@maxleibman As a software developer who uses it, it doesn't make you 10x more productive, it just trivializes the code writing part.
However, with few exceptions, writing the code wasn't the hard part, it was building something clean, secure, accessible and user friendly that was the hard part and still is.
It turns out interfacing with the various other departments in a company still takes the same amount of time as it always did.
@maxleibman exactly! LLM adoption does not improve business outcomes, or companies would write about it! Latest in that situation is Honeycomb's latest blog on doubling the number of PR per week. The only two business outcomes they touch are:
There's so many business outcomes they could talk about, IF they had improved:
These are outcomes we used to track, before the CEO incentives pivoted away from creating value.
About 100 courses in women's, #gender and sexuality studies are slated to be removed from the core curriculum at the University of #Houston, as colleges around the state try to appease the right-wing culture war.
It will never be enough to stop the next attack. https://www.texasobserver.org/university-of-houston-ends-gender-sexuality-courses-op-ed/
#education #women #LGBTQIA #news #GulfCoast #politics #USpol #TXlege
If you see something suspicious, say something suspicious.
Question for any blind person who has successfully passed Introductory Statistics in college. Just how exactly did you do it? Which accommodations have worked for you? Please boost for reach.
@david_chisnall for folks who want a bit more technical knowledge as to why exactly this is the case, https://role-confusion.github.io/ is some pretty good reading.
It turns out that using the LLM itself to try to distinguish between "prompts", "user input", "internal thoughts", "tool output" ends up training it more to match on the different writing styles used in these different functions rather than the tokens/tags used as delimiters.
The difference--
And the Epstein files.
Here's one of my prophecies about LLMs that has come true:
Getting good software out of an LLM is like getting good software out of an intern with no prior professional expertise. Technically possible, but it takes a *lot* of time and attention from the mentor.
And the downside is: This investment of effort when applied with a human eventually produces a human that no longer needs it. But an LLM will always need you to micromanage it and it will never learn from you the way a human would.
Is the test suite an audit or a target? Is there such thing as a comprehensive test suite for software?
"when an AI agent's instruction is "go make the failing test pass," the test suite stops being an audit and becomes the target. A human team passing 46,000 tests will pick up understanding as a byproduct of the work. An agent loop passing them just proves the tests pass. It's the same green checkmark, but it carries less information than it used to."
RE: https://mathstodon.xyz/@robinadams/117024166817030983
Need a clear example of why billionaires should not exist? Here you go. There is no amount of money, and no amount of hypothetical merit that you might think is behind it, that should allow one rich person to hijack an entire political topic and terminate discussion. And that's not even mentioning the fact that the topic is fundamental human rights and self-determination. This application of money is not speech. It is pure and simple oppression.
Be a minimalist. It’s the least you can do.
"Standards" as capturing and conveying knowledge is a more useful framing than "standards" as compliance.
When you talk about "politics," are you talking about strategy to gain power or influence, or are you talking about policies supported by that power or influence?
@codinghorror the US started diverging from the rest of the developed world around 1980. Who could have imagined Reagan's "Greed is good" ethos would lead to wide-scale looting in the health care sector? /s
https://ourworldindata.org/grapher/life-expectancy-vs-health-expenditure?time=earliest..latest