One of the features I am looking forward to is the LeoAI chatbot. We are told there will be a feature similar to what most of the LLMs offer. This means users (Premium) can prompt LeoAI.
This is being trained on Llama3.1 so it will be based upon the Meta model. On top of that, we will see LeoAI fed the data that is on the Hive blockchain (text), updating on the fly.
What we will be looking at is a powerful combination where users can do their traditional prompts while also getting Leo/Hive specific information.
Why is this a big deal?
The focus is activity. It end up translating into transactions. As that increases, we see a host of benefits arising.
There is a lot of activity going towards these models.
Each day, tens of millions of prompts are done (maybe more). This is a growing number as people get more comfortable with it.
For this reason, it is important that we incorporate it into LEO as soon as we can. It is a golden opportunity to bring value to the platform while also generating revenue.
When I was using Groq earlier, this is what I came across:
It is nearing 600 million prompts. This is a significant number. Of course, Groq is a fairly popular platform, having its own chip architecture while also incorporating some of the most well known models. For example, it has the 3 version of Llama.
The point here is we have a new pool to access. This is activity that can be pulled onto the platform. We can see how this can easily amount of thousands of prompts per day. While this is not at the level of Groq, it is a significant number that adds up over time.
It is also time spent on the site, a metric that is crucial when you get into advertising.
One of the most valuable aspects of a chatbot of this nature is that ability to generate a great deal of data. While this is synthetic, which is controversial with regards to the value of it, we can see how it could enhance the amount that is available on chain.
Here is a thread mentioning a feature that is going to be added:
@khaleelkazi/re-taskmaster4450le-2gfj2ukkg
What this means is the information provided by the chatbot can be posted to the blockchain. It will appear as a threadstorm, being immutable and permissionless.
Consider how much more data will be available if there are thousands of prompts per day.
There is another factor to consider.
One of the areas of debate is whether synthetic data degrades with each iteration, making it less valuable. While it is uncertain, there is a logic to this thinking. For that reason, it is wise to take this prospect into account.
However, with LeoAI, there is another angle, as shown in this thread.
@taskmaster4450le/re-leothreads-25a6erft9
The synthetic data can be compounded by human data, thus providing the knowledge graph with more powerful connections. This could help alleviate hallucinations while also providing context to the data. This is an important step in the overall path of these models.
There are many advantages to Leo adding its own chatbot to the platform. This is something that is best served by getting it out as soon as we can. Things are moving fast in this realm and those that are behind will have a tough time catching up.
A dedicated page can enhance advertising activity. We know this is a premium only feature that could drive more subscriptions. The amount of data that can be generated will help to drive value to the database, further enhancing the decentralized and democratized aspect to data.
All of this favors Leo in general.
