@3blue1brown I heard that for educational and documentary videos, AB testing from a different platform shows that people prefer the British accent to the American accent. I’m genuinely curious if you can get a 3b1b video to be dubbed by someone with British accent and see what happens
The U.S. flag packs 50 stars into a rectangle. I found a packing with stars 95% larger in diameter. It has central symmetry and is our best so far, though not proven optimal. I don't know if an optimal packing has central symmetry. AI-assisted search.
Last night I had a dream: I worked on a wireless communication problem with many antennas. The output at the i-th antenna was y_i=f_i(inputs). The problem is therefore named y_i,f_i, and pronounced WiFi.
@dntse If AI can solve some open network information theory problems, it may bring a renaissance of the field. It may only last a few months: by the release of the next AI model, the classical problems may be all gone.
@AlexGDimakis@DimitrisPapail There are tons of open problems in network information theory. I feel we have more open problems than solved problems (solved as knowing a single letter characterization). Will AI trump all of them? I haven’t seen signs yet.
@AlexGDimakis@DimitrisPapail I like the part “the field has largely moved on”. I’m one of those who moved on from information theory to industry. Back in my PhD I asked my advisor, what will happen to those open problems in information theory. He said, people would stop caring at some point.
@abakcus Twenty years of Google trends for “Roman numerals” in the United States. The peaks are super bowls. People want to know what number Super Bowl is it. The secondary peak in 2025 is to find out the meaning of Pope Leo XIV.
@__eknight__ I'm reading about this problem, and I think the stronger conjecture about orientable cycle double cover is aesthetically more appealing. There, each edge is covered by two oriented cycles in both direction. I wonder if one can ask GPT this stronger version as a follow up prompt