maxaw

I’m not worried about a fully autonomous attack, I’m worried about an attack by a crazy human coached by a frontier llm with no guardrails on how to make a bioweapon

What a cool article. I have good memories of being 13 and my cousin telling me about limewire. Between random pornography titles there was an artist called burial, which I downloaded cause I thought it sounded edgy. How lucky was I!

Great article. This last year I've been travelling around and have met 2 people running very lucrative vibe coding agencies. They vibe code websites and apps on behalf of people for whom writing prompts is too much mental overhead.

Your middle ground idea is interesting. And maybe inevitable in some ways, if the llm becomes even more of a universal interface to the outside world.

Regarding llm writing, it is disappointing. Especially as people could use it in the “opposite” direction, using the tirelessness of the llm to experiment with the communication with the aim to make it as clear as possible for the reader…

to play devils advocate - i doubt the articles that end up on the front page are one shotted (theres probably a sequence of back and forward refinement). and in any case, i feel the avg reader would actually not prefer to read the prompt, which would be very information dense

having said that, i'd still much prefer a norm of including prompt history with the article, or the codebase for that matter, so people can choose for themselves :)

Looking to experience life outside of software, but I don't know exactly what. In short, current filters are:

high total customer face to face time// high face to face time per customer// probably not in sales

as these are too abstract to map cleanly to traditional job board filters I’m scraping indeed and using deepseek to classify jobs according to this criteria, with an aim to discover really good jobs and then put a lot of effort into each of those jobs, like reaching out to hiring teams directly etc. works alright but worried coverage is an issue.

ps- can any one recommend a service or product that does this already? i should be able to set a city and then write my own filters like "this job involves dressing up like a crocodile" or "this job requires ballet dancer experience" and have each job posted in my city get assessed. maybe i get an email each day of matched and not matched jobs. i have tried to search myself but given there is so so so much slop in this space i find it very hard going. and most products do this just very poorly...

Given the the labs are trying to create [super] human-like consciousness, partly through the guidance of huge system prompts, and many philosophers are experts in textual descriptions of consciousness it makes sense

Just went down a chatgpt rabbit hole because I was also curious - seems it’s literally physical repackaging, no modification of the drug itself- as another commenter pointed out it’s not quite the same molecule but they have the same effect? Someone correct me if I’m wrong

this happened to me too. few things about the process made me suspicious. i downloaded the repo and told claude to "find the malware". took about 15 seconds. remote code execution that would have run upon npm install, iirc. many layers of obfuscation. in implementation, a little different to the op's situation but there are similarities. it was a "crypto startup". maybe they think people in crypto world are more forgiving of idiosyncrasies in the recruiting process? i reported the recruiter's profile to linkedin, with extensive details. they said they wouldn't look into it unless i opened a ticket in some other part of their site, lol. however it seems they got onto it, or someone else complained, because i can't find the recruiter "alice kenny" anymore. but the "company" she was recruiting for is still live:

https://www.linkedin.com/company/blockchainaustraliasolution...

>why are they doing this?? [the mafia game show]

Instantly thought of the big short: “they’re not confessing. They’re bragging”

I don’t know why people have to pick one side or the other. AI speeds up development at the cost of oversight. Whether this tradeoff makes sense depends on the real world consequences of getting it wrong and the quality of the foregone oversight, which is very much case by case

Yes! Glad to see someone else having this thought. Will be interesting to see how it plays out

Good to know my thoughts are in good company! It seems obvious to me that on some level I believe these llms to be conscious otherwise I wouldn’t feel the urge to type all caps rage in the first place - rather than robotically reverting state to my last prompt. so I don’t want to get used to treating what my brain thinks are conscious beings with anything less than kindness and certainly not habitually calling them the worst words in the language ..

I’d rather lose 4% accuracy and practice kindness! I’ve been actively trying to avoid raging at the bot because I worry about this behaviour leaking into real world interactions

Thanks for commenting :) The whole point is tighter scopes. You control what parameters the agent is allowed to pass when calling endpoints. This allows things like revealing only a few folders in your repo to your agent

I didn’t say it doesn’t exist, rather that it’s already taken into account. I’m also not sure what you are proposing- if mediation is required, and someone has more power than someone else, why would they voluntarily engage with a mediator who will reduce that power? Or if they are forced to use this mediator (eg by the state) then this means they never had the power in the first place

In reality people never have equal power over anything (what would that look like, physically?) so something like nash bargaining is an attempt to get closer to a notion of fair given this inequality

This is so cool. Even small disputes like roommate arrangements can feel very emotionally impactful at the time and it would be wonderful to have a tool for these moments

This is lovely. Thank you for sharing this

If you cant find the people, may not be the right market for you :) ideally you are exposed to target market directly, and daily. Or you team up with a cofounder who is. Re: building in saturated market - “prove” is a strong word. It’s another signal. And you can judge for yourself how strong :) i dont think anything can replace direct customer feedback

Begging is hyperbole, its as you describe. Just looking for the strong reaction

It is “simple” Find out what people want Make it Sell it to them Unfortunately, engineer brain loves to skip step 1

Ive recently become friends with a younger person who makes a lot of money off vibe coded mini saas. He is fanatical about step 1. If he can’t find n people begging him to make it he will go validate the next idea. He’s ruthless with this aspect and will drop an idea instantly if people dont care. It really woke me up to the reality of it all. Made me realise how much i delude myself into making things people dont want because i enjoy the making process. I will at best half ass step 1 and the proceed to spend a few months hand crafting some software no one wants. Meanwhile he spends two months validating and one month vibe coding something that people would be embarrassed to post on HN and then sell 100usd/month subscriptions to it. Its crazy

Proxy over GitHub’s REST API for fine-grained repo access – e.g. file-level scopes. For unpredictable agents :)

The big difference is that in many cases the people who support this are the same ones that are addicted. You’re telling addicts to stop their moral panic over their own addiction

The difference is that the media is 30 seconds not 2 hours so the feedback loop is shorter and the content pool is far far far deeper because it is user submitted so the content recommendation algorithms become so effective , and the experience so compelling, that it becomes addictive. And as a wise man once said “a difference in scale is a difference in kind”

I watched my 78yo step mother become addicted to reels so older people are definitely not immune. But she was able to go cold turkey as she only communicated with her sister over instagram so it wasn’t a problem to just continue with WhatsApp. Young people real life networks are too enmeshed with instagram to have the same option.

Also, what you’re describing sounds like when you’ve haven’t spent enough time on the shorts for the content recommendation algorithm to learn your preferences. Which I agree, is unwatchable. I saw it recently when my friend put on YouTube shorts on a guest account (on an Airbnb smart tv). it was bad. But spend enough time and that will change. But best you don’t!

I didn’t mention time? From Cambridge dictionary: ‘addiction: an inability to stop doing or using something, especially something harmful.’ I am in support of regulating things which are harmful and which people have trouble not doing

I don’t know a single person who after exposure to short form video has not had to exert special effort to regulate their consumption.

Screens on their own aren’t “uniquely and magically addictive”, but infinitely scrollable short form video delivered through that screen is, because a few companies spent billions on the smartest minds in the world to make it so.

Fully agree with you