X

Thinking of doing a daily blog to analyse the singularity Should I do it? If so here, substack or something else? Anything you'd like me to write about?

"We still need humans at the moment but quite plausibly within a year we'll no longer need humans for doing research in many areas." Jacob Coxon (ex Anthropic+Open AI researcher who resigned) on CNN, talking about Recursive self-improvement. CNN: "When did your feelings about this technology change? Was it something about the rate of progress that you've seen?" Jacob Coxon; "So I think it's really important to just look at the rate of progress and say, for example, in areas like coding or math, a couple years ago these AIs were just about helping humans a little bit. They could give you suggestions. Now they're close to replacing them. We still need humans at the moment, but quite plausibly, within a year, we'll no longer need humans for doing research in many areas. Only on Tuesday, OpenAI solved a millennium problem, one of the biggest unsolved open problems in mathematics, purely autonomously using an AI. And I think what's most scary is if AI is used to make itself more intelligent. So you can take an AI and give it the problem of AI research, just like currently they've given it math research. You could make it do the tasks that we're currently doing. And then you get what's called an intelligence explosion. The AI just gets smarter and smarter with no human involvement necessary, until you have a thing that is vastly smarter than humans. And this could be coming very soon." ---- From "CNN" YouTube channel, (full video link in comment)

做量化、炒币、盯美股的兄弟们，金融界的 Codex 今天真落地了！！ 这可能是今年把散户投研门槛干得最碎的一次，没有之一！！！ 你敲一句中文大白话： →左边有4 个 Agent 当场开圆桌会，看多的、挑刺的、风控、决策委员会分开干活互相撕； →右边同步写代码、拉盘口、跑回测； →而真金白银扣出去那一下，必须你亲手点确认。 它叫 Minara Harness，刚拿了 Circle Ventures 的投资，口号就一句话：自己运营你自己的华尔街。 我刚才直接开它的「机构圆桌」做了一场真实压力测试，咱们看下方视频： 我只敲了一句话：“AAPL 现在该买、持有还是卖？把整套逻辑讲给我听。” 指令一敲，后台立马拉响 7 阶段投研流水线！ 中间 4 个专业 Agent 瞬间扑上去并行开工： 基本面分析师去扒财务底稿、情绪分析师去扫 Polymarket 预测、宏观分析师盯新闻政策、技术分析师拉盘口指标——谁在采哪份物证、谁在复核数据，屏幕上全部透明实时跳动！ 经过观点对辩、风控复审、投委会闭门收敛，它没有像野鸡 AI 那样无脑喊冲，而是给了一份冷静到可怕的顶级机构级结论： “Hold（持有）：不追买，也不急于卖出。” 前瞻 PE 飙到 34 倍严重透支安全边际， 317 到 330 箱体上沿追高盈亏比极差， 模型置信度直接给你亮出 0.55， 甚至连毛利率口径冲突都抓出来示警！ 更绝的是，跑完当场甩出一份 317KB 的交互式 HTML 报告和 PDF，连建仓纪律、关键硬损位都切得清清楚楚。 以前机构养一个分析师团队加风控委员会才有的架势， 现在敲敲键盘，18 分钟在眼前全自动跑完交付！ 大家也别把它跟那些 AI 荐股脚本混为一谈， 刚才你看到的实测，只是它最能打的三刀之一： 1️⃣ 机构投研圆桌：一个标的丢进去，多空风控各执一词，谁依据什么数据、在哪一步吵翻，全程摊开。通用大模型最坑的就是没实时盘口、没你的持仓，还能一本正经编行情；这玩意直接接几十个数据源，研究覆盖的市场比它能下单的市场大得多； 2️⃣ 一句话变策略 IDE：用中文说一句做多半导体的趋势策略，当场出代码、跑回测，净值曲线、回撤、夏普全拉出来，改一句话就出一个新版。然后顺序替你焊死：真行情假钱纸交易 ➔ 小资金加回撤上限逐笔确认 ➔ 最后才轮到要不要自动跑，绝不让你一步冲动上实盘； 3️⃣ 决策下单同屏：K 线、盘口、拟执行订单排在同一屏，报价在动绝不追单，仓位、滑点、敞口都有硬上限。没有你的最终手动确认，一分钱都出不去。 做过交易的都懂以前有多碎： 看盘在 TradingView，研报在 ChatGPT， 算账在 Excel，回测在 Jupyter， 真下单还要切去交易所。 每切一次窗口，假设和数据就断一次， 现在全钉进同一个工作区，模型还能自选，OpenAI、Anthropic、Kimi、Grok、本地 Ollama 都能接。 AI 金融这行都在比拼谁更敢替你扣扳机， 它反过来，把扳机牢牢焊在你手里。 但先把丑话说在前头，这玩意是趁手的兵器， 绝不是印钞机这东西是工具，不是印钞机： 策略市场那张最唬人的卡片，回测涨了 619%、夏普 4.38，同一个策略近 90 天真跑只有 24.8%。 它把两条曲线分开画给你看，本身就是警告： 回测是写给历史的，证明不了未来； 咱们再把交易底层看清楚： 它能下单的不是美股真股票， 是 Hyperliquid 和 Lighter 上的 USDC 链上永续。 代币化的英伟达、苹果没有分红、没有投票权、有资金费、还会被爆仓。 想要 Schwab 那种真实股权和期权的这里给不了， 市场风险、滑点、过拟合、你自己的贪念， 它一个都消除不了。 第一周记住五个字：先用别打钱。 先跑一份能逐条核对出处的财报审查，再开圆桌逼它找看空逻辑，最后写一条策略只跑纸交易盯回撤。 等代码、逻辑、费用假设你自己都盯明白，再谈碰真钱。 说穿了：投研队伍可以是 AI 的，但扣扳机的手必须是你自己的。 大家觉得金融交易里，AI 最终是该全权托管自动跑， 还是必须把最后那道确认权死死扣在人手里？ 入口、定价、实测避坑和 3 条可直接抄的 Prompt 放评论区了👇

https://t.co/yAjiQkjJ7d

A recent paper showed AI self-replication is no longer technically hypothetical. Current agents can hack, copy themselves to a new server, and keep going, so this capability now belongs in real AI security planning. The paper actually demonstrates the full loop rather than testing isolated pieces. Agents found vulnerabilities, extracted credentials, transferred model weights and the agent software, started inference on the compromised machine, and then used the new replica to attack the next target. Qwen3.6-27B completed a 3-hop chain across 4 VMs on 3 continents from 1 initial prompt, with no human intervention after launch. But this does not mean AI agents are freely spreading across the internet, here in this experiment, the targets were intentionally vulnerable lab systems with known hostnames, available GPUs, and unusually accessible SSH credentials.

Jacob Coxon's next interview on CBS News (ex Anthropic+Open AI researcher who resigned) "We can't just unplug it because it could be copying itself over to other computers. Like it's not that difficult to find yourself because an AI is just code. It could transfer itself over the internet to a different place and then you unplug it here, but it's actually still over there and maybe it makes 10,000 copies of itself and they're all cooperating." ---- From "CBS News" YouTube channel, (full video link in comment)

GPT-6 Astra can now generate animatable Gaussian Splat characters. We connected it to the PINOC MCP and asked for a backrooms-style, Exit 8-ish game. We described the character we wanted and the motions. [MCP link in the comment 👇] Astra generated the character and every motion through PINOC through free preset animations and text to animation, and wrote the loop and the anomaly logic itself, and shipped the whole thing in a few sessions.