doctoboggan

Yes, I am projecting out into the future, not talking about the current state. I said "might be the first".

> A band can be entirely findable in 2006 and entirely gone by 2026 depending on which technologies they happened to pick.

This is interesting, and in some cases desirable. I wonder what technology choices one could make today if they wanted to have an internet presence now but not have it stored forever. Would that even be possible today?

I am not so sure that's true. Even if AI intelligence were to plateau at today's levels, there are still many gains to be had in speeding up today's intelligences. ASICs with burned in weights could become economical to invest in as they would retain usefulness longer than 18 months, and I feel we have only scratched the surface on possible usecases of local AI and what it means for almost any technology product or interface.

It's been very interesting watching Tao's evolution on his thinking on LLMs. Of course the LLMs have themselves evolved so that shouldn't come as a surprise.

The job of professional mathematician might be the first to be completely eliminated by LLMs, save for those who can make money from a patron. I am hoping they are able to figure something out to save their profession, as other professions could use it as a blueprint as AI comes for them next.

You are conflating "important" with "good", which I don't think the author was intending.

I am mostly at the same point right now you are, but I think in the future with those "gas town" ideas we might be managing even more agents each.

Also, I've recently begun experimenting with specific tasked agents running on a cron like timer for non-dev work. (checking emails, managing small business tasks, etc). Once I started using Claude code in this way, the number of agents I can imagine running has skyrocketed. So I guess what I am saying is that I look forward even cheaper tokens going forward.

Other than what the others mentioned about finding more efficient representations, you can also compress by pre-agreeing on some common terminology.

In many ways, we are communicating using a compressed channel (words) since we both have pre-agreed on the meaning of these words.

I honestly haven't done enough thinking on it, but it doesn't strike me a priori as ridiculous to limit people's ownership of these nation-sized companies, and to possibly limit the size of the company as well

If an amazing drug or other therapeutic falls out of this I bet they would be happy to own it outright.

I believe that you just want to make a really good harness, but is that all amazon wants it to be or do they want more?

Have you ever heard someone say “pace yourself” when you are eating too fast or otherwise rushing too much?

This is why the chrome driver plugin and similar tools are so important. The site shouldn't know or care if its me or my agent working on my behalf.

Someone should make the millennium problems for AI alignment so the frontier labs will actually try to solve that problem.

Does each separate “organ” here have any relationship to the two separate functions identified in “Thinking, Fast and Slow”?

> I thought that the standard line is that 3d printing metal wasn't developed enough for anything serious.

This hasn't been true for over a decade. High value relative to weight, highly complicated internal geometry, or repeated need for one off parts are all reasons to choose 3d printing today for production parts.

Nit, I know, but I don't think they are trained on too much pretentious writing so much as they are RLHF'd into it because that writing impresses people.

> a sort of epistemic performance art project.

Agreed, and it's a wonderful piece of art. I look forward to seeing the actual publication and reaction from the math community.

> forecasting = modeling = AI

I wouldn't go that far. Humans can forecast by modeling with their wetware, nothing "A" about it.

Use Claude code and the chrome plugin to access reddit.

I recently purchased some very rural land with a cabin that I put on AirBnB. The address is a different road than the driveway is and all the maps took drivers around 1 mile away with no way to find the actual driveway. I had to go through the process of fixing this on Google, Apple, and open street maps. OSM was by far the easiest and my updates propagated within minutes. Google was the second easiest but still took me deep into the “feedback” menu and then a week or two of waiting. Apple was by far the worst as I had to sign up for an Apple business account, upload incorporation docs, and then go through multiple rounds of back and forth with some approval agent.

The administration is way behind the curve on this one. Data center opposition is one of the rare areas that has bipartisan support.

v1 to v2 is the biggest leap, and each leap after that is smaller and smaller. Just look at the difference between the 17 pro and 18 pro.

Just a reminder to everyone, always buy the second version of any new product category, even from Apple. The first duo will look primitive in a few refresh cycles. It does look really cool though!

I use it for audiobooks. They are served up by Plex and I use the Prologue iOS app which is a great audiobook listening experience.

I think all he big labs are pretty explicit about when they do and don't train on customer prompts. Is the accusation here that OpenAI trained on prompts when they claimed not to? Or were the mathematicians using one of the interfaces that allows OpenAI to train on the customer data?

Claude will be able to help you with everything you need. If you don't have Claude code then the various self hosted subreddits are a good starting point. I had Claude do a sweep of the recent reddit posts to capture the current most recommended usenet accounts, he gave me two links, told me to sign up and paste in the API keys and he took it from there.

After reading a lot of reddit threads I settled on ReadMeABook[0] and it has been working great so far. One pretty cool feature is that it lets you connect an LLM API key and then you can have it make requests based on your library. I now have Fable 5.1 suggesting my next book.

[0]: https://github.com/kikootwo/readmeabook

Thanks, fixed

Step one is to just let Claude have ssh access to the machine you run plex from and let him do a full sweep. I had some conflicting profile rules and was not properly blocking malware and not properly marking failures so they were never retried. This fixed a big portion of the manual work needed, but the usenet account is really what got me to nearly 100% hands off. I think I am paying $3/month for the next fifteen months, and apparently if I pay attention during Black Friday there will be some pretty cheap deals. For more months of access at even less

EDIT: Claude was also able to look at my logs and noticed that nearly all of the exe or other malware was coming from 2 trackers so we just eliminated those.

Yes it really does, and I intend to switch eventually but from what I heard the last time I looked into it, the end user clients are a big area where plex still beats Jellyfin. Since that’s where my non technical friends and family mostly interact with the system, it’s the area I need to be the most polished and hands off.