There is a great deal of conversation surrounding AI and blockchain. We dive into it quite frequently on here also.
This is an important matter. It is possible we are looking at a technology that is more powerful than anything we ever created. What this means is trust is a crucial factor.
When we bring up this work, blockchain should come immediately to mind. How do you bring trust in a trustless environment?
Anyone who has been around cryptocurrency knows the answer to this.
For this reason, it is important to note how the future is going to be, at least to some degree, AI models on blockchain.
The Future: AI On Blockchain
This is not a novel concept. Some companies have been working on this for some time.
One of the major issues people are having is they do not trust the data these models are being trained on. In most instances, since we are dealing with proprietary systems, we have no idea what is taking place.
Of course, people like Sam Altman claim this is solved by trusting him. He seeks regulatory capture by having the government outlaw open models.
This is akin to trusting the banks. Anyone see what Wells Fargo has done over the years? Now correlate that to the likes of Meta.
Does anyone think these entities deserve our trust?
Here is where blockchain provides the solution.
Fortunately, there appears to be a breakthrough in this matter. AI on blockchain might be a reality already.
Fully On-chain AI Model
This is the claim be DecideAI.
Artificial intelligence firm DecideAI has claimed it has been able to completely place an older AI model onto the Internet computer (ICP) blockchain, marking what it says is the first time the model has been launched onchain.
The firm asserts it put GPT-2 on the blockchain. This is an older version, dating back to 2019.
It is, naturally, viewed as the first of what will be an evolution of AI models on blockchain. The benefits could aid a variety of industries such as education and healthcare.
Reverting back to the elephant in the room, surveys are that trust is a major factor regarding AI models. It starts with the data it is being trained on and then extends to the models themselves.
Here is where blockchain can assist. By going the decentralization route, we see the ability to offer the tools to people expand. At the same time, we are looking at data transparency in regards to the models, something that should aid in the trust level.
Using blockchain removes the need to trust an OpenAi, Anthropic, Google, or XAI. We often cite third party risk when discussing the financial arena. This is no different.
It is also something that is likely to receive blowback.
Regulatory Environment
The regulatory environment surrounding cryptocurrency is still uncertain. When we consider what the political class is trying to do with AI, there is a fight about to take place.
As mentioned, people like Altman want to be handed control through regulatory capture. A banning of open source models would hand the power to the few companies that were large enough to comply.
Is blockchain a counter to this?
Part of DecideAI's future plans are to extend this to Ethereum and Solana. This means we are looking at AI models, along with agents, being placed on decentralized networks.
How does the government counteract this?
We know the ability to shut down these networks is of a bygone era. Bitcoin has operated in the wild for more than 15 years, with nothing stopping it.
The idea of regulating blockchain networks is against the concept of decentralization. Once something is out there, it is there for all to use.
Of course, one of the major issues with AI models is the compute required. This is why the model used was an earlier version. The compute associated with GPT-2 is much less than the present one.
Hence, scaling is an issue. Of course, this is always the case with networks. Does that mean that blockchain models will always be lagging? Perhaps. It is likely in the near term as experimentation occurs.
Even if that is the outcome, as compute power increases with regards to price, more powerful models will be available over time. For example, perhaps GPT-3 could be deployed on this same network in 12-18 months.
Decentralized Internet
The concept of Web 3.0 is basically a decentralized Internet. We can see how these AI models are being incorporated into Web 2.0. Where we see a challenge is in what is being posted. How do we know if it is authentic or not?
We also see the benefit of blockchain’s transparent and tamper resistant data processing capabilities to enhance the user experience and reduce user friction.
Recommendations can come to users without them having to give over their privacy since the blockchain can be the source. At the same time, we can see the control of the algorithms and data spreading to more than just a few companies.
This is opening up completely new possibilities and could radically enhance the appeal of Web 3.0.
It is also the future.
Posted Using InLeo Alpha
