I only follow accounts I've disagreed with. Building something new. Co-founded @ComplianceEase | Alum @Caltech @Wharton | Ex-Post-Facto. Sometimes jasonr.eth.
In general, people's predictions regarding technology would be true, but only with "all things being equal"
Technology changes variables in unexpected ways, so that all things are rarely equal
https://t.co/ELBbagAtWR
so far at least, i'm pretty sure AI has been net job-creating.
this was not what i expected--although i was much less pessimistic than others, i thought by this level of capability we'd have seen some impact.
it is possible this direction keeps going!
Every exam in college was take-home and open book. One time, they accidentally posed a problem with no known solution. Basically, if these students weren't sobbing by the end, I fault the test authors.
https://t.co/RoHmSTdG4X
A Brown professor gave his students a take-home midterm exam. After suspecting many cheated using AI, he made the final in-person. The orange dots are the midterm scores and the gray dots are the final scores. Looks like all but 3 cheated on the midterm.
It's astonishing how incapable Google's AI tools are at completing tasks
I told NotebookLM to create a transcript with timestamps from a YouTube video
It couldn't figure out how to do it and had no suggestions for how it could be done
At one point, it reasoned it was impossible to determine timestamps because the video is from June 2026, which is "in the future"
With other, larger sources of cash flow, LLM labs like Google, SpaceX(ai), NVIDIA, Alibaba, and Amazon could potentially decide to subsidize inference indefinitely
Imagine working on 3G networks in the 1980's
"Sending three floppy disks of data per minute without a wire? Why would anyone need to do that?"
https://t.co/GmTjUA9dio
Mind boggling to me that I can make a thing faster and there's always people that ask "but why?" What kind of mentality is that? The pursuit of excellence does not need justification. Also, I find in so many cases, we can't know the impact of an improvement until we do it.
For example, one I've talked about before: Ghostty's high IO throughput has enabled terminal program (emulator and TUI) fuzzing at a speed thats incomparably fast to prior solutions. This has resulted in upstream patches to resolve issues in popular projects like btop, tmux, and more.
Speed enabled that anecdotally example that lifted the tides of adjacent communities that don't rely on Ghostty technology at all. I didn't predict this.
Make things better because they can be better and let the results naturally play out.
The closer coding is to pure translation, the less attention you need to pay to it
I usually have a detailed scope, technical spec, prd with user stories and acceptance criteria, and testing plan before generating relevant code
Same as when I was the one writing the code
I don't understand how anyone ever sees these replies because even a very small/fast LLM could very easily identify them as spam (same profile pic as OP, similar name, garbage content). They should appear for a few ms before they're immediately detected.
https://t.co/BL8T2Sg4L1
We really need one-click reporting of impersonations like this... if you have over 50,000 followers this happens all day long.
The algo seems to be getting it, which is great, but there should be a public database of these (like the facebook ad library) where you can just one click report "yeah, their using my persona to do bad shit").
Smartphones used to have a light that flashed for missed notifications, so we'd know if there was a reason to pick up our phone
Those lights gradually went away, so now we need to pick up our phone to see if there is a reason to pick up our phone
From what I hear, AI is a problem for designers, but not because they're being replaced
Executives use Claude Design or whatever in their spare time to completely redo products and the new job for designers is dealing with "why aren't we doing it this way?" all day
It's the 58th anniversary of Pink Floyd releasing their second album, A Saucerful of Secrets. Work on the album began in August 1967, before David Gilmour joined, and therefore the album included tracks which featured David Gilmour as well as Syd Barrett. Do you know which they both play on though? The album sports a cover design by Hipgnosis - their first for the band - which uses collage techniques to include elements from an edition of the Strange Tales comic book (issue 158 to be exact!).
We used to avoid rolling our own libraries to reduce risk of introducing security vulnerabilities
Now we avoid using third-party libraries to reduce risk of introducing security vulnerabilities