MinimalAction

Rambler for Pixel 11 Pro. Runs a local AI model Gemma 4. Eloquence on Apple store released by Google comes close too (most features run local models and you can cut off cloud access).

If you use an android phone, it's easy to find patches for GBoard keyboard for turning on this new feature that was kept exclusive to Pixel 11 series.

I am surprised by the title and the story. Understanding has always been the bottleneck; there is nothing new about it. The argument goes like we humans should understand so we can verify and participate. How bold! Maybe we should have been doing that all along...?

For what it's worth, I use Brave and it didn't render properly in a normal tab, but rendered correctly on a private tab! Underlying engine is Chromium anyway, so make of that what you will.

No. If they still draw an enormous amount of water, that's the finite resource that's getting wasted no amount of money could immediately retrieve. Paying for electricity doesn't mean the resources are not shared.

It is the very humanity that is absent in all of this. The power is still in the hands of a very few. I feel you're dismissing this with a grandeur vision of AI developement.

So, you're arguing that just because mathematicians are a minuscule, it doesn't matter for prosperity? Because, if so, that is an insane take honestly.

Absolutely not the same! People need jobs to bring in income. I don't believe those who profit off of this will share it with the world. The power is all concentrated in the few hands that decide whether or not the rest get any semblance of income in the long run. I don't believe UBS until it happens.

I hate this timeline. I might be excited for the kind of answers this AI builds for unsolved problems, and also for learning new things by talking to it. But, I feel like I'm in the minority of people here who feel this could be a net negative endeavor with this having to kill a lot of educational institutions and their ability to fund themselves in the long run. It's not worth that.

Thank you! I don't like substack at all. I hate the typography overall. It's not hard to power up a static website, especially these days using Jekyll, Hugo, or hundreds of other such framework in so many different languages.

The only thing good about Substack is to find new people to read from, but I'm less inclined to follow them for long because who knows how Substack changes things.

That's fair as well. It occured to me as I wrote my post that not everybody would glue themselves to screen while eating.

But how do you read? On a device?

Sounds like I'd enjoy those videos as well! Any particular recommendations?

I feel seen and heard.

I'm conflicted in feeling there's now a lot more to do with AI, but also if it ain't me doing most of it, where's the fun in that? Days feel meaningless, and don't know who will pay for what I want to do, after all these years of heavy training.

Disagree. In many professions, people were paid to endure the boring parts to be able to finally do fun parts. See academia; a professor writes grants, attends committee meetings, teaches (sometimes undesirably) and such, to finally go do research which is fun. If the fun parts are predominantly done by AI, and are removing people from getting paid, what fills one's stomach?

True, but many of us have grown identifying ourselves with the kind of work we do, the discoveries that follow from our hard work. Even in the case of rockets, we designed the rockets! Here, I've to let the oracle take care of things while I'm robbed of that delight of discovery.

As an academic, I concur with the findings. However, some points for discussion:

1. I don’t believe iPad doesn’t offer the same benefits of writing with pen on paper. I mean iPad writing can be made to be almost as realistic as pen on paper with good accessories. It can be slow, doesn’t waste papers, you can edit, color code it, annotate in a way that’s backed up. I definitely like writing on paper, but I don’t see too much of a difference with iPad.

2. I absolutely love using Pilot V5 RT pens, and haven’t found something that is as smooth and satisfying when put on paper. What are your favorite pens? Are they available on JetPens for trying? Thanks!

A random professor's academic website is usually barebones, no CSS, only HTML, mostly text. But the value one finds there is incredible.

As opposed to this, vibe coded slop which neither explains the stuff that's already there, nor is pedagogical. Why use AI for this?

Yes, of course mathematics is a tool to model nature as accurately as we could. That still doesn't mean mathematics is as trivial or frivolous as manipulating symbols.

That view is incredibly reductionist. It really is an efficient encoding of how nature behaves. It might be a human construct, but given how best it allows to understand nature (through principles of physics), it is uncanny to be any different from the language of nature.

Reminds me of Wigner's Unreasonable effectiveness of mathematics in natural sciences [0].

[0]: https://en.wikipedia.org/wiki/The_Unreasonable_Effectiveness...

Extremely cool. I'm sure they'll eventually be trained to read it, but it's nice until then to trick AI.

I'm mad at AI companies for stealing texts from the entire internet knowledge base and now privatizing those profits in some sense.

As a scientific adventure, SpaceX is a worthy company full of awesome people. But the management and VCs is another story, as usual. To price it at a market cap of $1.8T, somewhere double that of Walmart is insane.

I'm tired of these types of articles making it to the HN frontpage. None of these articles can say anything definitive; all they could do is speculate. We are trading away the old HN which was full of really interesting links that used to send me on rewarding rabbit holes for hours.

I liked the idea, but this literally could have been a list. The AI-generated pages with filler text create a stark contrast in utility compared to the influential nature of the listed papers.

It is insane to me that Apple couldn't build its own source of RAM given its 250B war chest. Could have bought some RAM company at an opportune moment.

That makes a lot more sense when you put it that way. Nevertheless, it is ever more harder to prove something is legitimately one's own work. That is worrisome when it comes to science where credits matter more than for glory; job markets only value publication outputs which may have been cheaper to obtain with LLMs now and harder to distinguish what's yours and what's not.

Given the advent of LLMs, I don't know if plagiarism is ever a thing anymore. Nobody is stupid enough to include verbatim unless citing those works. Feed into an LLM to get a paraphrased version conveying the same meaning.

It is worrisome that the scientific machinery as it stands needs an overhaul in LLM era.

What's the biggest draw of GrapheneOS apart from de-googling? Does it have a better battery life? And compliance with NFC payments?

Well, for some reason Pixel 9 series and also 10 pro is excluded?

Interesting that they disclose this right after IPO.

Often this is the conundrum in research as well. What should one spend their life working on? Especially if you want to make an impact. Choosing the right problem is often harder than coming up with a relevant solution.

Just looked up Chinese production capacity (~1700 GWh). That's orders of magnitude higher than the rest of the world. What did they do differently?