This is a very key point. Perhaps that discrepancy can be leveraged in image generation to save time.

It already is, which is why rads input window/level settings.

Chess reminds me more of programming given the set of defined rules in each. However, I'm biased as I work in radiology and program more as a hobby. So far I've seen way more tools to help me code than to accurately detect radiologic findings.

So far. Computer vision is currently lagging NLP, but I wouldn't expect that to last.

> but I wouldn't expect that to last

Do you have any links to research or work being done on computer vision that leads you to this conclusion? Would love to check it out!

You can compare best image synthesis and image understanding from two decades ago (SIFT / HOG), from a decade ago (CNN, SdA) and now (Transformer). Very rapid progress that went from being able to unreliability recognize a face to getting to outperforming human professionals (see MMMU) is quite remarkable.

AIUI, and I may be wrong, but each of the mentioned technologies was a "breakthrough" technology - not iterative improvement. Along this vein, I was wondering if there was some promising, novel research OP was aware of for image understanding.

The most recent of which you mentioned, Transformers, is used by both LLMs and image synthesis/understanding. The parent posits that while computer vision lags behind NLP, this may not continue. While your comment points out that image synthesis and understanding has improved over time, I'm not sure I follow the argument that it may soon leapfrog or even catch up with LLMs (i.e. text understanding and synthesis.)

I agree with everything you highlighted!

I'll add that my career isn't programming. However, Svelte let's me come back to my code base and understand everything, even after months. There is minimal framework related syntax to retain in my memory. Just have to know typescript well and that's it. Additionally, I don't need to retain all the rendering gotchas that other frameworks (eg, react) have. This has changed slightly with runes but I find them pretty easy to refresh on if I find im confused.

For me, this trade-off is well worth the limited availability of third party libraries. I wish someone would create a Laravel-like extension for Sveltekit. If I could just pull in user login flows, hashing, mail, etc all from organized Laravel-like docs, Sveltekit would be perfect for full-stack.

Awesome game! Simple feedback, your button icons are off center. At least for Firefox mobile.

Yes I use all default fonts and emoji's. For the font symbols as button symbols I think I need to use something like font-awesome, I'll add it to the todo list!

Bigger ecosystem for react. Svelte is easier to develop in rapidly.

I can make things much faster in Svelte, epsecially since Svelte is less complex than React. With AI code generators, I don't really even have to think, just tell it what I want.

Of note, my main career isn't programming.

How does this compare to goravel?

I'm a third year radiology resident. That speaker is misinformed as diagnostic radiology has become one of the most competitive subspecialties to get into. All spots fill every year. We need more radiology spots to keep up with the demand.

As a radiology resident, it would be nice to have a tool to better organize my dictation automatically. I don't want to ever have to touch a powerscribe template again.

I'd be 2x as productive if I could just speak and it auto filled my template in the correct spots.

As a former radiology resident, I totally agree. That's why we're building exactly this: https://md.ai/product/reporting/.

For the web, PHP remains #1. It's super fast, especially when using FrankenPHP or Swoole. Laravel has the fastest development velocity out of any framework available.

It's easier for me to spin up postgres and redis with automated backups using coolify than manually write a deploy/backup script.

Outside that I deploy apps outside coolify as it doesn't scale automatically.