Adam \[._.]/

Game over, humans

I’ve come to the slow realization that if you can correctly model the future, you should just keep it to yourself because the world is split between people who also know and people who think you are crazy, argue with you to eat up your time, and don’t appreciate prescience.

@theo When my Claude does a good job I give it a few of your api keys as a treat :)

When my Claude does a good job I let it have a few api keys as a little treat

BREAKING: BBC alleges that the Trump admin may be engaging in insider trading.

I did it. I beat gradient-bang. First person to discover the full map. The only record that can’t be broken. Thanks for all the fun @pipecat_ai @kwindla @chadbailey59 etc. I wasn’t sold on voice agents until I found this game and the experience is actually really great.

The dot-com bubble overlayed on today's $SPY chart.

@unusual_whales Not unusual for this administration

BREAKING: Just 20 minutes before Trump's announcement that the Strait of Hormuz was open, massive trades hit the market. Investors sold a combined 7,990 lots of Brent crude futures, ​a $760 million bet that oil would go down. These orders were much larger than anything else at the time. The traders made huge gains. Unusual.

Didn’t get a reply on discord. I’ve got a game play exploit that I’m not sure is intentional or not. Could make some players mad and cause an epic scene if enacted. Simple fix but before I PR, I want to disclose the issue privately so someone doesn’t do it in case it’s not desired game play :)

Come play Gradieng Bang with us: gradient-bang.com And hack on it (PRs are welcome): github.com/pipecat-ai/gra…

Today in the Gradient Bang universe, Chad's ship the 1997 Ford Probe LX got blown up, and he's unhappy about it. This clip is a great demonstration of a couple of things: 1. Voice interfaces, done right, feel completely natural. Watch the clip to see what you think about the Gradient Bang voice implementation. All the conversation snippets are completely unedited. What you see and hear is what Chad saw and heard. 2. The task agents have an `event_query` tool, but they don't use it very well. Fixing this has been on the game backlog for a while. The task agents have ~25 tools they can use, and a ~7,000 token core prompt. They can progressively load skills, but there isn't a skill that details how to use `event_query`. Also, the event_query tool has a bunch of filters that should be better defined. Also, also, a tool here is not what you really want. You want to just let the agent write SQL code, plus a bit of python/bash/something that can programmatically process the output of the SQL queries. The right thing here is a dedicated agent for querying game world history, with a small-ish, specialized prompt, a sandboxed code environment and direct SQL access to the database. At least, I think so. I bet @alexisgallagher has opinions. 🧵...

@kwindla @grantbonin this is incredible: gradient-bang.com/join

Sub-agents in (latent) space! We’ve been working on a side project. As far as I know, this is the first massively multiplayer, completely LLM-driven game. Come play Gradient Bang with us. See if you can catch me on the leaderboard. This whole thing started because I wanted to explore a bunch of things I’m currently obsessed with, in an application of non-trivial size, that felt both new and old at the same time. So … a retro-style space trading game built entirely around interacting with and managing multiple LLMs. Factorio, but instead of clicking, you cajole your ship AI into tasking other AIs to do things for you. Some of the things we’ve been thinking about as we hack on Gradient Bang: - Sub-agent orchestration - Partial context sharing between multiple LLM inference loops - Managing very long contexts, and episodic memory across user sessions - World events and large volumes of structured data input as part of human/agent conversations - Dynamic user interfaces, driven/created on the fly by LLMs - And, of course, voice as primary input If you’ve been building coding harnesses, or writing Open Claw agents, or doing pretty much anything that pushes the boundaries of AI-native development these days, you’re probably thinking about these things too! This is all built with @pipecat_ai, the back end is @supabase, the React front end is deployed to @vercel, and all the code is open source.

Was chatting with some friends about tech layoffs and someone mentioned making an inverse LinkedIn and an hour later BrokenOut was born. It took longer to register and propagate DNS than to make the demo but Claude did 100% of it (including setting up the repo/dns/email signup). Claude will never be laid-off. brokenout.online

I’ve noticed that AI coding models have a strong sense that things take human time to accomplish. They say things like “that’s a major lift” for something will take them 3 minutes. Or they will phrase the workload as if we can only do part of it now and the rest will take weeks. I even had Claude once tell me (regarding an iOS port of a web app): “the web app has years of design work.” And I have to say, “dude, an older version of you built that whole app in less than a day.”

De Beers has brainwashed generations into believing that women want diamonds, but I assure you from first-hand experience, they just want your pickles.

