This article is not an endorsement of the Lightchain AI project nor the potential for its token. It is included in this article for informational purposes only.
AI is going to be taking over the world. We are going to see this applied to the digital realm before extending into the physical. It is something we covered for some time.
Over the last year, we have beat the drum for AI being integrated to blockchain. One of my primary motives is to do what I can to pull as much away from Big Tech as possible. Naturally, this is a difficult poposition and will take a collective effort.
To me, one of the keys to the future is to ensure that ownership of the means of production, much of that which will come from automoation, is outside the hands of a few powerful technology companies. With technology, we face a winner take most situation. The power of network effects is well known at this point.
For this reason, when something has the potential to sway things in a different direction, it is worth nothing.
Here is where we get Lightchain AI.
The main reason I put the disclaimer at the top of this article is we are seeing Lightchain conducting an ICO. This is important for a couple reasons.
First, people need to do their own research to see if something is valuable. A white paper does not guarantee success.
A second variable is the fact we are dealing with a start up. The lesson from 2017 and the ICO craze was that, even when the intention was there, failure was the norm. It is not easy to build a successful technology start up.
The idea is to use Ethereum as the network for AI services. It seeks to model blockchain consensus, with a slight twist.
But what truly sets Lightchain AI apart is the introduction of its Proof of Intelligence (PoI) consensus mechanism. Unlike traditional consensus systems like Proof of Work or Proof of Stake, PoI rewards nodes for performing valuable AI computations. Tasks such as machine learning model training, inference, and optimization will no longer be limited to centralized systems. The future is decentralized, and Lightchain AI is making it happen.
What stands out here is the focus on decentralizing compute. Having machine learning model training outside the largest tech comapnies is a good step. Also, the fact that inference is being mentioned shows foresight into a major bottleneck.
In my opinion, inference will be decentralized since the amount of compute required will overwhelem even the largest of companies.
Here is the crux of it:
At the heart of Lightchain AI is its Proof of Intelligence (PoI) consensus mechanism. PoI brings a fresh perspective to blockchain consensus by rewarding network participants for performing AI tasks. Imagine a network that doesn’t just validate transactions but actively contributes to global AI innovation.
Bssically, it is taking the incentivization mechanism that blockchain utiliizes and applying it to AI tasks.
This is a blending of two important technologies. The focus is upon the foundation, something that can see AI agents and a host of other services built on top.
Lightchain AI leverages Ethereum’s decentralized system to facilitate secure, transparent, and tamper-proof AI-driven transactions. Whether it’s training an AI language model or executing real-time AI computations, Lightchain AI offers a scalable and efficient solution powered by Ethereum’s blockchain backbone.
It’s not just about what Ethereum can do for Lightchain AI—it’s about how Lightchain AI can take Ethereum to the next level by enabling intelligent, AI-driven computations that unlock new use cases for the blockchain.
To read the press release for Lightchain AI (where the quotes were taken), click here.
We will have to see if this is a viable solution. As mentioned, there were a lot of big promises made in 2017, some of them well intentioned. However, the variables required for success are many.
That said, this is encouraging. Developing systems that can incorporate compute for AI training and inference is crucial. This is where Big Tech has a distinct advantage. Smaller entities do not have the resources to drop billions for GPUs. Some of this is being offset by the fact that small language models are advacing in a way that makes them viable.
Inference is another story. Here we have a compounded issue.
At present, Big Tech monopolizes this. Anything that is generated is on their servers. The outputs, i.e. syntethic data, is fed back in, further enhancing their models. It is closed data that nobody else has access to. We can only imagine what OpenAI is capturing each day with all the prompts done using ChatGPT.
Jensen Huang mentioned how he believes inference is going to be a billion times larger than compute for AI training. This is logical if we consider a world where there are tillions of prompts being done each day. An automated world sends numbers flying.
We are dealing with outputs which the basic unit is cognition. This is reached by feeding in the fundamental unit of data (tokens).
One way to think about this is by framing it against Google Search. There are roughly 8.5 billion searches per day. Consider it just 10% of that switches to prompts. That is a great deal of data being generated, all of it feeding Big Tech.
Could something Lightchain AI change this? Certtainly there is the potential to build AI services on top of their infrastructure. By incentivizing participants, we can see the democratization of activity.
For me, however, the question is how does this interact with Ethereum and what about the transaction fees?
Nevertheless, it is positive that people are building alternatives.
It is only a matter of time before AI comes to blockchain in a significant way.
