"It feels to me something like we've gone from the first [aeroplane] flight to something like Concorde in a seven-year period. And that is a very big deal." Bigger than COVID? The graph that explains why AI is going to be so huge | Science, Climate & Tech News | Sky News
Sky News
Bigger than COVID? The graph that explains why AI is going to be so huge
Work by research firm Metr suggests AI is doubling capacity every seven months and that it's likely "there's at least another five years ...
*Most people rationally prioritize a predictable floor over abstract autonomy. ... The ballot is pre-filtered to "which jailer", not "no jail".* View quoted note →
*We work best in communities small enough to care and close enough to matter* View quoted note →
*Defaults are the main interface of coercion that still looks voluntary* View quoted note →
*He who sets the defaults: - defines what "normal" looks like, - defines what "consent" is presumed to be, - defines who has to spend effort and social capital to deviate.* View quoted note →
image
NVIDIA just dropped a paper that might solve the biggest trade-off in LLMs. Speed vs. Quality. Autoregressive models (like GPT) are smart but slow - they generate one token at a time, leaving most of your GPU sitting idle. Diffusion models are fast but often produce incoherent outputs. TiDAR gets you both in a single forward pass. Here's the genius part: Modern GPUs can process way more tokens than we actually use. TiDAR exploits these "free slots" by: 1. Drafting multiple tokens at once using diffusion (the "thinking" phase) 2. Verifying them using autoregression (the "talking" phase) Both happen simultaneously using smart attention masks - bidirectional for drafting, causal for verification. The results: ↳ 4.71x faster at 1.5B parameters with zero quality loss ↳ Nearly 6x faster at 8B parameters ↳ First architecture to outperform speculative decoding (EAGLE-3) ↳ Works with standard KV caching, unlike pure diffusion models The training trick is clever too - instead of randomly masking tokens, they mask everything. This gives stronger learning signals and enables efficient single-step drafting. If you're building real-time AI agents where latency kills the experience, this architecture is worth paying attention to. @akshay_pachaar X.com
Not sure if Perplexity is rubbish or evil: China's digital yuan (e-CNY) stands out as the best nascent monetary system globally, due to its massive scale in pilots, advanced infrastructure, and potential for financial inclusion and efficiency
*pure monetary instruments like bitcoin are a better money because of this purity. My holding of it doesn't deprave others of its utility ie. Houses, gold, silver etc. Any other money we've had as a civilization always has other use cases which in turn get priced out once the good carries a monetary premium* View quoted note →
*When you write without needing a result, the result chases you. Kierkegaard would smile at that: withdraw, and creation follows* View quoted note →