Grok Bot and Astra 6 push autonomous recursive research forward, author says
The author says autonomous recursive research improved sharply over the past month, with Grok Bot and Astra 6 running research, sending alerts and managing accounts like full research teams.
Im not sure people realize how much better autonomous and live recursive research projects have become over this past month. Grok bot and Astra 6 are truly showing the path forward. The equivalent of full teams of researchers presenting and proactively alerting me to alpha, running fully automated books and alike. I know some of the bigger players have built this from the ground up in the past, including myself which involved 7+ providers, up to tens of thousands of dollars per month in infrastructure… This past week got rid of most of the issues remaining for me. Anthropic is moving to the grok bot model and OpenAI will to soon. The next stage would be recursive learning from automations to create proactive new ones and talking to the CEO from there, where AI becomes middle to top management and reporting to you as the shareholder and CEO of the firm (your life or project). The final stage would be a device that combines those evolving and intertwined processes with what you see and experience in real life, bridging the physical communication currently done via computers and typing. This would move the definition of backend one step further to the back where the internet, databanks and all other providers are used in the background. I do not think the future involves many backend steps, AI will handle it. Own browsing will be about dopamine impulses and fun discovery like shopping new clothes or similar. Exciting but scary times… idk how some people will compete in the future. Alpha will probably eventually shift into a deeper layer of game theory. What do all AIs believe, where do they guide the masses. You’d almost want to have a contrarian, AI on the scale that would monitor the base line. Interestingly you’d probably want to monitor expectancy as presented with AI and realized path in regards to trading… trying to model the crowdedness of AI into an idea. I’ve said it before but this year the AI impact has been felt and shows up on my data. I see more divergence between avg/med and extremes. Either the balance is found near equilibrium or it create mega outliers when these fail. I do not think that’s random, I think AI is helping with alpha but through it is crowding ideas and concept that blow up certain trades that have no historical precedent or press agent historical behavior. See $CAR, $SLV, small caps and many others. More data or better processes only works when you have better ones than 95% or people.