As Iran's President Masoud Pezeshkian walked alongside the coffin of Supreme Leader Ali Khamenei in Tehran last week, some of the black-clad mourners surrounding him chanted not in tribute to the late leader, but directly at him " "death to the compromiser." Not far from that site, Abbas Araghchi, Iran's top diplomat who negotiated a ceasefire with the Trump administration and lifted some sanctions on the Islamic Republic, was forced to flee the funeral after a mob pelted him with rocks amid death chants accusing him of being a "traitorous sellout." read more
The first defendants since daycare fraud became a highly talked-about issue in Minnesota have pleaded guilty. That includes the woman who received more taxpayer money than any other childcare provider in 2025. Fahima Mahamud was arrested in February after she closed down her daycare and, that same day, booked a flight to London. The 50-year-old ran Future Leaders Early Learning Center since 2015. That daycare received $3.7 million in childcare assistance program money last year, more than any other daycare in the state. read more
Maine Democrats have until July 27 to choose a replacement candidate to take on Sen. Susan Collins (R). read more
Bryan Johnson, the former tech CEO who's become famous for his biohacking pursuits, says he has autoimmune gastritis. That means, he says, his stomach is "eating itself." read more
New mortality data from the federal government suggests that life expectancy probably hit another record high in 2025, as death rates have continued to fall since the height of the Covid-19 pandemic. read more
The AI question has everything to do with the nature and possibilities of alien intelligence, which is science fiction.
#104 | Posted by Zed at 2026-07-23 10:02 AM | Reply | Flag:
So no more hard science fiction then, back to speculative.
Since you've been reading about AI, and all of it fiction, and you won't watch a simple statquest video, here's something to read about AI that's real: The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.
The foundational paper causing all your fretting.
Deluded to think any prof is reading what one of their 500 us history after 1877 students are writing. It's 2026, they have TA's, unless they're at a podunk school.