02-Sep, 2024 - morning links, etc.

links, etc Harder Drive: Hard drives we didn’t want or need a delicious waste of network bandwidth and linux network drivers. quite amusing. Have we stopped to think about what LLMs actually model? • The Register no. tags: linux, ai, hacks

September 2, 2024 · steve ulrich

01-Sep, 2024 - afternoon links, etc.

links, etc Why A.I. Isn’t Going to Make Art i read pretty much anything that ted chiang decides he wants to publish to the world at large. he’s been particularly critical of AI over the past few years and i have a difficult time arguing with the ground that he’s staked out. this article is well worth the time to read. it’s full of solid, nuanced perspective on the topic of art and human engagement with art....

September 1, 2024 · steve ulrich

26-Aug, 2024 - afternoon links, etc.

links, etc Crafting AI-Complementary Skills and Bulletproof Assessments more in the theme of … you need to wrap your head around this and figure out what makes the most sense to use, discard and leverage. there are some great examples of questions/tests that dig a few layers deep, and can be effectively worked with AI as a complement. tags: #ai

August 26, 2024 · steve ulrich

19-Aug, 2024 - afternoon links, etc.

links, etc Northern Virginia is world’s hyperscale datacenter capital • The Register it’s not even close. tags: #2024, #links, #datacenters, #neteng

August 19, 2024 · steve ulrich

21-Jul, 2024 - afternoon links, etc.

links, etc Smaller, Cheaper, Faster, Sober | Drew Breunig i’ve been intrigued by the small models that run fast and are increasingly useful for mission specific tasks. this seems like the thing that will be durably useful in all of this AI flotsam. Enterprises are foregoing general models for open ones trained for single tasks. And it’s not just foundation training, but fine-tuning training. Teams are learning that small models trained for a single purpose can outperform the best general models out-of-the-box....

July 21, 2024 · steve ulrich