4,357 karma · joined September 22, 2013
which can now solve Advent of Code challenges with minimal input
Citation? The best writeup I found of people trying this for AoC 2023 had a mediocre success rate on the first 12 problems:https://www.postgresql.org/docs/current/ddl-schemas.html
I assume the article's solution involves some different set of tradeoffs versus native schemas, but that would have been good to include in the writeup...
Maybe he needs more "ruthless self-reflection"
This interfaces with the CPU's bus and produces very low-level clock & data signals to the LCD panel (CL1/CL2/FLM/MB/D1) - basically a stream of pixels along with synchronization pulses.
Nowadays most "graphic LCD" modules will incorporate a similar chip on-board, so that they can present a simple 8-bit parallel interface.
Capturing + translating the signal would be tricky - it's not _terribly_ fast (240x64x60Hz) but there's no flow-control or error-correction, basically low-resolution video capture
What groundbreaking logic it had implemented?
We consider a lot of what it introduced as "just videogames" nowadays, so it can be hard to spot. Comparing can help - consider Doom (1993) versus Wolfenstein 3D from the previous year:* walls at angles besides 90 degrees
* floor and ceiling textures
* multiple kinds of ammunition
* lighting
Why it is ported to various platforms
It was open-sourced in 1999, so it's partly about availability. I suspect it's also because it's _possible_ to port to lots of things; it benefits from specialized hardware for graphics acceleration but doesn't require it.https://www.nytimes.com/2005/09/12/technology/oracles-chief-...
Or is he only worried about the behavior of people who aren't billionaires?
Your health insurer, on the other hand, already HAS your money. Their goal is to do exactly as much customer service as will keep you (or more likely, your employer) from switching companies. The more efficiently they let you make claims, the more you cost them.
What these results tell me is that it is probably possible
for users to modify their slicer settings to get similar performance.
It took an LLM to notice that 170k lines of G-code was probably not typed in by hand? Despite numerous tries, I couldn't get an answer about what slicer
settings to change to get the G-code I produced using their slicer
to perform as well as the G-code generated using their slicer.
Did the paragraph where the LLM identified those settings get lost on the editing-room floor? It’s a clever, technically sophisticated solution to a hard
problem, and OpenAI has had a working prototype for two years.
Sure, and I've got an infinite free-energy generator in my garage. No, you can't see it or any performance results. But it definitely exists and works, TRUST ME /sSimpler explanation: the prototype doesn't work.
For instance, check out the demo on slide 10: now you can have janky sliders that don't match the platform's version on every platform!
“.NET developer”? Meaningless. Copilot, Cursor, etc can
get anyone conversant enough with .NET to be productive
in an afternoon
Until something breaks. Or needs an API after the LLM's knowledge cutoff. Or has a bug in the business logic that requires domain understanding to even see.My prediction: garbage handwaving like the above will be used by corporate bigwigs to defend paying less and less for experience, because "it's just typing in code".
sell that data to dental insurance companies
I know of at least one startup that tried this the other way - started with a networked toothbrush and created a dental insurer that tracked how much you used it.They've pivoted away from that to general benefits management (basically every flavor of job-connected insurance that isn't major medical), FWIW...
https://top10proactive.owasp.org/v3/en/c10-errors-exceptions
Well-run sites will have some kind of logging / exception-tracking behind the scenes that captures all the details about the error (stacktrace, request details, etc) for later review.
You could manually read and review all the code the LLM writes
If you're committing code you haven't even READ, maybe you should be replaced with a small shell script...* figure out what you'll use to check to see if the LLM is correct
* go study THAT instead
`cond` still returns the same output even though its input args
have changed. So `cond` is not a function anymore!
It sounds like you're holding the definition of "pure function" backwards: a pure function can't give a different OUTPUT for the same INPUT.A feature like "cond" is usually implemented as a macro (since the arguments are only evaluated sometimes) but that's ultimately just a shorthand for a bunch of extra "quote" wrapping.
This arguably should make Meta subject to the requirements
of the Online News Act.
LOL gotta get rid of that last 15% of engagement, I guess. Maybe ban even mentioning the NAMES of news outlets just to be sure /sA breakout board like this one will likely be helpful; it includes the needed connector, some power-supply bits, and an SRAM (since a 300x400 bitmap would eat a substantial fraction of the RAM on a small microcontroller):
AI systems are designed to execute these functions with high accuracy
LMAO AI’s Deterministic Approach
ROFLCOPTER For example, when AI is used to automate security protocols, the
exact parameters of those protocols need to be meticulously defined
so that the AI can implement them without errors.
Either this is the most out-of-touch startup ever, or some sort of advanced satire.