to use this new API.
