Good morning cool cats! I've been remiss in sharing, but after bouncing from coast to coast to coast (back in CA again) I've found a jet lag moment to compile some of the reading I've found worth sharing.
Before I turn you loose on some fun coffee reading, I want to share an early idea that I'm working on. That in classic left brain human fashion, we're perhaps applying the wrong metric in evaluating Ai. An intelligence is just that, Intelligence is not a calculator. Repeatability and predictably and reliability are the hallmarks of machines, automation, six or seven sigma level error rates. Intelligence however, is prone to mistakes. How many of us can reproduce a sentence, verbatim, we just spoke? Will some part of it not be changed? The tone, the timing, an inflection, a breath? Even the best actor, having memorized and rehearsed their lines ad nauseum is challenged to replicate the dialog exactly night after night. Intelligence is messy, chaotic, exploratory.. hallucination is a hallmark of Intelligence; we do it every night and a dozen times a day. So, that's something I'm thinking about...
Ok. On to the fun stuff.
Don't be intimidated by the number of links. The longest is a 15 min read, most are a few minutes at most.
The subject of time has come up a lot recently, mostly for the wrong reasons in my opinion, but nevertheless it stands out. Any one of these articles separated by some distance of time would fail to make the impact of knowing they're largely concurrent; the pace of "progress".
And that's my segway into the first article. (Order: My Description or comments follow below the links)
- lesswrong.com
- In the weeds look at trajectory toward the super intelligence and the singularity, with a very self aware statement about predictions.
- spj.science.org/cbsystems.0336
- The study. (So cool! )
- techxplore.com
- The synopsis.
- (Synthetic neural thinking machines for Ai.)
- dailydoseofds.com/mlops-crash-course-part-4
- Also, in the weeds, but fascinating look into how we are refining and making machine learning more efficient.
- marktechpost.com
- Tower of Babel ... the global access to Ai takes a huge step, as does the connection of humans to humans in removing language barriers.
- venturebeat.com
- Self explanatory
- clickhouse.com/moosestack-does-olap-need-an-orm
- They have heavy bias, but that's OK. It almost reminds me of the Vertical Thinking movement of the late 80's.
- reuters.com
- It's not new news that Meta is playing something between Texas hold-em and roulette. Can you imagine putting your team through this successfully?
- marktechpost.com
- Prompts are dead. Long live the Prompts.
- arstechnica.com
- Local LLM. Pocket sized. Lightweight. Low power.... tik tik tik....
Share your thoughts on this decoupage!
- Have an awesome week!
- M