alecco

5 ms·

31.
by alecco 3mo ago
https://archive.ph/AvNa3 > “We told a couple of the customers who were being very aggressive with pricing at that time that this is not constructive,” he said, without naming Apple, adding that low prices discouraged cap
32.
Micron blames Apple and customers for the lack of memory capacity(wsj.com)
6 points by alecco 3mo ago|3 comments
33.
by alecco 4mo ago
https://archive.ph/aiJiq EU allies join US pact to break reliance on Chinese AI supply chains Jacob Helberg, architect of Pax Silica, tells the FT the American-led group will boost innovation --- European AI people not too
34.
EU joins US pact to break reliance on Chinese AI supply chains (no sovereignty)(ft.com)
8 points by alecco 4mo ago|3 comments
35.
by alecco 4mo ago
https://archive.ph/btGBY > After Zhipu’s release, Elon Musk wrote on X that he expects China to match the abilities of the current frontier by early next year. It “won’t take that long”, Tang Jie, Zhipu’s cofounder, shot
36.
by alecco 4mo ago
Also our brains can't keep up with the Joneses at a worldwide scale.
37.
by alecco 4mo ago
AI detector 96%
38.
by alecco 4mo ago
Yeah, not a great apples-to-apples comparison. I think the point stands: MoE, a myriad of complex attention approaches, shared layers, you name it. And making it all work together well is a huge trial-and-error pain even for small models, n
39.
by alecco 4mo ago
Grammarly and GPTZero say 0% AI.
40.
Stop Saying Half of 2026 US Datacenter Capacity Is Canceled(newsletter.semianalysis.com)
2 points by alecco 4mo ago|0 comments
41.
Barret Zoph is out at OpenAI again after just five months(theverge.com)
3 points by alecco 4mo ago|2 comments
42.
by alecco 4mo ago
He was "VP of Research (Post-Training) at OpenAI".
43.
by alecco 4mo ago
https://x.com/jukan05/status/2067865844689056191 > "Some neo-clouds worry that they can’t stray from buying Nvidia’s full stack of hardware for fear of being put in “Jensen jail,” meaning they might lose t
44.
by alecco 4mo ago
Allegedly OpenAI is struggling to jump to bigger models and had serious issues in the past (4.5) and also allegedly Shazeer is just the right guy for that. OpenAI is having issues hiring talent as most SF-style people want to go to Anthropi
45.
by alecco 4mo ago
https://www.threads.com/@stockmktnewz/post/DZui-v1ipMV/the-e... The European Union downsized plans for data centers to support AI, outlining a call for tenders that includes smaller-scale facilities than ori
46.
Europe Scales Down AI Ambitions with Smaller Data Center Tender(bloomberg.com)
2 points by alecco 4mo ago|2 comments
47.
by alecco 4mo ago
As a story, yes. But Terminator failed on a basic premise: Skynet becoming self-aware. The future seems more like Blindsight [1]: hyper-intelligent, completely unconscious systems outperform, out-manipulate, and out-compete human beings p
48.
by alecco 4mo ago
But it's not like "shoving the logs into CNC machines". You have to understand what they are doing and point them into the right direction. LLMs very often lack common sense once you move out of easy things.
49.
by alecco 4mo ago
Why is it bittersweet? Carpenters probably didn't cry when their tools improved. It will be bittersweet when there's no human needed at the wheel but IMHO we are far, far from that. These models/agents are just mimicking huma
50.
by alecco 4mo ago
A perfect example of a coding agent guided by a human with domain knowledge.
51.
by alecco 4mo ago
Yet another Ars article adding nothing to Ed Zitron's post 2 days ago and the corresponding Financial Times article. "Exclusive: OpenAI Losses Increased Nearly 8X in 2025, With Spending Hitting $34 Billion" https:/&#x2F
52.
by alecco 4mo ago
"The Cost of Excess" The American Council of Trustees and Alumni (ACTA) (2021) https://www.goacta.org/wp-content/uploads/2021/08/The-Cost-o... "How Much Is Too Much? Controlling Administra
53.
by alecco 4mo ago
Would that align with their VC-backed incentives?
54.
by alecco 4mo ago
Consider using decrementing score order (best on top)
55.
GLM-5.2 is the leading open weights model on the Intelligence Index v4.1(xcancel.com)
1 points by alecco 4mo ago|0 comments
56.
by alecco 4mo ago
> 4× RTX PRO 6000 Blackwell Workstation (GB202, 96 GB GDDR7, 600 W) Those are SM120 so no tmem/tcgen05 and lack of support in main libraries (it's like everybody is focusing on B300/SM100). For that money I'd buy a si
57.
by alecco 4mo ago
Bellard seems to be at the extreme tail of the distribution of talent x grit/perseverance.
58.
by alecco 4mo ago
Could be that the generated regression tests create actionable exploit code.
59.
by alecco 4mo ago
They are hands down the best at serving their model via API. Currently due to caching and their special Engram technology.
60.
by alecco 4mo ago
Good summary: https://xcancel.com/jukan05/status/2066752564809605584 DeepSeek Completes First External Funding Round, Raising $7.4 Billion - DeepSeek has completed its first external fundraising round, raising mor

More ›