
Dev at @whogivesacraptp. Writing music (guitar/drum/samples) as Monopolyopolis. Watching tiny squares at https://xitter.gq/t.co/LQdpEmUBMX
Harrisonburg, VA
Joined December 2009
- Tweets1K
- Following456
- Followers149
- Likes724
Replying to @arithmoquine
Wow – it got within 15mi/25km of the actual location with nothing but this to go on, no signs, no horizon view, nothing.
Replying to @ElliotMurphy91
Can you click on “reasoned for a few seconds” and see what the thought process was? 3o-mini knows that it can go both ways, but goes with the more natural option. I bet if you prodded a little and had some dialog, you’d get an answer (although it’s being concise to a fault here)
Sounds like 4o can recognize when a word is made up (but pronounceable) and knows that those are often used for “exotic fauna” in sci-fi or fantasy, and then it takes cues from the supposed origin of the word to determine characteristics.
4o can jump straight from a generated mockup image into implementing the design in code. Made this with just these 2 prompts.
Result: 4o-glassmorphic-dashboard-de…
PS kinda funny that it decided to hardcode the time. Simple example here, so might not do well with more complex stuff
Replying to @strangestloop
First side project in a while that feels like it actually has some legs to it - dayprogress.com, little minute-by-minute progress bar of the day, makes it easier to see where you are in the day in relation to sunrise/sunset.
I tried this out. The hypotheses are all things we know cause other problems (genetic disorder, autoimmune disorder, viruses, etc) so you're right, it's nothing new, just remix of training text. But, germ theory existed in 1880 so I have to imagine it'd be in 1880 chatgpt's list.



















