The Golem Network: One AI to Rule us all?

Golem Network

An All-Seeing eye?

Are we about to have a real Skynet in the near future? Not quite, but the Golem Network reportedly wants to eventually create an AI system that can do and see just about anything. The Golem Network is a blockchain project out of Poland which hopes to be, in its own words, “the Airbnb of computing.” As of now, Golem exists as a decentralized network of computers that share resources to complete projects that take a lot of computing power. The only working part of this on the network at this time is related to CGI rendering or 3D computer graphics that are impossible to do on your everyday computer. You might be saying now: well, that’s all great, but how does this have anything to do with Artificial Intelligence? It’s true that this isn’t clear upon first glance at the project, but when one considers the overall implications related to having a global network that is effectively one large supercomputer, the answer to this question becomes clearer.

Rising Costs, Rising Power

It’s often reported that any project in Artificial intelligence drains a lot of computing resources, particularly related to the types of computer chips that it needs to function well. One Computer Science professor pointed out that the biggest stumbling block for AI in terms of computing power lies in building learning or predictive models. Think of this like an AI model that is designed to recognize certain things in pictures. The hard part of designing this model is designing the math behind it so that it consistently works with a very low percentage of error. More precisely, AI scientists often use supercomputers as well as powerful GPUs or Graphics Processing Units to be able to do the math behind Machine Learning. It should be noted that this isn’t the only thing that these powerful computers and GPUs are used for in AI, but it is a prime example of what they are used for in AI.

All in all, this process involves large sunk costs that the company won’t often win back, at least not quickly.

In comes, the Golem, forged from the Ether

Here is where the Golem network could possibly prove to be beneficial.

At its core, if we keep in mind what was mentioned above, the Golem network is actually like the Airbnb of shared computers. People rent available time and available power on the network.

If you’re offering up your computer, you receive Golem tokens as a reward.

As was also mentioned above, AI teams, especially those that deal with complex algorithms, require more computing power than averagely priced computers can give. If you know a bit about the cryptocurrency world and how mining works, think about starting an AI project as being somewhat analogous with the difficulty of setting up a mining operation.

Except, it’s more difficult and more costly.

Both need a lot more than your laptop can give.

One World, One Computer

All in all, with Golem, you have one giant supercomputer.

Just think for a second about what AI projects could do if the network reaches a critical mass of even a million computers. All in all, everything could possibly become faster, more efficient and easier to accomplish. For example, the iterations of testing required to help an AI system reach its ideal output data, could become far shorter.

Even more importantly, as some experts already suggest, if the problem of needing such special groups of computers could be erased, then AI would finally be something that any company could work on. As of now, the power is reportedly in the hands of those that you’d probably expect it to be in.

Google and a few others own almost all of the AI capability, as of now.

So what could bring it out into the light?

The answer lies in Golem connected with another, previously discussed advancement in computing.

Golem 2.0: Quantum Edition

We’re back to Quantum Computing! If data can exist in two places at once, then everything with computers could be two times faster, at least. If the Golem network becomes a standard, then the speed that it already brings could be enhanced yet again by a factor of two or more. With the biggest limitation to AI being professed to be computing power, it is interesting to theorize what these two advances could bring. Perhaps Golem plus Quantum Computing could be the answer to the singularity, perhaps not.

All in all, with Quantum computing already existing and Golem already hosting a successful project in the vein of CGI, this is not a completely far-fetched supposition to make.

On the Edge lies another

As with any form of analysis, however, when we weigh the advantages, we have to weigh the other side of things as well. Edge computing appears, at first glance, to stand as the biggest risk to the success of Golem. In Edge Computing, those involved are trying to develop network technology to make the internet largely obsolete. It should be made clear, however, that Edge Computing doesn’t want to do away with the internet. However, the mass-adoption of the computing practice would equal more of a reliance on newly developed local area networks. Essentially, computers will have new and novel ways of reporting and analyzing data, offline. This will exist in opposition to existing practices such as sending large amounts of data to the cloud or specialized data centers of some sort, according to NetworkWorld.

Therefore, the spread of this new form of computing could stand in opposition to the Golem network for the simple reason that the Ethereum network and therefore, Golem’s entire way of doing business could become obsolete. As a proxy, AI could also face major problems due to its aforementioned reliance on supercomputers as well as its possible, future reliance on networks of computers like Golem. We’re then left with a very specific question. What are Golem and Ethereum doing to prepare for Edge and Quantum computing?

More importantly: what are AI teams doing to prepare for this? For the answer to this next question, stay tuned for our next pieces.

References

AI and Computing Power: https://www.quora.com/What-are-the-technological-limits-to-artificial-intelligence-and-computing-power

What is Ethereum?: https://www.coindesk.com/information/what-is-ethereum/

Golem Network main site: https://golem.network/

Speculation on Golem’s future: https://www.reddit.com/r/GolemProject/comments/7qtbdv/golem_vs_skynet/

Golem ICO: https://www.forbes.com/sites/rogeraitken/2016/11/12/fintech-golems-airbnb-for-computing-crowdsale-scores-8-6m-in-minutes/#790899183583

Golem’s Roadmap: https://blog.golemproject.net/golem-building-the-worlds-most-powerful-supercomputer-on-blockchain-4ccb44c328a

Golem Project from CryptoBasic Podcast: https://www.cryptobasicpodcast.com/episodes/

Possible future threat to Golem Project: https://www.networkworld.com/article/3243925/lan-wan/artificial-intelligence-may-not-need-networks-at-all.html

What is Edge Computing?: https://www.networkworld.com/article/3224893/internet-of-things/what-is-edge-computing-and-how-it-s-changing-the-network.html

About Ian LeViness 113 Articles
Professional Writer/Teacher, dedicated to making emergent industries acceptable to the general populace