LLMs are hard to create a moat around
it's stateless compute that you can switch overnight when a better/cheaper option shows up
all the commotion you see is downstream of this fact as companies flail around trying to fight this
@lauriewired This won't change because of speaker tech. Bad acoustics dominate differences in speaker quality. Most living areas have dreadful acoustics and would require significant effort and cost to justify high-end speakers.
@AmandaAskell Most flight schools offer a ~$200 hour-long "intro flight" in which potential students (you) fly a Cessna 172 from the left seat, instructor in the right. Great way to determine if you'd want to start training beyond the sim (I was immediately smitten). Go for it!
@adam_kranz On the other hand, computation seems to be independent of interpretation: the result of a given computation might be interpreted as a truth value, or an unsigned integer, or an ASCII character, etc.
Computations transform state, interpreters associate meaning with those states.
@dearmadisonblue By 'properly conceived' I think @getjonwithit is pointing out that statements like "Simulated water is not wet" belong to a metalanguage outside the described system.
*Within* the sim, "Water is wet" can be a true statement.
Use digits for the ages of people, animals and objects.
For example: a 6-year-old girl; an 8-year-old law; the 7-year-old cat.
Use hyphens for ages expressed as adjectives before a noun or as substitutes for a noun. A 5-year-old boy, but the boy is 5 years old. The boy, 5, has a sister, 10. The race is for 3-year-olds. The girl’s 6th birthday. The woman is in her 30s. 30-something, but Thirty-something to start a sentence.
@47fucb4r8c69323 It's not that math is more "true" than a natural language description of a planet's orbit, rather it is more *general*. Math gives us a clear and consistent language to express only the details that matter.
@rivermillion@cam_d32@bzogrammer And words only have semantics with respect to a given dictionary or language. Yet we still find it useful to distinguish English words from arbitrary strings of letters, ie, no one would claim there are 26^5 distinct 5-letter words in English.
@rivermillion@cam_d32@bzogrammer No, I showed upper bounds for *programs* that are strictly less than for *enumerated bit strings*. We can debate how many 32-byte programs are trivial (most) vs not, but there can be no debate that the number of 32-byte programs is strictly less than 2^256.
@rivermillion@cam_d32@bzogrammer Again, you're describing how to enumerate bit strings with extra steps. There are indeed 2^256 distinct 32-byte bit strings. For any non-trivial ISA, there will be less than 2^256 valid 32-byte encodings. In general, there are even fewer encodings that generate distinct outputs.