Wish I read Chinese honestly I wonder if they've got the same shit.
1 Like
I dont know what youre doing but that seems like a fake job to me
I probably have the most tangible job out of any of you
SBIR is basically crypto trading.
He’s looking at us military contracts that they offer online
Contracts maybe not the best word, they just offer out shit they want and you can bid on it to do it
nah these are grants, not contracts.
I’m still wondering if you can get those contracts to supply ammunition or guns, I read some of them and it seems they only want to give it to people who are manufacturing ammo/guns domestically probably to avoid brokers
I wonder how the ammo business is these days it seems like the price of bullets is always going up
nah they want to reindustrialize the base. every single gubment employee says "reindustrialize" every single day.
1 Like
there's massive amounts of money in "casts " and "moulds". can't even do that anymore.
Read it, I'm mildly into it
I generally think that people think too highly of LLM apps
For example they think they can do things without human in the loop/final stage human reviewer. Which is not correct, LLM itself is not actually "reasoning"
However if I look past the things I don't agree with I can see there is a project here, probably already planned and written by/for someone specific like you said
RAG my codebase and tell me what to fix is a real use case that basically everybody is working on. In this case (compliance) it has the issue I mentioned before (that your results can't be trusted without human review, therefore you have to do all the work you're already doing with humans if you need to certify a codebase has no vulnerabilities) and also because everyone's working on this you can probably wait 3-6 months and just use some startup's project to do this for less money/time spend (or 6 months to a year for it to become a managed AWS service, run it inside govcloud, whatever other nonsense regulatory shit you have to deal with when you're the federal govt)
In the end once this current phase of everyone rushing to build the obvious applications of LLM is over, this tool specifically (when it's built by some startup & available to the broad market) will just become another one of those compliance things that I ignore for as long as possible until somebody sends the "we have 16 vulnerabilities!!" email. Compliance and security are fake problem spaces because they're inhabited (within companies) by people who are worse at whatever task they're looking at than the actual people they're auditing, so their contributions are more noise than signal and this AI shit will only increase that problem
In fact I already got this email 6 months ago, 300 line excel sheet with "here are the list of vulnerabilities in a vendor app you own, can you tell me what we're doing about this" - open the spreadsheet and each line says "this description was generated with AI." Thanks for wasting my time. The answer to these emails is always - I won't be doing anything about this, fuck off and leave me alone - and you figure out how to say it politically in a way that gets heard and doesnt get you fired. Which I did
Point being compliance is a fake field full of incompetents and it is resolved more than politics than actual technical skill
All apps will continue to be vulnerable, with or without AI review, and that doesn't matter at all
People who work compliance and security should be ashamed of their contributions to society and immediately pivot and find something actually productive to do with their lives
My posts are only going to get longer from here on you guys better go to bed.
Too late to go to bed im staying awake
Might go get a bagel but i dont know if anything is open
Vyvanse?
Cease all drug use
1 Like
I think we have the best potential to upend bigasses life if we ping him with stupid shit an hour after he takes his adhd med and make him type 20 paragraph responses to everything
I will never stop taking zyn
I dont even care if it causes cancer