RE: More hints about the Tron Foundation plans for AI
then using the rewards to fuel its automated trading strategies.
I think that the idea of "automated trading strategies" may be drawing a distinction that doesn't necessarily make sense with a blockchain-trained-LLM agent. Theoretically an LLM can include financial transactions in its repertoire of ways to interact with the world the same way we humans don't just use words to communicate in face-to-face interactions but also incorporate physical gestures, facial expressions, etc. It seems like the idea would be that humans tend to learn about finance like learning a second language as an adult, but an LLM could be as fluent as a native- or bilingual-speaker.
In theory the Steem blockchain could be interesting in that respect, since there are a lot of transactions besides buys and sells that could help an LLM learn in interesting and nuanced ways. In practice I suspect a lot of the "data" in the chain wouldn't be useful from a practical perspective because the economy has always had a lot of distortions, weird power dynamics, etc.
I agree. That situation can always improve, though. It just depends on the goals of the dominant stakeholders.
This is an interesting point that I hadn't thought about. An LLM that's trained on Steem operations as a "native language" might even reveal interesting voting strategies, especially if coupled with reinforcement learning. I wasn't thinking of LLMs in those terms, but maybe JS is. If so, I'm sure his first training set would be the Tron transactions. It might be interesting if he included Steem operations too, though. Not sure if that would be much different than a pure ML approach, though... We've had ML voting bots here for years.
0.00 SBD,
0.03 STEEM,
0.03 SP