This is something we discussed from the perspective of data.

Few can question the explosive interest in artificial intelligence, especially generative AI. With LLM, image creators, and text-to-video applications appearing all over the place, there is validity to the excitement.

That said, we know there are some challenges.

To start, we are told there is a likelihood where we lack sufficient data to keep training these models. This is something that was absurd a couple years ago, considering the growth rate of the data produced. Nevertheless, these training models consume so much data that many are predicting a time when synthetic data is the only option.

Another issue is compute. We see the makers such as NVIDIA pumping out chips as quickly as possible. However, the demand is off the charts, with orders lined up well into 2025.

Finally, there is the centralized nature of all this. We are seeing more power being handed to the major players. In the West, names such a Google, Amazon, Meta, X, and Microsoft (OpenAI) keep appearing. Mega-tech, it would seem, is only getting more powerful.

Image made with Ideogram

Blockchain Moving Into Decentralized Machine Learning

When we look at the next evolution of computing, what comes to mind?

A case could be made the next breakthrough is going to be at the edge. When we look at some of these issues, edge computing offers some solutions. It also brings massive potential.

Some are theorizing that we are going to see devices at the edge which are more powerful than the standard smartphone. There are a number of initiatives to address this problem.

One who already looks to be taking the lead is Apple. The company is going to upgrade the chips it offers, potentially addressing some of the global needs.

Apple Silicon, the chips that power Apple devices, also promise to unlock massive global computing resources. Grieve says that research into Apple M2 and M3 chips shows that the hardware reaches parity with mid-tier, current-generation consumer Nvidia RTX GPUs.

Source

This would be a major step forward.

If we consider the fact there were over a billion iPhones sold, this is a huge market to tap into. It is something that could be enhanced by the concept of making money with the device. We will get to that in a second.

As the hardware evolves, so does the software.

Blockchain is going to be vital to the future of decentralized machine learning. There are a few ways this will enter the picture.

To me, the basis of this starts with incentive.

Why would anyone give their compute for AI training? While some might do it on principle, the reality is most are going to require payment. Here is where we see cryptocurrency entering. If someone's compute is required, it is necessary to get paid.

The other is to essentially mimic what Satoshi Nakamoto and Bitcoin did in the early days.

Gensyn Building Decentralized Infrastructure For Machine Learning

Before diving into this, we have to state this is not a recommendation for Gensyn in terms of an investment. Anything written here is purely for informational purposes.

With that out of the way, we see a project that was raised $50 million from Andreeson Horowitz. It started in 2020 and has the intention of creating infrastructure that allows edge devices to use their processing to help train AI models. These devices are connected with others around the world for training.

Here is where the answer to the centralization problem could enter. When we look at compute, there is a lot available in edge devices. The problem is harnessing it. This is what Gensyn seeks to resolve.

“How can you peer with another device and train a machine learning model on that device where A), the device is untrusted? B), your training model can’t fit on that single device. And C, you want the achievable scale of the entire system and unit economic outcomes as good as AWS,” Grieve said.

Gensyn’s litepaper describes the protocol as “a layer-1 trustless protocol for deep learning computation.” The Network directly and immediately rewards participants for availing computing resources to the network and performing ML tasks.

Sounds very similar to Bitcoin. In fact, it is the entire premise behind Web 3.0 infrastructure. This is simply taking it to another level, pushing the compute further out.

What we are seeing is an ongoing trend of development with the focus being decentralization. There is obviously great debate in this area. However, there is one concept that is becoming evident: decentralization is going to be a necessity.

This have nothing to do with ideology, hatred of government, or fear of mega-technology. Instead, there is a practicality to decentralization that does not appear on the radar of many.

The solution to resource constraint is decentralization. It is by harnessing the processing, storage, and bandwidth of these devices that we can keep moving ahead. There simply is not even being produced by the centralized entities to keep pace.

Again, think back to the data situation. If I said, in 2020, that we are not generating enough data, most would have thought me nuts. Only the few who foresaw what was going to happen with these AI models would have seen the issue.

Now we are facing the issue of processing. It seems absurd considering that Nvidia is churning out $50K H-100 chips like cereal. Yet, as we can see, the demand is there.

My view is blockchains are going to succeed for a couple practical reasons.

To start, the open nature of the data is imperative to many companies going forward. Second, the storage capabilities of said data is also needed. Finally, these system will provide the ability to enhance the edge capabilities. It is likely blockchain is a part of the solution in what ties all these devices together.

Things are moving quickly and a lot is in the works. Gensyn is one start up seeking to unlock a massive amount of compute.

These are major issues that the AI world is facing. While the centralized entities will keep expanding as quickly as they can, a lot more is needed.

Here is where blockchain can step in. It can offer solutions to massive problems by providing the necessary infrastructure.

---

What Is Hive

Posted Using InLeo Alpha