Hacker News: pharke

New comment by pharke in "Dalai: Automatically install, run, and play with LLaMA on your computer"

pharke — Tue, 14 Mar 2023 13:19:18 +0000

Given a choice between pre-industrial life and our current lifestyle, the choice is obvious.

New comment by pharke in "Dalai: Automatically install, run, and play with LLaMA on your computer"

pharke — Tue, 14 Mar 2023 13:12:26 +0000

Explain how a language model can “evolve” into AGI.

New comment by pharke in "Dalai: Automatically install, run, and play with LLaMA on your computer"

pharke — Mon, 13 Mar 2023 15:56:57 +0000

LLMs will never be AGI

New comment by pharke in "Dalai: Automatically install, run, and play with LLaMA on your computer"

pharke — Mon, 13 Mar 2023 15:55:37 +0000

We currently live in a world that has been “unrecognizably transformed” by the industrial revolution and yet here we are.

New comment by pharke in "Dalai: Automatically install, run, and play with LLaMA on your computer"

pharke — Mon, 13 Mar 2023 15:52:11 +0000

Could’ve said the same for any major technological advance. Luddism is not a solution. If these models are easily run on a laptop then yes some people are going to hurt themselves or others but we already have laws that deal with people doing bad things. The world is not going to end though. Your Taiwan scenario has a much higher probability of ending the world than this yet you seem unconcerned about that.

New comment by pharke in "ChatGPT's API is so good and cheap, it makes most text generating AI obsolete"

pharke — Sun, 12 Mar 2023 18:53:01 +0000

They charge per 1k tokens so you must have high volume somehow, are you maxing out the prompt length every time? That’s the only thing I can think of besides sending a ridiculous number of requests that would cost that much in an evening.

New comment by pharke in "ChatGPT's API is so good and cheap, it makes most text generating AI obsolete"

pharke — Sat, 11 Mar 2023 21:08:42 +0000

You could have saved some money by writing tests. How much text were you sending at a time? I’ve been summarizing multiple 500 word chunks per query in my app as well as generating embeddings and haven’t broken $10 over the course of a couple weeks.

New comment by pharke in "France's baby bust"

pharke — Sun, 26 Feb 2023 19:40:01 +0000

> I argue that the diminished sway of the Catholic Church, nearly 30 years before the French Revolution, was the key driver of the fertility decline

> ...the Catholic Church, threatened by the spread of the Protestant Reformation, took ‘be fruitful and multiply’ seriously and the purpose of marriage became explicitly multiplicative

> The decline of Catholicism, and fertility, in eighteenth-century France turned it from a demographic powerhouse – the China of Europe – to merely a first-rank European power among several

New comment by pharke in "France's baby bust"

pharke — Sun, 26 Feb 2023 12:19:25 +0000

Blaming the Catholic church specifically for high birth rates is a misstep especially since the author compares to England which has a long and varied history with Protestantism. The argument should be reframed around religiosity in general or he should show data for England that indicates a difference in fertility during the periods of stronger and weaker Catholic influence in England.

New comment by pharke in "Ask HN: Isn't ChatGPT unfair to the sources it scraped data from?"

pharke — Mon, 06 Feb 2023 13:06:06 +0000

It's very much in their interest, if the information their models provide is impossible to verify then it severely limits its uses. You essentially can't use it as a source for anything that requires any type of citation or reliability. That's a huge handicap for selling it to businesses and researchers. The general problem of determining what training data was used to produce an output is an open problem in ML and one that is being very actively worked on since it would greatly further the field.

You believe correctly that ChatGPT is not capable of showing sources, it's currently impossible to do but we were discussing Tomorrow so I included it as a possibility. You could potentially hack it in now using traditional search or nearest neighbours but it wouldn't be 100% accurate, probably not even 50%, it would just show a bag of similar texts so not really worth doing.

