On the last couple AMAs, we heard about the translation tools that Leo is building. This is due to the implementation of AI to the platform, something that is going to expand.
The first feature was changing the Leo template to whatever language one desires. There are a number that can be chosen in the settings tab.
What this does is change the basic page layout to whatever language is selected. The threads and posts are still in English (or whatever language they are posted).
This is going to change with LeoTranslate 2.0. It will not only bring utility to the site, it will add more data while helping with bias.
So what is LeoTranslate 2.0?
This will be the ability to utilize AI to translate threads, posts, or comments into a different language. It means that something that is posted in English can be translated into Spanish.
One of the keys to this is once something is translated in that language, it is there. This is important because then others can get the content in the alternate language without having to translate themselves.
How will this work?
There appears to be a split system being built. One is for general users with another for premium.
Premium users will have an unlimited number of translations helping to build some value there. Non-premium users will have a limit, perhaps 10 or 20 that can be done. The difference is due to inference costs.
As we can see, having more translations done by other users could enhance the experience of non-premium members. This means content that appears on For You or Trending could already be translated.
We spoke a great deal about data. It is crucial that we keep adding data to LeoAi.
LeoTranslate can help this in a couple ways. The translations are going to be posted on-chain, providing more data. This is the first level of impact.
Another is the conversations that take place.
If someone posts a thread in English, then someone else translates it into Spanish, the reply is likely to be made in Spanish. This is crucial, with more of it on a bit.
As the conversation proceeds, people using two different languages can interact. The replies will be translated into the languages for each.
A big factor here is to get more content outside of English. The translation will help although that is synthetic data. That is one phase. The next step, the interaction with replies being made in that language is crucial. It is here were the synthetic data is enhanced by human interaction.
One final fact that can not be overlooked is the fact this helps to address one of the inherent biases in AI models. Since internet data is used to train these models, the bias is towards English since that is what most of the content is. Having people engaging in their native languages will help a great deal.
This is going to be one of those features that adds to the utility of the platform while also increasing the amount of data.
