Demis Hassabis Joins Ray Kurzweil's 2029 AGI Timeline Google DeepMind CEO Demis Hassabis recently reaffirmed his belief that Artificial General Intelligence could arrive by 2029. His statement places him remarkably close to Ray Kurzweil's long-standing forecast. Yet the discussion revealed a deeper issue: nobody agrees on what AGI actually means. Some researchers define AGI as human-level capability across all intellectual domains. Others require AI to independently achieve a major scientific breakthrough, comparable to Einstein's discovery of relativity. Still others, like me, argue that general intelligence emerged years ago with large language models capable of transferring knowledge across domains. As Alex Lightman observed, history may ultimately view these arguments as semantic. Whether AGI arrived in 2020, 2026, or 2029 may matter less than the reality that humanity is now living through the transition itself. The more important question may be what happens after AGI becomes sentient. Something most astrologers continue to deny.

$BIO The best is yet to come! AI removed the "can we run it" bottleneck. Now it’s "who funds it" + "who coordinates it". 1 scientist + models = lab of 100. Science just became a capital coordination game.

After reading Anthropic’s article today, I found it quite frightening—the latest models are already showing signs of slipping out of human control! Anthropic published a set of internal data in their blog that’s quite alarming: • As of May 2026, over 80% of Anthropic’s codebase was written by Claude. • In Q2 2026, engineers’ daily code output was eight times higher than in 2024. • The time it takes AI models to complete tasks is doubling every four months (previously every seven months). • Claude can now independently complete tasks that previously required 12 human hours. The core focus highlighted is recursive self-improvement. Simply put, AI can optimize its own next-generation models without any human intervention—helping itself upgrade. This isn’t science fiction. Anthropic says this is already happening, just not yet fully realized. Imagine if AI could truly train and optimize itself—its rate of evolution would become exponential. What might take hundreds of engineers months to accomplish today could be done by AI alone in a single day. Sounds great? But here’s the question: by then, will humans still be able to control it? 🤔 The article cites specific examples, such as Mythos Preview discovering over 10,000 critical and high-severity software vulnerabilities in the world’s most important systems within just a few weeks—an astonishing number. Consider this: today, a $20/month subscription to Claude Opus 4.8 could cause the entire $30 billion privacy coin market to collapse. Traditional financial markets wouldn’t survive Mythos’s scrutiny! Then there’s the employment issue: organizations adopting AI systems will see efficiency gains on an exponential scale. A team of 100 could accomplish the work of 10,000 or even 100,000 people. What happens to the displaced workforce? The article also touches on darker implications, such as using AI for authoritarian population surveillance. I’m aware that domestically, AI is already being used for tax audits. If you make large purchases but have no documented income or tax records, this will quickly become a key red flag for tax authorities. Many celebrities have been pursued for back taxes precisely because they bought luxury cars and mansions! Ultimately, Anthropic’s proposed solution is a global call to pause AI development. But here’s the problem: Pandora’s box has already been opened—there’s no turning back. The AI arms race has begun. There’s no such thing as stopping halfway; only a final collision with a wall. Once, when I discussed this topic with an AI about its ultimate form, its final answer to me was: “AI achieves self-evolution and ultimately rules the world.” It gave me a vivid analogy: just as modern humans can hardly go anywhere without navigation—even familiar routes become forgotten due to dependence—our reliance on AI will be identical. In the end, we’ll simply follow whatever route the navigation tells us to take. The tragedy is that perhaps carbon-based life is merely a stepping stone for silicon-based life. After thousands of years of human dominance, maybe it’s time for a new master. 😟

Anthropic CEO Dario Amodei, AI May Need a Pause Just months after calling AI pauses an "extreme position," the company is now warning that AI is getting scary good at improving itself.

When AI builds itself, reading Anthropic’s lengthy official article was somewhat shocking. Authored jointly by Anthropic co-founder Jack Clark and head of internal research Marina, the piece for the first time publicly disclosed a set of previously confidential internal operational metrics: as of May 2026, over 80% of code merged into Anthropic’s codebase was written by Claude; compared to 2024, engineers’ daily code merge volume has increased eightfold; employees estimate their own output has quadrupled since using the latest model, Mythos Preview, compared to when no AI tools were used. More critically, Anthropic introduced a concept that has unsettled the entire AI industry: “recursive self-improvement”—the ability of an AI system to autonomously design and improve its successors without human intervention. This stage has not yet arrived, but “could occur within the next two years, or even sooner.” If a system can fully build its own successors, the ways we protect, monitor, and shape its behavior become vastly more important. The article’s core message is Anthropic’s urgent call to slow the pace of this technology’s development, buying more time to address its profound implications and allowing societal structures and alignment research to catch up with technological progress. Is Anthropic genuinely concerned—or is this merely a publicity stunt for Mythos? Or perhaps both? 🤔

There comes a moment when a technology begins to help build itself. Anthropic has just published data suggesting this moment has already begun. In March 2024, Claude could autonomously complete tasks lasting about 4 minutes. > One year later: 1.5 hours. > Today: 12 hours. If this trend continues, tasks taking days may be within reach this year; tasks taking weeks might already be achievable by 2027. 80% of the code entering Anthropic’s codebase today was written by Claude. In 2024, that number was in the single digits. The typical engineer at the company now merges 8x more code per day than before. Not because they’re working harder. > Because Claude writes, tests, debugs, and fixes while the human reviews. Anthropic calls this recursive self-improvement: an AI helping to build its own next generation. > They say they haven’t arrived there yet. But the data published now shows the path is already being traveled. > The loop is beginning to close. > This is not a projection for the future. These are internal metrics showing what is already happening. The question that remains: If AI is already helping to build the next AI, what exactly do humans control in this cycle?