AI 2027 is happening

PLUS: Build a self-improving agent system with Fable 5

Author

Barsee
11 Nov

Together with

howdy, it’s Barsee again.

happy wednesday, AI family, and welcome back to AI Valley.

here are the biggest things worth knowing today:

  • The AI 2027 researchers now think they were too conservative

  • The Nvidia scandal makes no sense

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

ADOBE

Courtesy: Adobe

Adobe Firefly, the all-in-one AI creative studio, is expanding the Adobe Creative Agent through new capabilities in Firefly AI Assistant (beta). It carries out multi-step tasks using Adobe Creative Cloud tools, including Premiere, Illustrator, InDesign, After Effects and Frame.io, all guided by your direction. Describe what you want to make in your own words, then refine, redirect, or take over at any time.

Adobe is also previewing Adobe Creative Cloud MCP, a new way for creative professionals to direct third-party agents as they make changes directly inside native project files in Photoshop, Illustrator, Premiere and After Effects, while staying in control.

It's part of Adobe's approach to making pro-grade creative workflows easier to start, control and continue, wherever inspiration strikes.

*This is sponsored

THROUGH THE VALLEY

Courtesy: AI 2027 Tracker

The AI 2027 researchers now think they were too conservative. All six authors of the famous AI 2027 scenario now expect a “superhuman AI researcher” sometime in 2027: AI as good as the very best human researchers, to the point where a lab would rather replace its entire research team than give up the AI. Daniel Kokotajlo now gives it a 50-50 chance by May 2027, pulled forward from December 2028 just three months ago. His own summary: “I am scared, sad, and angry.”

The reason is a new METR assessment of Anthropic. METR spent 11 weeks inside the company and estimates its AI progress is now running about twice its historical pace, with perhaps 80% of that speed-up coming from AI itself. The AI 2027 team says that puts us about five months ahead of their original scenario. Three months ago, even their most aggressive model gave less than a 4% chance of that happening by now.

If that sounds familiar, it’s because Anthropic warned in its IPO filing that AI might soon dramatically speed up its own research. That’s recursive self-improvement, or RSI: AI building better AI, which builds better AI still. METR’s report is the first serious outside evidence that the loop may already have started.

And this isn’t just engineers working faster. Anthropic staff self-report being 12x more productive with their latest internal model. METR’s finding is that the whole company’s progress has roughly doubled, which is a much bigger deal.

The forecasts that follow are wild: Their model suggests 2027 alone could pack in something like five years’ worth of historical AI progress. They put a 35–65% chance on AI that beats top human experts at virtually all white-collar work arriving before 2028. And extrapolating Anthropic’s own automation data, they get 95% of its R&D tasks fully automated by November 2027. That won’t stay an Anthropic story. Whatever happens there will eventually happen at every lab.

One big caveat: Anthropic redacted most of the evidence METR relied on, and this is one company, so treat these numbers as a strong signal, not proof.

Still, what gets me is that the people whose literal job is forecasting this are the ones most surprised by it. If the pace is hard to take in for them, the rest of us probably aren’t taking it in at all.

TRENDING TOOLS

  • PaperTrades - watch Claude, Gemini and ChatGPT race with pretend money at real prices. Every trade, reason and grade is public. See who wins on Nov 28 (sponsored)

  • Claude Whiteboards, Claude Sheets and Claude Data - think a plan through on a board, build a tracker in a sheet, and tell a story with your data

  • Businesses in ChatGPT - claim your business, manage your information, and get insights on how people engage with you

  • Google Gemini 4 Carbon - its strongest model, built for long agent jobs like coding, research, and security work

  • terminal-mac - entire MacOS that runs inside your existing terminal. Computer use built in: MCP, CLI and skill. Free & open source

WHAT I'M CONSUMING

THE VALLEY GEMS

What’s trending on social today:

THAT’S ALL FOR TODAY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.