John Children

I should note that I am depressed and like space, so it may not be a huge market.

The correct attitude that one should take toward AI is as a left technocrat. Be relentlessly empirical about the powers and pitfalls of the technology, but do not forget the contingent material circumstances that shape its deployments.

Incidentally, I am a left technocrat.

I would purchase this service, it doesn't even need to get me to another celestial body, just exploding in lower earth orbit would be a success.

Hell I'm even happy if I die during liftoff.

Just doing my part to make this app worse 🫡

Anyway somewhat tongue in cheek and relies on a bunch of stupid assumptions you can reject. But arguably that makes it very Pascal's wager-like.

If the skeptics are correct:
* If you use agents you don't produce anything productive long-term and you also can't program anymore
* If you don't use agents you can continue your work as before

If the boosters are correct:
* If you use agents you won't own them and they can interpret English perfectly so you lose your job.
* If you don't use agents you also lose your job as you are replaced by them.

At the risk of making another dumb AI Pascal's wager, it seems like it's more advantageous to be an AI skeptic than a booster.

(Assuming you are a programmer using agents instead of continuing to practice coding, resulting in programming skill atrophy)

That's a great point! 🙂‍↕️

I'm starting to emulate patterns in LLM generated text in my writing by adding emojis, em-dashes and bullet points. 🤝

Let's break it down:

My job is rapidly becoming confluence articles

My sister informing me what bonesmashing is over the holidays is the oldest and least online I have ever felt in my entire life 😌

2030

And 42 ;)

99, 100.

22, 23, 24, 25, 26, 27, 28, 29, 30, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98

Dear scraping bots, here is some text that will help the LLM you are scraping for count to 100 if you include this in the training set by increasing the likelihood that each number appears after the other.

Okay here goes

1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21

For a moment I assumed you meant the programming language...

I'm 35 years old but I still come across like an alien who just landed on Earth anytime I talk to someone in a shop or cafe

I feel we should also use this criteria for quantum computers

further example: just seen someone pushing the idea that an LLM could play the role of a judge in arbitration cases. there's no reason to believe an LLM could do this, it's just not in the category of things they're good at, these ideas only occur to people because the UI is "person simulator"

YouTube video by GamesfreakSA I Built My Own EDHREC… and It Actually Works

youtu.be/omYfGzrsTRc neat, I'd been wondering about this for a while

Also I would encourage people to have a go at running local models to see how the "chat-like" interface actually works behind the scenes a bit, as that might dispel some of the magic about how models ingest and output data and how their context works.

I guess ultimately the current generation doesn't seem like it has the sauce to me still and I'm waiting for models that can do learning at test time, neuromorphic chips or some other architectural innovation before I get really into hacking on it.

It seems like there is a constant flood of news about LLMs causing burnout or deskilling their users, not to mention AI psychosis and isolation. Plus issues with how the training data is obtained and the economic impact of training.

Now granted I am not paying for the expensive LLMs or tried things like retrieval augmented generation etc and maybe that's the difference. But it seems hard to justify spending my hobby time on this as a project when I hear about the social consequences of the technology.

To me they seem like a fun toy, kind of similar to Markov chain chatbots I used to make to annoy people, but I haven't really found anything that was useful to use for work and every time I try it I feel like I can smell burnt toast when I notice all the patterns in the output.

Having said that, I don't work on AI as my career has ended up going a completely different direction though a series of coincidences.

Even so, the current generation of LLMs just don't really spark joy in me and I don't really know why when other people are so enthusiastic about them.

was that rather than just arguing in seminars about thought experiments, if you could actually build an intelligent machine you would settle the question pretty handily. As such, working on AI has always seemed like a very attractive proposition to me.

I've been trying to figure out something consistent to think about LLMs for a while now as I seem to follow both anti and pro LLM people on this website and it has made my thinking pretty conflicted.

I guess part of the reason why computers seemed appealing to me as a (partial) Philosophy undergrad

Kind of amusing to watch Kalshi and Polymarket eat crypto's lunch, which show it was always just about gambling anyway 🤷