ChatGPT4 ... that was unexpected ...

[BudToaster]

shouldn't we work harder on our real intelligence before releasing an artificial variety? it might help us see the problems inherent in the artificial stuff. it's not like having a calculator in your pocket meant that you understood how mathematics work.

isn't that what homo sapiens has been doing for the last 3 million years? and see what it has brought us to ... maybe we need a mandate or three. we saw how that technique goes.

imho it is way too late to stop or slow it down - competitive pressure is driving it forward at a rate i just can't believe. and i am convinced this AI has been going for a decade already. tuning the LLM must have required that the bits that have been exposed since mid-march to start collecting data on how humans react to it.

i consider this time to be equivalent to the moon landing, the internet, and the birth of the microprocessor chip. amazing to be able to witness the birth of such incredible industries.

being a computer programmer (software engineer) for 57 years, i am all in. i was in the middle of installing AutoGPT (w/GPT4) when my spouse had her hip injury so that has been on hold for 3 days (nagging at me like a splinter in my mind - for 3 days). i have a very specific use case for sales and marketing. sure is a lot of hype. this will let me find out what is real.

copilot is supposed to work with ms office apps and dev tools - but i'm not on the short list - signed up and waiting for access. AutoGPT is accessible to everyone. Open AI fee is reasonable ($100/yr) for an individual. but i need to point it at my code base.

soon!

 

[armedoldhippy]

i consider this time to be equivalent to the moon landing, the internet, and the birth of the microprocessor chip.

it may turn out to fit right in with those examples. the more people that use it in various ways, the faster it will adapt and begin to see what we require from it. but will it be "controllable"...? they have already done studies on it, and apparently iit is already good at lying and misinformation...like we need more of that. it's going to be very convincing, and damned hard to spot as it gets "smarter"...

 

[BudToaster]

so i was thinking about not seeing or hearing anything about Mitch McConnell and it occurred to me that even if he were to appear on msm, there is no way to know i am not just watching a chatbot (if he starts speaking in Farsi, well, that's a tell). it will all look so legit, no one questions it.

but the larger point is in reference to all the blather about being uploaded into the cloud. if it looks like a hippie dickie, and it sounds like a hippie dickie, hey, that's good enough for most people. no questions asked.

me? i be on the outside trying to hack a virus into the skynet and bring it the fuck down. it seems to me that with a hive mind, just a thought-thread could be enough to fry the system - like in Snow Crash, the pattern on the tv screen was tuned to fry programmers' heads.

AND today, after a week of trauma and dealing with the medical system, i will finish my install of AutoGPT and try to determine what it can do for me and my projects. instead of a team of co-workers, i can just spawn agents. which work cheap. and they will all be mini-me.

i have great hopes. Hopium by the gallon.

 

[OpenPollenSeeds]

our children will use this for school
the world as we know it is gone

 

[Brother Nature]

Kids have been cheating at school since well before chatGPT.

 

[JKD]

A.I. Is Getting Better at Mind-Reading

In a recent experiment, researchers used large language models to translate brain activity into words.

www.nytimes.com