machiaweliczny

AFAIK the whole morphology is decided in distributed computation fashion via electrical potential changes, at least according to experiments by Michael Levin

Yeah compared to animals we have a lot of extra bootstrapping outside of physics/chemistry alone via culture and stored information similar to how cell DNA bootsraps via physics, human mind boostraps via stored information in human "network" (talking, internet, books) after being born.

I can see a perl comeback

You can also use "repomix" tool to bundle whole source of godot into single file and tell it to search it when uncertain

Just add godot example games nearby and it will learn functions / usecase from them. Just say in instructions BTW you have example games in "examples" directory to check

Vision is very far from solved IMO, simple modifications to inputs results in high differences still, lines aren't recognized etc..

I think it's due to improvements in vision basically, the arc agi 2 is very visual

It's also easy to spot as when you are tired you might misrecognize objects, I caught myself with this when doing long roadtrips

20 more IQ would be nuts, 110 ~ top 25%, 130 ~ top 2%, 150 ~ top 0.05%

If you ever played competitive game the difference is insane between these tiers

Is there a way to do hybrid search that combines vector similarity with scalars fast using pg_vector? Or do I need to migrate to other tool?

Still doesn't understand physics as in that cover should be over the wheel, which should be easy if it used 2D space reasoning

It was also only model that was good with coming with something creative at all, like brainstorming startup ideas etc. for me - they were grounded as in reasonable compared to other I tried

It's the same as saying if you could plunder and pillage why wouldn't you

If there was something like that guarantees pseudonymous ID that can't be reversed via rainbow tables like stuff that could be useful as general login ID.

I would like to have ability to choose if I use new one for this provider or one that can be linked with other services. Kinda like with apple anonymising emails. Are there any efforts in this directions?

It's general hard to have discourse online without some good authentication as bots dominate everything. That's why normal people use mostly facebook groups. This needs some solution but gov ID isn't it, unless it's some cryptographic thing that is passed to other services that link data with their own salt or something like that.

Having said that I like how in PL we have very nice auth to all government services that finally works well (it's like 5th implementation of this system). It makes it very easy to use gov services and before had to login with bank accounts or create some crappy accounts. Now it just works with faceID and done.

You can also login to 3rd parties as well. For example I was able to get my medical health results info from commercial provider by sharing basic info from this auth app without creating account or anything (they get access to Gov ID).

Just take those ADHD meds ;)

No React won and I choose it in 2015 because FB used it for their own app while Angular was by google but never really used by their top engineers so it was crappy. I think they still use Clojure compiler to this day. FB had correct incentives to keep it stable and good. Now that Vercel overtook I am more sceptical of incentive alignment

That someone can initialize a state in if block is not something good. React won with Angular 1.0 because noobs abused two-way binding making fking mess everywhere. Now in react they abuse useEffect but it’s a bit easier to control. I work currently in Svelte and never use 2-way binding and are careful to package state mgmt well but I like it. It’s similar to react with mobx but more performant although has no good component libraries. SvelteKit is also generally fine

Distribution of wealth and disposable income need correction. It’s urgent political issue.

They already do 1000s a day AFAIK but for long range they use iranian based design which costs more

To add on top of it. As anyone in ML know the hardest part is preparing data. If the AI hype will lead to good permissive datasets, we will get a great progress once someone fixes algorithms.

AI coding diffuses the knowledge of common 80% of programming. So it should make programming more accessible actually. As it can remix well it basically can extract for you any implementation it has seen on github which I believe is great for humanity progress.

Examples of bit harder things I can do thx to AI: * write code in less known languages * do research about things to implement i had no clue about * vibe code games - this is good example of what I say below. At the moment you need to iterate: write feature, optimize code, write feature, optimize code - because it takes crappy implementation that does job but then is totally capable of fixing it

If someone would ask me for example if for python I prefer UV or AI I would tell UV for now but python had ultra bad tooling.

I believe that LLM have more potential than they currently show. As someone explained they don’t use knowledge links properly due to prediction objective. Eg they don’t attack problems from “tangents” just go in straight line, RL somewhat fixes that but it’s not there yet - it still just trick with perhaps, let me consider etc. This might get improved upon IMO and some new aproaches add tree search on top.

It would be great to have ability to customize it for other frameworks and languages in the client/server way and headless if possible. Then it can be a part of bigger platforms.

You can just tell it to read library code in npm_modules or wherever you have vendored libs in your framework. I for example give it whole demo examples and just say look at @demos/ how to do this. Cursor / CC authors don't add these prompts as this would be costly for them (and they run at loss likely now).

Most people use Vite these days or Nextjs

Change to robot mode

Why not library?

https://suno.com/song/ca5fc8e7-c2be-4eaf-b0ac-8c91f1d043ff?s... - this one about em dashes made my day :)

This. My younger brother thinks it’s crap but if you know state of the art + research it seems like things still are moving quite fast. Also tons of product work on top already.

I think it’s mostly because they are incentivised to answer verbatim as medicine students and not with their own understanding. RL methods change that.