facu17y

6 ms·

1. 31.▲
   by facu17y 3y ago

   There is no AI risk. There is only the risk of bad or "unfavorable" actors using AI. AI in a killer drone unleashed on civilians? The bad actor is the one who deployed this weapon. AI given agency and goal maximization ending up g
2. 32.▲
   by facu17y 3y ago

   mvcc is already taken care of by fdb, no?
3. 33.▲
   by facu17y 3y ago

   you can do it for free with tesseract.js
4. 34.▲
   by facu17y 3y ago

   yes, seems to be a network issue, not a service issue
5. 35.▲
   by facu17y 3y ago

   seems to be a network issue, not a service issue
6. 36.▲
   by facu17y 3y ago

   TruTime is, in Google Spanner, using atomic clocks
7. 37.▲
   by facu17y 3y ago

   This project reminds me of this now-ancient blog post by @marcfawzi from the Evolving Trends blog (does not seem to be maintained): https://evolvingtrends.wordpress.com/2006/06/26/wikipedia-30... It took more
8. 38.▲
   by facu17y 3y ago

   "98% of the time, and the other 1% of the time" Is this a strange take on "off by 1" error?
9. 39.▲
   by facu17y 3y ago

   As long as the synthetic data is good, how can you tell the difference between it and human generated data? This paper has one huge hole in it: it assumes that content on the internet is not moderated and that the training dataset will neve