czep

I find this experiment inspiring, but probably not for the reasons intended by the experimenters. To let loose a swarm of agents in a sandbox game with the vague goal of "build a thriving civilization", and then to observe them flounder in risk averse strategies is precisely what separates us from them. There is no risk-taking, no irrational spark, no neurodivergent "because it's there" impulse that led us out of the caves in the first place. Perhaps it's impossible for a fully rational intelligence to build a civilization, as analysis-paralysis would exponentially dominate all decision-making. It's only happened because of happy accidents, ultimately stemming from what would in the moment be thought of as very bad ideas.

I am equally intrigued by how to augment Minecraft game play using agents as, essentially, team members to command. Given the ability to communicate with agents who are autonomous enough to play the game well enough, could they assist a human player to achieve very ambitious goals? Spawn in with 50 "friends", and instead of letting them wander aimlessly, give them some concrete tasks that a human player knows to be valuable steps on the tech tree. Build a cobblestone generator, an iron farm, lock in a Mending librarian, find a stronghold, loot the End, a sugar cane farm, gunpowder for rockets, now everyone has Elytra, on to raid farms, a perimeter, slime farm, witch farm, etc.

We don't give agents a rack of servers and an objective "build the next unicorn", but we direct them to solve tasks we as humans know are valuable in our own domains. Same story here and I am seriously daydreaming about a server in which I've got 50 full Netherite agents at my command ready to assist me in whatever grand build I have in mind. "Ada, I need 20 shulkers of black concrete. Bram, bring me 14 stacks of blaze rods. Cleo, we need more blue dye, find the optimal gradient in the flower forest and build a shifting floor farm for cornflowers..."

The economic question is whether the average company will have the time or talent to roll their own models instead of eating the cost increases. The firms in question are exactly the same that have already decimated their teams. Can they so quickly pivot to self-hosted models if their AI workloads suddenly cost them 10x more? I bet most will simply start shoveling themselves deeper.

It is really beautiful to see distilled to the core algorithm. Despite AK’s claim that everything else is simply optimization, I am sure there’s a bit more to training a useful frontier model! Still, an excellent teaching tool and nice way to spend an afternoon walking through the code.

While true, it's still a user-hostile move. You kinda have to meet your customers where they are. If people are clicking ads without knowing it, that's a serious design problem. Yes, people should learn to read, but the risk of placing too much burden on users is that all it takes is one ambitious product manager to push an A/B test that generates huge revenue wins while enshittifying the product for everyone else.

> You need to know that your data and conversations are protected and never sold to advertisers.

> we plan to test ads at the bottom of answers in ChatGPT when there’s a relevant sponsored product or service based on your current conversation.

There is a severe disjoint between these two statements: the advertiser now knows what your conversation was about! This gives a lot of leverage to ad campaigns to design the targeting criteria very specifically crafted to identify the exact behavioral and interest segments they want.

If you’re referring to body shop consulting agencies this may be true, but IME as an IC consultant in DS/ML, my rates are well above Staff+ at FAANG, and nobody has ever tried to leetcode interview me. Yes, I have to do a lot of smooth talking, but performance is extremely transparent: if I don’t deliver, I don’t get paid. Honestly I doubt I could pass a leetcode style interview, and I’m glad I don’t have to do that anymore.

You're right, of course, there are suitable alternatives that work well. There are some things that will just be easier to manage such as Parental controls, backups, and Family features. Additionally, I've lost basic tech support skills on iOS. Part of my motivation for originally switching to android was to gain familiarity with the ecosystem, but now I've fallen behind on the iOS world and can barely use my kid's ipad.

I lean towards your viewpoint as well. Their assumptions (axioms, postulates?) are highly controversial, while the actual principles seem quite sound to me.

The only issue I can see is with #5. I would argue for decision making, you absolutely need a single metric, otherwise the process collapses into bickering over which measure is more important at the time (often for political or interpersonal reasons). The point is a bit vague on what exactly is being evaluated (product quality, which means what?). For launching products or running A/B tests, aim for a single metric as your decision framework. If you must have more than one, then be explicit about the tradeoffs in a flowchart: e.g., "if X is > 0, we launch. If x <= 0, but y > 2%, we launch, otherwise no launch".

There are too many insightful and thoughtful points in your comment for a single upvote to repay, so I must thank you with a reply. Becoming so demoralized with the present media and political landscape, it is comforting to find that there still remain semblances of reason for which I may again briefly restore some measure of sanity. Thank you.

Think of the people you've helped get their jobs done over the years. That's worth more than author credit in the colophon of a man page.

Bluff? Putin wants to live out his golden years in unabated luxury after anointing a loyal successor. If he nukes the entire world, he will live in squalor and misery while being hunted by the entire human race. He won't do it. Me hopes. Me also hopes someone in his inner circle gets a nice clean view of the back of his head and does the world a favor.

Any recommendations for best practices in "modern C++"? I have to resurrect an old codebase and would like to refactor it using more modern idioms, but its been a loooong time since I wrote any decent c++.