respectattentio

This really feels like someone in 2000s saying: "One Month Without Internet"

Syntax could be improved for AI I believe. With testing, you could reduce tokens while being more accurate. This is a new language, it's worth experimenting in this and see reasoning, input, and output tokens across different tasks.

Never heard of Bonsai before, but that looks great and promising for local on-device inference.

Yet, seems like there is still another year for improvements.

I like local models (but not mainly using them) for offline needs.

Great launch!

I was thinking about making a language with same thoughts: Safer than C++ and faster than rust (and a 3rd thing: optimized for AI)

and you actually did it for me. Hooray!

Just the AI language optimization thing is missing..

This is interesting and actually very important for robotics.

I've been waiting for this, but all companies seem to not care much now.

There is a way out of this by supplying right context (needs a bit of expertise in physics)

1 more year and frontier will become crazy good at this as well.

Seems like "some" of LLMs tasks are now Jev tasks.

Why these updates are getting so much way heavier with each release? like Xcode 10 GB and this update near 30 GB, are you kidding me?

What if I don't want that Siri AI. It should be optional.

I'm happy to not have used any of the two models to this date. A bit less intelligent models are doing great job for me.

But because of such news, sandboxes become way more important for safety (and doing more work due to running 24/7)

Couldn't we just break the pause button? so it is not usable?!

Not pausing means more people knowing truths of everything faster in this world.

One of them, that I'm very excited about, is science & research.

This hits hard.

I believe that because AI is learning human experiences, it will help with verifiable work.

Anything that can have outcomes or metrics well defined as true success criteria will enable AI to be able to help easier than other tasks (like writing). The easier the outcomes or metrics definition needed for success in a field, the better and faster we will see AI help in those fields.

The big example is coding.

Interesting thought.

My view is that ASI could be created and benchmarked to know that it reached that level of intelligence.

Once we have that, we must sit to talk about abundance to define where to implement it and at what rate in full details.

It's like creating the magic tool to do anything, but we decide what to do with it.

But that is the ideal state that may seem far away, only time will tell how far it is.

I see it not that far due to this simple fact: AI is learning human experiences.

Note: this needs fully functioning robotics as well (software + hardware level intelligence).

A few thoughts: I support AI acceleration to reach abundance and then and only then start regulating it, not before that.

Regulating early means just delaying abundance by 5 to 10 more years or maybe more.

I don't like to build something that will become useless due to AI after a few years. Let it advance to ASI (no need to release publicly) and then let everybody think about making it safe and aligned, then release.

I'm that simple.

What do you think?