👋🏻 Hi, I’m Nathan, a UK-based technical professional working in DevOps environments, with a focus on cloud-native systems, Linux, Docker, Kubernetes and AI.
@thsottiaux@MiaAI_lab But “influencers” don’t represent all of us. Randomise early access and get broader and more accurate feedback across many use cases. Follower count should not determine who’s feedback matters most. We are paying customers, our feedback matters too.
@REALFSDDU@thsottiaux 100% this. Some AI “influencers” are insufferable. @OpenAI@thsottiaux I am not an influencer. I’m a regular person trying to get shit done. I’ll give you real feedback.
I am no one and this will probably get nowhere, but OpenAI needs to stop giving early access to the clickbait whores.
They need to give it to real people with no following. Feedback would be legitimate, useful, and free of this hyperbolic nonsense.
Sol/Astra 6.2 are really good with a few rough edges. However the carry on from these clowns who had early access make you think its akin to GPT 2!
Be honest… does anyone else genuinely love spending an entire day at home, completely alone, with no plans, no visitors, and absolutely no desire to see another human being? Because that sounds like heaven to me.
Now seeing dots in the UK, at last! @ChatGPT@thsottiaux apparently it’s paused until I upgrade my plan, despite being on a pro plan which is what I thought was the minimum?
Someone wrote a bash script that emails Sam Altman every day their RAM sits idle on a Linux box.😄
It's a reply to what Altman said on John Collison's Cheeky Pint podcast on Oct 29:
"From a global allocation perspective, actually, it's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer, and being used maybe 5% of the time."
The project is called Ram-Altman. A cron job reads /proc/meminfo every five minutes and counts used memory as MemTotal minus MemAvailable. If more than 10% of RAM is free, Altman gets one email a day with the real figures.
@thsottiaux Yes, make it more accessible. Pro 50 would be a sweet spot for a lot of plus users. I’m on the $100 pro plan but I did have to think twice when I upgraded from plus. I don’t regret it because Codex is worth it for me, but I do understand why people are reluctant to upgrade.
How does AI actually work?
Let’s talk about what’s happening under the hood of agentic chatbot models like @claudeai. I’ve heard from my students that understanding the two components of these types of models has been the biggest unlock for them in using these AI models confidently.
In the prior video, we discussed how LLMs (large language models) work, and how they are “probability machines.” Today, we’ll discuss the second component: agent harnesses.
This was created in partnership with Anthropic for other platforms, but I'm posting this to X on my own.
At 40, you may be behind on energy. But you've got something they don't: 15+ years of experience, the discipline to grind through the boring parts, and the courage to start over in a new field. That's your superpower.