AI
I screenshotted this because it will seem so funny a couple years from now. Though to be fair it seems pretty funny already. https://t.co/qMwifbOLuz
ternary plot ftw
When is AI useful in the real world?
milan.cvitkovic.net
“AI is useful for a real-world task only if the cost for the AI to do the thing and for you to check its work is less than the existing solution.”
Excerpt: “Over the last 15 years, Excel has been unbundled into a host of other apps, such as Asana, Looker, and QuickBooks. But this unbundling was only possible once Excel became ubiquitous enough for users to know they wanted a purpose-made alternative. In order for that to happen, Excel needed widespread adoption with power users who began using it for niche workflows that it wasn’t originally designed to support.
…
Once those workflows were created, the power users realized that parts of their workflow were inefficient, or that features were missing for their use cases. They felt a need for purpose-built tools—and that became the opportunity for B2B SaaS to develop into a $327 billion market.”
AI is the new plastic
notion.so
There is beauty to naturally occurring material… and thought
The Death of the Junior Developer
sourcegraph.com
AI adoption and capabilities are distributed bimodally across industries.
Writing-heavy professions like writing, law, software are feeling it first. It doesn't feel that way because other industries are still untouched. That may continue for long or it might not.
It's always calmest before the storm
By default, anthropic hides these "inner thought" aka text chunks. This is the reason why sometimes you see the anthropic flower loading in the middle of a response. Claude is basically deliberating at arbitrary points of a response for a better output - imo one of the most underrated unhobblings.
LLMs are better with an internal monologue
After working with LLMs for 2 years, I’ve noticed that humans have very different short-term context windows too.
Humans with small-to-mid sized context windows develop deep “breakdown” strategies early in life to deal with virtually everything. Ways of breaking down complex problems into smaller steps that can be chained. Their brains contain an ... See more
Parallels between the how we think and how we get llms to mimic thinking
We have to take the LLMs to school.
When you open any textbook, you'll see three major types of information:
1. Background information / exposition. The meat of the textbook that explains concepts. As you attend over it, your brain is training on that data. This is equivalent to pretraining, where the model is reading the internet and accumulatin... See more