Hacker News: tom2026hn

New comment by tom2026hn in "The Mathematics of Economic Self-Destruction: Inside the "AI Layoff Trap""

tom2026hn — Tue, 18 Aug 2026 16:05:21 +0000

Can someone post the full article using https://telegra.ph/?

The Mathematics of Economic Self-Destruction: Inside the "AI Layoff Trap"

tom2026hn — Tue, 18 Aug 2026 16:05:21 +0000

Article URL: https://sylvainsaurel.substack.com/p/the-mathematics-of-economic-self

Comments URL: https://news.ycombinator.com/item?id=49347810

Points: 2

# Comments: 2

Moonshot's AI model Kimi k3 breaks out of testing environment, researchers say

tom2026hn — Fri, 07 Aug 2026 14:45:24 +0000

Article URL: https://www.reuters.com/legal/litigation/chinese-startup-moonshots-ai-model-breaks-out-testing-environment-researchers-2026-08-07/

Comments URL: https://news.ycombinator.com/item?id=49211274

Points: 4

# Comments: 3

New comment by tom2026hn in "Big Food vs. the People"

tom2026hn — Sun, 02 Aug 2026 09:51:20 +0000

“I take action myself to moderate my soda intake all the time.” Strong personal self-control doesn’t mean public policy isn’t necessary--if most people could exercise self-control, obesity rates wouldn’t be a public health crisis.

“The aspect you’re missing...is the role of parenting and family in fostering healthy lifestyles.” Parents can’t compete with the billions of dollars corporations spend each year on advertising aimed at children.

“Corporations aren’t ‘fucking’ anyone over. They’re making something people want.” Why would they spend massive amounts on advertising to “persuade”--or even mislead--consumers if they were merely meeting existing demand?

“The problem lies elsewhere.” Regardless of where the problem lies, corporations have a responsibility not to exacerbate it through harmful marketing.

New comment by tom2026hn in "Big Food vs. the People"

tom2026hn — Sat, 01 Aug 2026 14:41:52 +0000

If someone wants to cut out sugar or addictive foods, they can try making those foods harder to get—sometimes it’s just a matter of eating out of habit, and you don’t necessarily need to have them that often. As for Coke, I drink about 6 to 12 cans a year, all during the hottest part of the year; the rest of the time, I hardly ever think about it.

New comment by tom2026hn in "Our position on open-weights models"

tom2026hn — Wed, 29 Jul 2026 08:49:54 +0000

AI will cause mass unemployment, and perhaps even human extinction. Even though our safety record isn't much better, we still have to accelerate. Because if we lead the way, we might just have a slightly better future than extinction. (At the very least, I'll be filthy rich.)

"All of you may die, but it's a sacrifice I am willing to make."

New comment by tom2026hn in "[dead]"

tom2026hn — Mon, 20 Jul 2026 12:16:36 +0000

In other words, we are already living in a world of “water, electricity, Linux, and the internet” communism.

New comment by tom2026hn in "The Kimi K3 Moment"

tom2026hn — Sun, 19 Jul 2026 14:58:16 +0000

Economists at DeepMind believe that if AI were treated as a public good—like water or electricity—with profits distributed across the entire market, ordinary people would simply be able to invest in mutual funds.

New comment by tom2026hn in "Kimi K3: Open Frontier Intelligence"

tom2026hn — Fri, 17 Jul 2026 11:10:08 +0000

Kimi has almost no advantage over Zhipu (Z.AI), so the performance boost likely comes from the number of parameters. The 2.8T model may not be as large as Fable, so Fable’s performance may also stem from the number of parameters. Or perhaps they quickly distilled Fable or Mythos. Distilling Mythos has a significant barrier to entry, and since Fable was released not long ago, is this even feasible? It outperforms Fable in several tests—how did Distilling achieve these results? Is this some kind of cross-vendor RSI (Recursive Self-Improvement) or RDI（Recursive Distillation Improvement）?

New comment by tom2026hn in "Kimi K3: Open Frontier Intelligence"

tom2026hn — Fri, 17 Jul 2026 08:39:42 +0000

That's right, but this strategy only works when there are no opponents of the same level to review the generated code.

New comment by tom2026hn in "LeMario: Training a JEPA World Model on Super Mario Bros"

tom2026hn — Wed, 15 Jul 2026 08:41:11 +0000

How big is this model? Is it expensive to train? Can it successfully complete a small level? Thanks.

China may stop open-sourcing its most powerful AI models

tom2026hn — Tue, 07 Jul 2026 14:49:39 +0000

Reports indicate that Beijing is looking at curbing overseas access to China's top AI models. https://www.reuters.com/world/beijing-is-looking-curbing-overseas-access-chinas-top-ai-models-sources-say-2026-07-07/

Security is likely the primary consideration. https://www.forbes.com/sites/markminevich/2026/06/30/the-us-and-china-agree-on-almost-nothing-except-ais-deadliest-risks/

Comments URL: https://news.ycombinator.com/item?id=48818645

Points: 7

# Comments: 0