15:[[48609465,{"by":"bix6","descendants":20,"id":48609465,"kids":[48609585,48611668,48615743,48609999,48610987,48609908,48609962,48616585,48611441,48610297,48610065,48616454,48610644,48611875,48613107,48625059,48609768,48615281],"score":10,"text":"What do you care about these days? The world feels so misaligned directionally and work is all in service of some esoteric AI that just strips the beauty from various crafts. What are you finding joy and purpose in for your own life and wellbeing?","time":1781965545,"title":"Ask HN: What do you care about? What is your joy and purpose?","type":"story"}],[48588648,{"by":"uejfiweun","descendants":14,"id":48588648,"kids":[48593432,48629281,48597627,48588704,48593036,48609154,48608469,48594898,48588754,48601221,48597553,48591252,48589536,48618041,48595773,48604452,48599667,48592344,48596685],"score":14,"text":"I'm having trouble reaching that golden "builder" zone when I use things like Claude Code. It's cool to be able to conjure software from scratch using these tools, but the output... I don't know, it just doesn't really feel like something I myself did. Maybe it's the lack of effort required, or maybe it's the fact that I know in the back of my head that basically anyone could copy this with the right prompt, that software in general is no longer a special thing.

Has anyone been able to find that "zone of fulfillment" using these tools? Because I'd love to find it again.","time":1781803620,"title":"Ask HN: Do you find vibe coding / agentic engineering to be fulfilling?","type":"story"}],[48589935,{"by":"hnthrow10282910","descendants":9,"id":48589935,"kids":[48590184,48591164,48590680,48616259,48611765,48590186,48608128,48605341,48611438,48611305],"score":10,"text":"I’m starting to realise it’s hard to find people who say useful things at work or speak concisely. (I’m also at fault here). I partly blame layoffs as people feel the need to constantly look very productive.

What’s the best way to improve my speech in such a way where I’m concise and the words I say are intentional? I want to keep people engaged when I speak. Trying to be a man of few words, but also still seem engaged despite being quiet.","time":1781809413,"title":"Ask HN: How do you effectively communicate or present?","type":"story"}],[48598508,{"by":"silent_butagrim","descendants":5,"id":48598508,"kids":[48604110,48604131,48601052,48604182,48604609],"score":3,"text":"Do you consider that Monte Carlo Tree Search implementation in an opensource environment (like codex, claude code) for AI can improve the code generation accuracy?","time":1781876632,"title":"Ask HN: Open-Source Intelligence","type":"story"}],[48601097,{"by":"vantareed","descendants":9,"id":48601097,"kids":[48602993,48605332,48602653,48614066,48612903,48603269,48603893,48640350,48601129,48601138,48602877],"score":16,"text":"AI and agents are taking up most of the oxygen lately.

What recent non-AI technology progress are you excited about, and why?","time":1781891225,"title":"Ask HN: What is the coolest tech progress outside AI?","type":"story"}],[48538686,{"by":"Protostome","descendants":5,"id":48538686,"kids":[48555599,48548821,48547731,48541604],"score":17,"text":"I've found that doing basic biology research with Claude has become extremely difficult. I'm not talking about anything remotely related to developing a bioweapon, just straightforward immunology questions that almost always end up being flagged.

I completely understand the need for guardrails, but in practice they can significantly reduce the tool's usefulness for legitimate scientific research. It sometimes feels like the model is better suited for "vibe coding" than for serving as a serious research assistant.

What's more funny is that you can use claude code together with molecular design models like RFdiffusion and Alphafold to develop a real bioweapon, but since this is in the realm of software, the prompts do not get flagged at all.\nThose models can even run on consumer GPUs so the limit of access is almost not there.

Has anyone else had a similar experience?","time":1781515061,"title":"Tell HN: Claude is completely unusable for biology","type":"story"}],[48536531,{"by":"the-mitr","descendants":7,"id":48536531,"kids":[48539874,48537357,48540468,48537533,48539674,48537163,48548529],"score":6,"text":"What models and hardware you are using? For what purpose? What were the challenges? Any tricks that helped you in doing this?

This might help new users like me setup theirs.","time":1781497050,"title":"Ask HN: What does your local LLM setup looks like?","type":"story"}],[48488990,{"by":"fractalf","descendants":9,"id":48488990,"kids":[48506418,48494264,48536138,48490482,48539096,48535819,48506348],"score":6,"text":"I've tried many different models and without doubt the code coming out of them differs a lot when it comes to "quality". Some of that is subjective for sure, but there are objective sides to "good" code.

I wish this was a metric for the AI benchmarks so I could choose a model based on this, because honestly it's one of the things I care most about.

Problem: How can you measure such things, whats the metrcis?

...maybe there just isn't a way to do it, since that metric isn't in the charts..","time":1781177979,"title":"Ask HN: Is there a metric for AI code quality?","type":"story"}]]

user:	PJHkorea
created:	June 14, 2026
karma:	2
about:	English is not my native language, so please bear with me if my phrasing is a bit awkward.