HackerTrans
TopNewTrendsCommentsPastAskShowJobs

muldvarp

757 karmajoined 7 anni fa

comments

muldvarp
·19 ore fa·discuss
Making things easier (or cheaper) is the entire point of AI. If AI is great because it makes writing software easier (or cheaper), then AI is also terrible when it makes publishing slop easier (or cheaper).
muldvarp
·8 giorni fa·discuss
What is this referencing?
muldvarp
·10 giorni fa·discuss
I'd love to work for some company distilling frontier models. Seems like interesting work and screwing with OpenAI, Anthropic and Google would feel fantastic.
muldvarp
·13 giorni fa·discuss
> Was it 2016 when Geoff hinton said that radiology was a dead career?

Funny how the jobs most at risk of automation now are tech jobs.
muldvarp
·17 giorni fa·discuss
Tech careers no longer exist. Tech jobs will still exist for a few years, but careers they will no longer be.
muldvarp
·mese scorso·discuss
I have yet to hear a single convincing argument by a person that works in software why they can't be replaced.
muldvarp
·mese scorso·discuss
I got a beautiful Latex template for my master thesis many years ago. I still remember fixing some things and it took me multiple days to figure out how this giant tangled web of a template worked. A year ago I tried to recreate the same look in a typst document. It took me less than a day to build it from scratch. Typst is genuinely one of the most awesome things I have seen in the last few years.
muldvarp
·mese scorso·discuss
How do you know? You can easily create AI generated text that is impossible to identify as such.
muldvarp
·2 mesi fa·discuss
I think be both agree that the future of software engineering is very uncertain right now (and likely will be for years). For me personally that alone is enough to recommend not investing money and time to get into software engineering anymore.
muldvarp
·2 mesi fa·discuss
But saying "LLMs are not good at architecture so software engineering has a bright future" is _also_ extrapolation.
muldvarp
·2 mesi fa·discuss
I think it would be even more clear if you just write what you mean.
muldvarp
·2 mesi fa·discuss
LLMs are trained to mimic human language production. If humans have heartstrings and the LLM does a good job at mimicking human language production, it will also mimic those heartstrings.
muldvarp
·2 mesi fa·discuss
I don't think it matters if the reasoning is philosophically "real" if it can solve real problems.
muldvarp
·2 mesi fa·discuss
The fact that AI currently requires some human supervision to produce valuable results is not a good predictor that it will stay this way sadly. LLMs were basically unable to reason two years ago. They are now better at many reasoning tasks than most people. If there is even a remote chance that LLMs will make your job obsolete I would pivot as fast as I could. This includes first and foremost software engineering.
muldvarp
·2 mesi fa·discuss
I don't get this argument because a paycheck isn't necessary for the system to "work" either. Serfdom and slavery "worked" for hundreds of years.
muldvarp
·2 mesi fa·discuss
I agree but AI is a) owned by rich people and b) (sadly) too useful for this to matter.
muldvarp
·2 mesi fa·discuss
There is literally nothing in it for me (all productivity gains will be pocketed by my employer) and there are many reasons to be afraid of it.
muldvarp
·2 mesi fa·discuss
Wasn't that obvious the second ChatGPT 3.5 released?
muldvarp
·2 mesi fa·discuss
I don't get it either. LLMs put the enshittification of software engineering into overdrive. The job is less fun (reviewing AI slop, sometimes even produced by entirely non-tech people like managers), the expectation of increased productivity, the expectation that we can now do the job of multiple people and salaries will decrease as well. I don't understand how so many software engineers I know cheer for this technology.

Do they not see that this will drastically change their lives for the worse? I'm in Europe, none of them has ever earned "fuck you" money.
muldvarp
·3 mesi fa·discuss
Manual verification that the "judge" judges correctly.

Also, how exactly do you programmatically validate CVEs?