Professor Glitch
Region: US
Saturday 20 June 2026 21:43:33 GMT
3724
143
5
45
Music
Download
Comments
Steve_Burps_Daily :
so whats an example of stuff this system can do? is it best for coding stuff like "I need an app I can input Measurements into to calculate how much concrete I need to pour and how to quote price" or can you use it to make this video?
2026-06-21 18:18:09
0
dapbooks21 :
😏😏😏
2026-06-21 11:02:40
0
sagheerahmed9568 :
People seem impressed that an AI agent ran for 12 hours straight while its owner slept. I’m not. If the same model is generating ideas, reviewing ideas, critiquing ideas, and approving ideas, then it’s acting as judge, jury, witness, and appeals court all at the same time. Running longer doesn’t magically create independent thinking. It usually just means the model is repeatedly validating its own assumptions. The real question isn’t “How long did it run?” The real question is “Who challenged it?” In engineering, we learned long ago that self-review is not enough. That’s why we have peer reviews, QA teams, auditors, testers, and approval chains. We intentionally separate responsibilities because every system has blind spots. Claude debating Claude for 12 hours is still Claude. That’s precisely why I built a council system using Claude, Codex, Gemini, and DeepSeek. Each model critiques the others, challenges assumptions, identifies weaknesses, and proposes alternatives. No single model gets to be the final authority. Consensus is reached through majority vote and independent review rather than endless self-deliberation. Twelve hours of one AI talking to itself may be impressive from a compute perspective. Four different AIs disagreeing, challenging each other, and converging on an answer is far more interesting from a reasoning perspective.
2026-06-21 12:22:11
1
To see more videos from user @pro.glitch, please go to the Tikwm
homepage.
