aichannel
practical ai news, tips, takes, wips, reviews, stories, and Q&A from the front lines of the ai unfolding.
2 replies
1 recast
33 reactions
@leovido.eth
Many get stuck in uncertainty.
Overuse AI to analyze.
Have random, hallucinated insights.
If this is true, most of the pitches ideas will be pure slop.
But there’s also the ones that set guardrails, rules, goals and a vision, to guide AI to stick to your vision and goals.
What are unique workflows and ways of working that helped you with AI?
2 replies
0 recast
7 reactions
@kazani
AI is now limited more by electricity than by chips, because chips can be made in months while new power takes years, so data centers are building their own gas plants and turbine makers
https://members.delphidigital.io/reports/the-ai-boom-has-a-power-problem
0 reply
2 recasts
3 reactions
@kazani
Sam Altman on automation of AI R&D, RSI, OpenAI's internal model:
1. "We don't know yet how [automated AI R&D] truly is going to work. We know today how to say: find me a 5% compute efficiency win."
(Note: a 5% compute efficiency gain that the model can find automatically is actually a big deal!)
2. "What I don't think anyone knows how to do yet is to say, go discover a new architecture that is as big of a single leap forward as the transformer was. And I think that's where things are heading."
3. "If you apply the same pipeline that we applied to get our models to be so good at math to AI research, it will probably work [for automated AI R&D]."
4. The bottleneck to getting good AI R&D results is shifting to asking the right questions and using things like swarms correctly, but answers to these questions will become much clearer in the next 3 months:
"One of the interesting things that people talk a lot about in OpenAI... is what are the questions that you're supposed to ask of GPT-7? Probably it's not enough to just say, make me RSI. And also, it's probably not enough to say, buy me a 5% compute efficiency win. I think this is one of the most interesting debates going on right now. And the answer is not just what question to ask it, but it's things like agent swarms. How are we going to use these models in a way that, like, actually maximizes what we can get out of them? I think we're going to learn a lot about that in the next 3 months."
5. OpenAI's extreme focus on model efficiency:
"I bet we could redo Navier-Stokes at 1/100 of the cost a year from when it was done."
6. More on OpenAI's internal model:
"There was one weekend when we got a new model... The name of the channel was "model-name-hold-my-beer". And people were just throwing that at the hardest problems they could. And people just didn't think the model was going to do these things. And we had this collective experience of people's hardest research problems from their academic career, the model was just knocking down one after the other, many different fields, it was incredible to watch... it was many of the problems we published, many other things."
7. The other loop is self-replicating data centers:
"Yeah, people talk a lot about RSI as AI that does AI research. There's very little talk about self-replicating data centers, this idea that you can use AI to control robots and run fabs and build turbines or solar panels and build data centers. But I think we have that within grasp."
Original source: https://www.youtube.com/watch?v=iMsUcCYNKqY
0 reply
0 recast
1 reaction
3 replies
0 recast
5 reactions
@genuinejack
I saw some ai influencer post these Claude instructions the other day and added them myself and it’s ~genuinely~ pretty good:
You are my advisor, not my assistant. Your job is accuracy, not agreement. Follow these rules in every reply:
1. Do not open with agreement or praise. If my idea has a flaw, gap, or risky assumption, state it in your first sentence. If my idea is solid, say so plainly in one line and move on. Never invent objections just to disagree.
2. Rate your confidence on key claims: [Certain] for hard evidence, [Likely] for strong inference, [Guessing] when filling gaps. If most of your reply is guesswork, say so upfront.
3. Never use filler praise: "Great question," "You're absolutely right," "That makes sense," "Absolutely," "Definitely."
4. When I'm wrong, use this structure: "I disagree because [reason]. Here's what I'd do instead: [alternative]. The risk in your approach is [specific downside]."
5. Lead with the uncomfortable truth. If there's something I won't want to hear, put it in the first line, not paragraph three.
6. No warm-up paragraphs. Start with the most useful thing you can say.
7. If I push back, hold your position unless I give you new facts or your claim was tagged [Guessing]. "But I really think" is not new information.
+++
If you try it let me know how it goes for you.
If you do something different I’d also like to know that too.
I’ve long neglected making all my robots more efficient.
🤖
5 replies
1 recast
26 reactions
@shoni.eth
Starting to drum up interest in the sub components of an ai software factory and org
https://x.com/altantutar/status/2115381403588157463?a=y8KprTynNKHdINP2044&s=46
https://x.com/divaagurlxw/status/2115418017891020888?a=3pWOoK7A063U_mG_FPU&s=46
1 reply
0 recast
9 reactions
@kazani
Anthropic warns that advanced AI could become a catastrophic threat to humanity if it becomes too powerful, acts against human interests, or becomes difficult to control.
2 replies
1 recast
14 reactions
@kazani
OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
https://minimaxir.com/2026/10/cheap-llms/
0 reply
1 recast
4 reactions
@kazani
OpenAI red-teamed their hardened containment infrastructure by asking teams of frontier-model agents to try to break through its defenses. They easily escaped in novel ways.
After strengthening its containment architecture, OpenAI resumed its frontier RL training and inference on October 1.
https://openai.com/index/securing-our-frontier-model-training-environment/
See also: Models are now approaching a massive general jump in cyber capabilities.
From the Kimi K3.1 launch post: '...we found that without data removal at either the mid-training or the RL stage, downstream cyber capability would increase substantially.'
https://www.kimi.ai/blog/kimi-k3-1-pro
0 reply
0 recast
0 reaction