Roger Johansson
@RogerAlsing
Followers
6K
Following
21K
Media
2K
Statuses
21K
Creator of @AkkaDotNet and @ProtoActor. Building IoT and EV-Charging solutions.Interested in observability, distributed tracing, distributed systems
Stockholm / Sweden
Joined March 2010
What are your thoughts on ChatGPT "Codex"?. My first impression is that it´s fantastic. task based code maintenance. I´ve already merged some 10 AI PR´s, typo fixes, improved tests, code clean up etc.
0
0
0
RT @Uncommonsince76: Holy Insanity. In order to cut down on third world immigrants stabbing people with Machetes, Australia has set up “M….
0
8K
0
Anyone skilled with ChatGPT Sora?.How do I force animations to not rotate?. Tyring to describe a scene. no matter what I do, it rotates around the thing Im describing.
1
1
0
When dad gets access to AI video generation. My kids will have to endure so much nonsense from me. Sorry Alex :-)
1
0
1
instead of bloating the code by default, it could be shrinking the code by instead pushing for using standard libs. I mean, I´ve probably copy pasted various text/string functions a gazillion times. or various crypto functions. Because I´m too lazy to try to find a lib.
0
0
2
What if AI coding was different, what if instead of spewing out new random code. it was used to find patterns, . e.g "it looks like you are writing a A* path-finding algorithm, there is a proven implementation in this library/dependency, do you want to see how we can use that?".
2
0
8
Why can´t AI models do proper "in-paint" ? .Why do they need to recreate the entire image?. ChatGPT can´t do it. it does have in-paint, but it still alters the entire image. Sora can´t do it. same thing there.
0
0
0
0
0
0
when using @ChatGPTapp agentic mode, is there a risk of prompt-injection?. Say an agent scans Github repository issues. If such an issue were to include prompt information. Could that leak into the agents context and be executed?.I mean it will read the text and reason about it.
1
1
0
On the current Tea-Spill leaderboard. all the top 5 positions are inanimate objects. (it´s based on images from the Tea app leak)
1
0
3
More AI facial reconstructions - Homo Nadali. Yes, AI is biased towards modern human faces, and will strive to make it look like such. And the AI probably miss out on important facial muscles. But I still find it weird how many reconstructions of old human species looks so bad.
0
0
0
What in the fresh hell is this?.They glue meat scraps together in the US?.And sell without disclosure???.
@KeithM1963 @WallStreetApes Yes, transglutaminase ("meat glue") is legally used in US restaurants to bind meat scraps into uniform steaks like filet mignon, often without menu disclosure. Packaged versions must be labeled as "formed," but restaurant practices vary. It's safe if cooked properly, but can trap.
0
0
0
@ChatGPTapp does have "inpaint", but it still regenerates the entire thing. its bloody hopeless. "edit this specific area" okidoki. changes a lot of stuff elsewhere.
0
0
1
just allow it to mark a section of the image where the edit is supposed to go. don´t remake the entire thing.
1
0
1
I wonder how much compute power @ChatGPTapp just throws out the window for incorrectly generating images. e.g. "change only this part" -> "I rotated this face and added a clown nose while at it". "No. no rotation, no clown nose. just change that part". "Here is a top-hat".
1
0
1
Homo Erectus Dmanisi Skull 5. I´m trying out AI for reconstructing faces. This might be bad science, but most recreations of extinct human races are often complete BS. #homoerectus #anthropology #ai
0
0
1
I honestly believe this is true. I don´t really have a better suggestion, but most people that vote, have no idea what they are voting for. they might have a vague idea about lowered taxes. or more funding for healthcare. but that´s about it.
@triggerpod 'Democracy' is massively overrated and doesn't create better outcomes than many other systems. But it makes people feel like they're in control, even though they're not, so it's revered and has value to the population and those in power.
2
0
4
We will end up in some form of AI polluted world where we can no longer identify what is real and what is not. Which is the real one?
1
0
1