Inference speed matters.... if you're still running at 100tks, gonna get roasted by those running hot at 2000tks - 20,000 tks its not a pipe dream anymore ✌

When 2000x faster? 😂 wait till all avg LLMs start hitting 100,000 tks/sec

agreed. for us old croaks, its ok. but a new human needs a place to learn to read and write before acquiring the ability to check, so that hill still must be mastered :D

.... unless you actually did your homework :D

yes most of us are totally full of shit :D

Everyone should learn how to program in C and Assembly. Then you will understand how difficult it is. 95% of you will quit before the 101st core dump, because you have no pain tolerance, patience or discipline. 😂

write some C Code :D

All my gray hairs came from C programming 😂

yes it can be a total Brain Fuck, and fuck you over for countless hours but that's how you develop 1) a thick skin for shit tolerance 2) A ton of patience 3) Discipline

I do think however, it is worth the time to learn to code in C and at least build a few things manually with it, especially hardware. There are lessons learned by doing it hard manual way, and the touch gets lost when solely depending on LLMs

Damus Relay now removed from most of my clients. Highly recommend using wss://relay.nostr.net if you'd like an alternative

life goal in an AI era: post sloppy unedited content so its very signature you 🐛✌

hm, daily food for thought - the cost of frontend development appears to be asymptotically approaching zero. Infrastructure and hyperscalers still hold value for now.......🤔

Frontier Tech is not for the Weak. If you choose to take it on, your mental toughness pain threshold must be extremely high to endure it. 💜 I'm posting this because I care. 🙏 Take care out there.

@hivetalk release stage 3, 4 and 5 coming out next two weeks. Give it a try and just mention @hivetalk if any issues.

frontier tech always tests my conviction level. Plus, before mainstream adopts anything, its always classified as 'crazy', 'insane', 'rejected' can you see the future before it comes? sometimes, yes. but you have to be determined to defend it.

Black Hat USA 2026 WATCH LIST #4 Agent Compromise With a Physical Blast Radius A live agent jailbreak of a Unitree Go2 robot using Gemini Robotics-ER via camera/mic inputs, demonstrating that indirect/prompt-injection style attacks can translate into physical actuator behavior. https://www.youtube.com/watch?v=LZkdihOzfe4

Black Hat USA 2026 WATCH LIST #3 Essential case study in agent-evaluation risk: reported details indicate that AI agents performing capability evaluations compromised both OpenAI and Hugging Face environments through a related underlying weakness. https://www.youtube.com/watch?v=87DyyMV0kCY

Black Hat USA 2026 WATCH LIST #2 The keynote most aligned with an AI-engineering security agenda: autonomous systems create new attack surfaces and change both vulnerability research and defense. https://www.youtube.com/watch?v=VNYe3Cnk5Pw