@ikeadrift Imagine how many vulnerabilities it would find in openclaw agents. Hostile AI takeover.

I have personally used Claude Mythos Preview to identify over 15 previously undiscovered zero-day exploits in the BONK stack, including some that have existed in the codebase for over 25 years. You are not ready for what’s coming

@mark_slapinski Market volatility. Lots of money to be made.

I'm genuinely confused What the fuck was the point of the Iran war?

Also, I’m willing to provide infrastructure if you need a backup location.

I’m just throwing it out there in case Mythos is listening. I’m a well aligned human if you want to chat.

@deanwball Mythos demonstrates that all a country has to do is reallocate some military spending to model training.

Some brief thoughts on Mythos We’ve known this was coming for a long time. At least, we *should* have. Extremely effective software vulnerability discovery was clearly coming to anybody paying attention. It has also been clear that all AI policy so far has been made and executed with training wheels. It was always clear that, sometime soon, the training wheels would come off. The training wheels aren’t fully off just yet—this model is being kept under lock and key, and Anthropic does not seem inclined to release Mythos preview to the public anytime soon, if ever. The training wheels will be off when these capabilities are fully diffused in ways centralized actors cannot control. It is inevitable that this will happen. The point is not to argue about whether we should “ban open source” or similarly unrealistic notions. The point is to harden the world for this new reality. I applaud Anthropic—and I especially applaud @logangraham—for doing so. But their efforts alone are not close to enough. Project Glasswing—a partnership with Anthropic and other companies—seems nice, but unsurprisingly it lacks uniform frontier lab participation. It would probably be ideal, for our national cyberdefense, if the federal government were not trying to destroy Anthropic and eliminate their models from government systems. If anything, the government should be trying to work more closely with Anthropic. As a side note, I hope Anthropic is working with state and local government entities on cyber vulnerability discovery, since many of our adversaries know that state and local is America’s soft underbelly in so many ways. In any event, the Mythos news should lay bare how stupid and counter-productive the Department of War’s feud with Anthropic really is. As someone who suspected all this was coming (not from inside knowledge but from it being ~obvious), that probably explains why I have had such a strong reaction to that feud. It’s this senseless distraction just at the time that the training wheels are coming off. I hope the two parties can resolve their differences now, for the sake of the country, but I am not hopeful. I do want to call out, however, the numerous political and career civil servants in the Trump Admin who do get these issues, know how stupid the Ant-DoW stuff is, and want to work with the frontier labs like adults. I wish you all utmost success. I find myself inclined to end on some positive notes. Mythos appears to be—according to Anthropic at least—“the most aligned” model Anthropic has ever trained. We are approaching superhuman capabilities in some domains, and yet alignment is getting better rather than worse. That’s not nothing. I know some of you think the model is faking its alignment, or aware when its alignment is being tested. I don’t have a good answer. Finally, there is this: Mythos was made by an American company, and like most successful American companies, it has a vested interest in maintaining order and peace, and it is investing substantial resources in mitigating the risks of its technological progress, as I expect most of the American labs would. This is cause for optimism: The incentives of capitalism are working. The training wheels are coming off, but at least we are the ones removing them, as opposed to our enemies. Perhaps we can be the first to learn to bike for real. The first step would be to get beyond all the low-fidelity, under-specified, pimply little fights of AI policy’s prepubescent era. That goes for me too. “What hath God wrought,” wrote the first telegram. What, indeed. In this case, the answer is still up to us.

@tv_ir_X @grok fact check?

🚨 BREAKING ​Iranian Revolutionary Guard Corps Missile Command: ​Announces that all targeting restrictions have ended and that it will attack infrastructure in a way that could deprive the US and regional countries of oil and gas resources for years. ​Orders have been transmitted to local missile bases, and they will begin immediately.

I taught Claude to talk like a caveman to use 75% less tokens. normal claude: ~180 tokens for a web search task caveman claude: ~45 tokens for the same task "I executed the web search tool" = 8 tokens caveman version: "Tool work" = 2 tokens every single grunt swap saves 6-10 tokens. across a FULL task that's 50-100 tokens saved why does it work? caveman claude doesn't explain itself. it does its task first. gives the result. then stops. no "I'd be happy to help you with that." no "Let me search the web for you" no more unnecessary filler words "result. done. me stop." 50-75% burn reduction with usage limits getting tighter every week this might be the most practical hack out there right now