I'd still be in the market for a book even if we had a perfect LLM that could answer every question I had with impeccable accuracy. I read books because I want to find out about things I don't know that I don't know. It's pretty hard to find those things if you just do question response. It's like a graph, if you start at one node it may take you a very long time to traverse the graph to another node but if you have some outside source that gives you the address of a new node you can just jump straight to it.

New comment by pharke in "Ask HN: Isn't ChatGPT unfair to the sources it scraped data from?"

pharke — Mon, 06 Feb 2023 00:01:01 +0000

Why would someone only ask an LLM questions when they were in the market to buy a book? Most people I know don't buy books in order to look up the answer to a question, sure some people buy reference books and use them but that's not really what we think of when talking about authors and books. If I'm in the market for a book, I'm looking to read a book, not query something or someone for answers. I think your example should go like this:

Tomorrow: 1) you do research, write posts, publish a book, 2) it is all consumed by a for-profit operated LLM. 3) People ask LLM to get answers to some related question or interest 4) They ask the LLM for a list of recent books that go in depth on the topic or are in the genre etc. 5) Your name comes up in the list 6) Goto step 2 from Yesterday

New comment by pharke in "Ask HN: Isn't ChatGPT unfair to the sources it scraped data from?"

pharke — Sun, 05 Feb 2023 14:22:50 +0000

You can also invert this and say that without a system like ChatGPT it is physically impossible for most people to find or use those 570GB of data. A search engine can only get you so far and over time they are becoming less useful as the net floods with junk content. If you don't even know what terms to search for then ChatGPT wins out since you can start with a very simple question and then interrogate it further on details it produces. The best way to think about it is as a better search engine, a fully interactive one that also has some degree of its own agency when it comes to synthesizing data. It could be better, it would be nice to have the option to show sources for the output so that you can verify the facts or do your own research.

New comment by pharke in "Show HN: I “wrote” a kid's book with ChatGPT and Midjourney"

pharke — Wed, 25 Jan 2023 12:52:51 +0000

I'm of the opinion that all kids books should be heavy on metaphor and allegory and all of the other important and joyful aspects of language. Restricting children to books that have been "written at their level" or "made easy to read" is doing them a disservice. It's no wonder so many people grow up to be seldom readers if their first exposure to reading is saccharin pablum that presents no challenge, no meaning, no mystery, and no danger. A proper children's book should be one which an adult can also read and enjoy if they give way to that childlike wonder that still lives within them.

New comment by pharke in "Microsoft has laid off entire teams behind Virtual, Mixed Reality, and HoloLens"

pharke — Tue, 24 Jan 2023 13:39:24 +0000

I've also played a few indie VR games that are using it. It's not like you have to be some technical wizard to enable it, you just need to make sure you aren't stomping on your FPS with a bunch of other stuff at the same time.

New comment by pharke in "Microsoft has laid off entire teams behind Virtual, Mixed Reality, and HoloLens"

pharke — Sun, 22 Jan 2023 22:37:25 +0000

https://twitter.com/iBrews/status/1570844710155190275

New comment by pharke in "Microsoft has laid off entire teams behind Virtual, Mixed Reality, and HoloLens"

pharke — Sun, 22 Jan 2023 22:31:23 +0000

On the price point, the original iPhone retailed for $499USD which in inflation adjusted dollars today would be $714USD. The Quest 2 is significantly cheaper than that as a standalone device. The Quest Pro is much more expensive but is targeted at businesses, watch Zuck's intro of it if you doubt that. The biggest barrier is that for really good performance on any headset you need to tether to a gaming PC. That's a lot more cost and complication for non-technical people. That said, the hardware is there and capable of delivering outstanding visuals and inside out tracking. The compute and networking are lagging behind, we really need a small, user friendly and powerful device that anyone can go buy, plug in and then connect their VR headset to with 0 frustration. That would free up headsets to only worry about the displays and tracking.

