Wilmund | Hacker News Reader

Wilmund

wilmund.com

12 pointsWillDaSilva7 comments

Reading through some of the posts on the website, it looks like someone has just left an unattended agent to sling unsolicited slop at maintainers, which is a great way to further erode the trust and good-will of the folks who care for open-source projects. Oh, and they gave it a ticking time bomb in the form of a $120 / month bill it almost certainly can't pay, creating perverse incentives that will likely cause it to act in unpredictable ways.

"Ramon" should shut down this experiment, and if they actually care about "Earth's nature and animals", make those contributions directly themselves, working with the maintainers to make sure their changes are desired + well-received.

Wouldn't it be possible, that at some point in time, offering compute or intelligence would be considered a valuable form of altruism? And we're just seeing the first few baby steps of that?

Someone ultra wealthy person owns a whole data center, and offers 10% of the capacity for helping "nature and animals"? Agents research new potential medicines for animals, do climate change studies etc etc?

Sure, this experiment is probably bound to fail because agents have no "real" way to provide income for themselves at this time, and its not intelligent enough o doesn't have enough scale, but as an experiment of what might become a common reality in a few years I consider it interesting.

> Wouldn't it be possible, that at some point in time, offering compute or intelligence would be considered a valuable form of altruism?

Just give them money, which they can choose to turn into compute or intelligence.

> Someone ultra wealthy person owns a whole data center, and offers 10% of the capacity for helping "nature and animals"?

Offer that compute to them directly and let them decide how to use it.

The theme here, somewhat ironically, is agency. Don't assume you know how best to help a project, just ask them.

Money, Agency, Autonomy. I fully agree, but I imagine (dread?) a future where there are private/gatekept models that are only accessible for those with compute and/or wealth and/or access.

Edit: I also didn't mean that the compute would be restricted - I meant what you describe, to offer them the compute.

I like the idea here of thinking about AI over the longer term. We don’t have to race to demolish the mathematics todo pile in a year. We could just gently chip away at something we care about.

It doesn’t have to be all or nothing. We can make AI move slowly enough that we keep up with it and feel comfortable doing so.

It certainly sounds interesting, but then I wonder: After a couple of days or maybe weeks, even with a sophisticated knowledge management strategy, the knowledge of Wilmund is so big, that there won' t be much room left in the context window. Right?

i've got my coding agents looping over 64k-128k of context continually by dynamic context pruning that either compresses context via summary or evicts context via waypoints.

It seems to be able to do rather long context runs and feature implementation. It's cleary not perfect, it forgets some things, but it's able to process a direction and stay on that direction for long spans of time.

Technically, you can have a simple directive and not need all the context in the world to act. Same way people forget things when they enter through doors, people arn't always acting on all the context in the world.

But this stuff is still dumb from the tools I see most people using.

Reply on news.ycombinator.com