Does anyone else feel really uncomfortable when a coworker is using copilot/chatgpt in a meeting where he is sharing his screen and without any shame or reservations types out infantile questions and swallows the always very certain and almost always inaccurate answers as "the truth"?I just don't really know how to respond to it but I feel really uncomfortable, like I want to be somewhere else.Does anyone know this feeling?
if chatgpt isn't searching for web it will most likely output unverified midwit takes
>>108554680And it answers in this style "ah, this is a typical case of ....", reinforcing the retard using it that experts are overrated and you can just rely on copilot
>>108554665I've been using the free models available on websites. They are wrong about 50% of the time but it's still helpful and time saving.
>>108554700It's just googling with extra steps
How do you know it isn'tt the truth?Lets be very clear:1. What did your coworker claim.2. Why is it infantile.3. So he is being confidentally wrong. AI is good at being confidentally wrong. So are people. Who do you know who likes being wrong?
>>108554695and if you question why it spits out false information by default it will tell you it's optimized for helping, not evidence-based conversation
>>108554720Stupid questions
>>108554723And end with condescending questions
>>108554711To some extent you are right. If you look at the action log, an AI agent it will literally detail to you its logical reasoning / steps it did.. Basically>OK. I need to solve a problem.>I understand that I need to verify this information, let me find smoe sources>Let me search the web>Searches google.com or other sources, you can pick which one.>OK let me read the source to see if it supports <claim> or does not, and explain why.Congratulations, you should have learned this in 9th grade. At least I did.If you have something you want to communicate to someone, transparency is the best policy. You can literally just read these logs if it sconfusing as to how to do so something. But AI can do it faster than any person can, so it isn't a fault in a person to use the tool to quickly do something faster than any person ever could. IMPORTANT CAVEAT: You must always have a responsible person to proofread/test/check the work on some objective metric.For programming it can be "does it run?" or "does it pass lint?"For AI/Art, it's subjective (how to quantify if art is good: not possible, lol)For AI in customer service: It doesn't need to be very good, but be useful at best.People who aer luddites about it are worse, really people who are luddites abou tit are worse. Most of the luddites are people who support orange man, and yes, fuck you if you did. Period. Full stop
>>108554756I wish the internet went back to being exclusively high IQ liberals with a techno optimist mindset. Everyone without just one of these attributes is too fucking stupid to function in the modern world. If you have none then you’re a certified retard who probably idolizes the unabomber.
>>108554665Is this the nu-/g/ version of "getting uncomfortable when driving into the countryside"?What the fuck is wrong with you pussies?
>>108554665Liberals hallucinating.
>>108554665actually it's smart because work doesn't matter. from the perspective of an employee with an IQ over 80, it's made up talmudic nonsense. your coworker is treating it as such.
>>108554776man it really must fucking infuriate you motherfuckers that i just keep on posting, and i'm still not deadmore black and white thinking - that is your fallacy. i dont care for what reason you do it. stop
