To be honest, I was less interested in the news about the burning of 22M $KGEN than in what stands behind and is being built on this decision.
In crypto, we have already seen a very large number of projects that simply conduct huge token burns. But usually this only creates a short-term hype, but after a while everything returns to its usual state, especially if new issues continue to enter the market.
In the case of @KGeN_IO , I think something else is important here. The project itself is trying to build a model where the reduction in supply is not related to a one-time event, but to the development of the business itself. If new AI contracts actually generate real income, and part of this mechanism is used to reduce the supply of the token, then there is actually a direct connection between the growth of the product and tokenomics.
Another interesting point is simply the emphasis on transparency. A lot of people talk about revenue-backed models, but few are willing to show confirmed figures. If auditing and on-chain publication of revenues work as described, it will give the community an easy opportunity to evaluate not promises, but real results.
That's why I like this, because it's experiments like this that show whether a token can be part of something big, part of the progress of a project, and not just a tool for speculation.
@KGeN_Community #KGeN @humynlabs
And just like that, only 24 hours remain until Season 1 goes live!
The Cellar doors open tomorrow. Over 7.7M $BCN in rewards, new cosmetics, and Kraken Klash awaits.
Bring your friends, they'll thank you later ๐
https://t.co/yVyZqVo0LX
S1 Details ๐
https://t.co/9I80TJmsb7
@TheSerafim_ It is very interesting to hear your opinion, and you are right, real conversations always look different from perfect test scenarios. That is why the training data should be as realistic as possible.
What struck me the most here is that a large number of people don't even think about how different the real world sounds. Many people don't speak the "perfect language" in real life - people are always mixing up words, switching between languages, having a lot of different accents, or simply interrupting each other during normal conversation.
And that's why one important metric like WER could never properly show the full picture. In real life, when communicating, two people can say the same thing in completely different ways, but to the system it can just look like a mistake, like an error, because it doesn't fully understand.
I really like that BRIDGE doesn't just compare regular models to each other, but tries to look much deeper: how AI behaves in different settings, where people communicate naturally in their own words, and not read pre-prepared text in a quiet room. All this looks much closer to how things actually work in everyday life.
It's purely my opinion that if AI is to be truly global, it must learn to listen to every word, every accent, not just the data that is "convenient" for it. Because nowadays the real world and its communication are very different from what is written in books - now it is not just one accent, not just one language, and there may be definitely not just one model of communication.
@humynlabs@KGeN_IO@KGeN_Community $KGEN
Season 1 of @The_Beacon_GG goes live May 25 alongside $BCN!
Use my code and get a 50 Gobloonz + 1 Umbra Chest head start on everyone else ๐๐ผ
https://t.co/GNd3WTQKhQ
@TheSerafim_ They told it very interestingly and extensively, I agree with your opinion. Real people and their real activity have always been important
@TheSerafim_ Well written, brother, you explained everything very clearly, and you are absolutely right: as soon as problems arise, most projects just run away. I have encountered this more than once.
@TheSerafim_ You are right and I agree with you, bro, the rapid growth of AI is very good, but few people are thinking about what the cost of mistakes might be in the future