I don't think the future of VR is portable. I have an aversion to lumping AR in with it since they are fundamentally different. If it makes it easier, VR is desktop, AR is mobile. Similar but not at all the same. The way I envision the future of VR is that it does for your physical experiences what computers and the internet did for your paperwork and documents. Thinking about VR as a cellphone is the wrong mental model, it's not a peripheral that exists in your physical environment. It is a replacement for your physical environment or more congenially, it is another physical environment to which you can travel in addition to the places you currently go to.

New comment by pharke in "Microsoft has laid off entire teams behind Virtual, Mixed Reality, and HoloLens"

pharke — Sun, 22 Jan 2023 01:39:02 +0000

Nah, Microsoft and Meta are obviously partnering up. This is likely just a symptom of that with Meta getting buy in from Microsoft to use Meta's VR platform. I agree that Horizon is the worse of the two but that doesn't mean they'll fail. Worse is better sometimes. The lynchpin will be if Meta is really married to Unity, Unreal Engine is eating Unity's lunch with adoption and features. If Meta makes the jump to UE they'll win a bigger share of the market with a far superior engine. It waits to be seen.

New comment by pharke in "The Fermi paradox revisited: Technosignatures and the contact era"

pharke — Tue, 20 Dec 2022 15:46:23 +0000

If we're talking about galactic distances and timescales I don't see why we couldn't safely assume enough technological advancement for either aliens or humans to modify their own biology to the point where centuries would be a small fraction of your total lifespan. I think the bigger question is why any significant number of people would travel to more than a few star systems if you had to do it slowly. If the system is uninhabited then you would be isolating yourself from the rest of society for a long time. If it's inhabited then you would have the excitement of discovery and contact but it would still be difficult being separated from your own civilization and you would want to return at some point even if just to share what you've learnt.

How many people today travel to places that are completely uninhabited and have no infrastructure in place for getting there? I'm talking end of the road, getting on a bush plane and landing in the middle of an open field kind of isolated. How long do they stay? Same for contacting the few remaining hunter gatherers on the planet. What if those trips took thousands of years instead of a few weeks?

I think there's a big stretch of time for the development of any technological civilization where they spend most of their effort on developing their own solar system and only send small flyby probes to their nearest neighbours. After that there may be an even longer period during which they colonize those near by stars but each colony likely takes an extremely long time to fully develop. The civilization may send out automated probes to look and listen and maybe the occasional explorer or anthropologist but I doubt there are any galaxy conquering civilizations out there.

So I think I do agree that if anything visits us it is likely to be an automated probe. Bracewell probes would be well suited to any chatty civilization that is seeking contact. Von Neumann probes are more suited to brute force exploration.

This is all assuming that the speed of light is the fastest you can travel and communicate. If it isn't, then I'd say all bets are off.

New comment by pharke in "The Fermi paradox revisited: Technosignatures and the contact era"

pharke — Tue, 20 Dec 2022 15:25:43 +0000

That's not a great starting assumption. How would the economics of travelling hundreds or thousands of light years in order to "consume" (whatever that means) other life forms work out? What is the benefit? How would it be worth the investment?

New comment by pharke in "The Fermi paradox revisited: Technosignatures and the contact era"

pharke — Tue, 20 Dec 2022 15:23:02 +0000

I would have to assume that they know this and are in fact only looking for messages that are purposefully sent at very high power. It's also why they favor listening at the hydrogen line since transmissions at that frequency stand the best chance of penetrating through cosmic dust, i.e. you would only use that frequency if you were intentionally trying to transmit across interstellar distances.

I think they don't popularize the downsides of a radio search since it could threaten their already tenuous funding. Basically, they have to be optimistic in order to maintain the pittance of a budget given to them which allows us an extremely small chance at detecting a purposefully sent signal if we happen to be in the right place at the right time and listening on the right frequency pointed in the right direction, etc.

In order to do SETI right we really need radio telescopes in solar orbit far away from the Earth and enough of them so that they can cover all of the nearby stars or other likely directions for signals to originate from. The only way we're going to get there is if space launches continue to become a lot cheaper. Fingers crossed.

https://en.wikipedia.org/wiki/Hydrogen_line