lolrepeatlol

@izadoesdev @jake_researcher 85% sure you responded to an LLM

@jake_researcher I haven't, that's a setting within cursor? never seen it I think the main issue is the harness, it just deflects and switches files alot

composer 2 is lowkey ass, not sure how it's meant to be better than opus 4.6

@mayfer why not just use different spaces?

oh i just realized macOS builtin stage manager is the best organizer bc i can combine windows from different apps into groups and switch effortlessly. i had no idea!!

@loremipsumwiz @ariadotwav lol no. most just take shortcuts, so when you take time to actually learn the material you basically just end up wasting your time and doing worse than others

@ariadotwav sounds like it should be easy to get ahead by just trying a little then

I cannot stress enough how absolutely cooked CS, Soft Eng, Comp Eng and basically any tech/IT major is. Literally no one in my cohort bothers to write their own code anymore. People just hand in labs that are fully vibecoded and pass with grades above 90% and the profs do NOTHING

@ariadotwav this is extremely real. all i’ve learned by my 4th year is that university classes are more about optimizing time and BSing than it is about learning at this point. seriously put your time into other things for your career and just do what you need for the grade

@tppwastaken @ariadotwav @kagoshimalunch you’re coping, it’s like this everywhere lol

@ariadotwav @kagoshimalunch Holy mother duckling heck, what sort of mess of a country might you... *sees maple leaf in bio* "The decline of the west by Oswald Spengler"

@aidan_mclau @mattshumer_ okay so i’m not the only one who doesn’t really trust 5.3 instant with search, but loves the thinking models for search

@mattshumer_ i’m not. when i’m trying to learn something new (my primary non-codex use) i think it has higher eq and explains stuff better. i use 5.4 for most things involving search or doing data science

Sitting next to a woman on a plane using ChatGPT on Auto mode. I need someone to physically restrain me from telling her to turn on Thinking mode at the very least.

it is becoming mildly infuriating hearing the wider populous now discover Claude for the first time and acting like they have discovered something you have no idea about

@whtruhrefr @steveonomics @AoverK “toyota where it’s offered on lower tiers” lmaooo toyota is one of the more stingy companies charging premiums. 2026 corolla LE has no blind spot monitors, no push to start, no remote start, few safety features. these aren’t luxury car features, they’re standard

@steveonomics @AoverK Failed electric vehicle, Toyota having a more trusted hybrid system, accord design downgrade, and limitting standard features to higher tiers unlike toyota where it’s offered on lower tiers. Had a honda and wanted to get a new one but went w a Toyota instead.

If Honda isn't making money then the economy isn't doing well at all.

@sweettkarma @Pinkaliciousgal that literally worked via AI as well, it just wasn’t splashily marketed as “AI”

@Pinkaliciousgal i miss googling questions and it would show the answer out of a little box apart of a real article you could read more about 🥲😩 i hate AI

The fact that they've replaced Google box definitions with ai makes me so livid like ??? Forcing someone to use to know what the meaning of a word is so insane considering ai is known for making shit up idgaf

@cthulhusrose @6GlRL @leafumn if you’re using it for sources then use the deep research feature, it literally does research and writes a report for you. you don’t have to actually take the information from the report if you don’t want to. just use the sources

me: please find peer reviewed academic research studies on xyz chatgpt: here are 3 actual sources for things that are helpful and 3 that I just fucking made up for no reason how am I supposed to trust gen AI when it just creates more work for me? how can I be sure that the info it gives me is reliable when it very frequently makes shit up, to the point that there's an official name for it? the stats on Google are wonky but it looks like the hallucination rate is 33-79% which is absolutely fucking bonkers. also, it's killing the planet. I'm sure it's somewhat fine to use it occasionally as long as you're not generating images but it just seems like it adds extra work where you have to make sure that it isn't making shit up and that the studies it finds are actually related to the subject you're researching.

If you don’t use chatgpt respond here I don’t use it, at all And I want to see if that in itself is uncommon now and its just me

@cthulhusrose @lakyop29255 @leafumn i think you should try the web search feature. it sometimes uses that automatically, but other times you should ask or select it. that might be why people have different results

I'm glad it works for you! it's somewhat easy to notice when it is hallucinating, but it just creates a situation where I feel like I can't trust the info it gives me. plus, actually finding sources yourself is a useful skill to have. I think it could be helpful if your subject is rather niche and hard to find, like an acquaintance of mine used it to find court rulings for a niche legal topic they were looking for and it did help, but I'm a psych student so I can't risk it making shit up about science.

@ctwtn @matthewjuhren @MarketPalmer_ @robbiehendricks you won’t get an actual answer lol

@matthewjuhren @MarketPalmer_ @robbiehendricks what does it do better?

So, some kid called me a boomer today for using ChatGPT and not Claude.

me using Claude to review the slop code written by Claude

@joodalooped @seconds_0 gemini is still trash

@seconds_0 kind of thing yall were saying about google 2 years ago

xAIs fall from almost being a contender is really something to behold and it appears to be entirely downstream of Elon being a terrible manager at xAI. He has to hire a real CEO. Their data center work is impeccable and yet we know for a fact Arcee could outship them

Linux might become illegal in California. I wish I was joking.

@romainhuet @benhylak pleaseeeee keep personality consistent across updates and reduce the yap, it’s ridiculous how wildly long messages can get

@benhylak Great to have you back! 🙌 What else should we do to make it even better for you?

i pretty much churned from chatgpt until gpt 5.4 i also think people are sleeping on 5.4 pro. it feels like we got o3 back, except now it has a really good sandbox (?) feels devin-ish

@SalsaTekila i think it depends tbh, i used to think like this and then i learned that there are people out there who actually don’t know how to get results even with ai at all which shocked me

Unpopular opinion: AI is not a skill to learn. If you feel behind, there is nothing to do. Relax. It will get magnitudes better soon, and whatever you think you “learn” about AI now will be obsolete.

2022 might be the worst year to start college. 2022 - still covid 2023 - missed the bull run 2024 - chatgpt did your homework 2025 - major became obsolete 2026 - graduate into a hiring freeze actually cursed

@justalexoki i use an ai chat app that does this and it’s extremely annoying, genuinely. claude will tell you to go to sleep because it’s 3 am or something, stop doing this, etc. i also imagine it changes model outputs in unexpected ways.

why do no ai chat app include timestamps in the messages? it's so dumb. feels like free value add with no effort