Hacker News Reader

Comment by dsjoerg | Hacker News Reader

I feel the opposite. Interacting with humans, I definitely pay a social tax - I have to negotiate the feelings of the people involved. With LLMs there is no social tax, I can be as blunt as I like. But there are other taxes to pay with LLMs; they don't learn, they BS relentlessly, there is less fun and camaraderie.

But "social tax"? No, there is not a social tax.

View on HN

Agreed, I haven't yet internalized them not remembering feedback from a couple days ago, which humans typically would. The process of remembering to update the context isn't entirely natural to me to this day.

I agree, but I get the impression talking to others and from articles like this one that everyone's experience is real, it's just that the diversity of experiences is so vast that there are some really good ones and some really bad ones and the best thing we can do is share more on how to get the better ones more often. I'm in a weird situation where my day job involves running a 400 person dev team and my evenings and weekends involves playing with my one person startup and my virtual team of AI agents, so I get a lot of heavy in person stuff and plenty of AI stuff too. The key for me in this last few months on the virtual side, which has been anything but fatiguing, is to use a framework a friend of mine made that treats agents not as order takers but senior collaborators who push back, have explicit autonomy levels for different tasks, and are by design proactive and helpful. It's made me able to confidently do things like product management and marketing that I lacked confidence in while leaving me to be as pedantic and particular as I like as the engineer on the team. This has been energising, not fatiguing. The framework is here: https://github.com/normannoble/agent-framework

The first-order social consequences aren't truly there but i think the format in which you interface with the tool can trick you into behaving as if they were, which is taxing it itself.

There are ways to put a superficial personality on the LLM which can break up some of the mannerisms that get old. Give it a new accent, allow it to keep a memory of things that nudge it into one personality or another, make it behave differently on weekdays vs weekends. Its still superficial, but if you work with it any reasonable amount of time, I think having the ability to change some of the characteristics can make it more..."fun" I guess.

One giant issue with say Claude Code is the speed, say you ask it to do something and you want to review the response, then I want to ask questions , I want to chat about the changes , but the slow speed of responses it just breaks the flow.

Unfortunetly now vibe coding is demanded to be used, to produce 10x more code or else some other developer will take your place.

I agree. I’m a social person but I also like my purely functional conversations with LLMs, and I have a lot of them going at once!

Doesn’t replace conversations with other people but at the same time I feel no tax talking to the LLMs.

> With LLMs there is no social tax, I can be as blunt as I like. But there are other taxes to pay with LLMs; they don't learn, they BS relentlessly, there is less fun and camaraderie.

My experience is the opposite. They bullshit a lot, derail the conversation/coding session, get defensive, and gaslight. It is not “social tax”, but they mirror sometimes the worst human behaviour.

They only respond to the inputs. Try adding some communication preferences to your system prompt.

There was no system prompt, I asked claude thru the CLI to implement something. It did not agree that I need that feature and offered something else. I started debating with it, why did I need it, and it started derailing the conversation, and explaining why I was wrong. And this is not the first and only case.

Reply on news.ycombinator.com