Google says AGI by 2030

👋 Good morning/evening (wherever you are). It’s Wednesday.

Yes, AGI is coming. What should we do? Who will get there first? 

Google DeepMind said this:

Read more: Taking a responsible path to AGI

Also, Epoch AI evaluated Gemini 2.5 Pro on GPQA Diamond and found a score of 84%, exactly matching the result reported by Google.

Gemini 2.5 Pro is very good. And it’s free.

Note: There’s a rumor that Google is releasing a new SOTA model that’s even better than 2.5 Pro (Code name: nightwhisper).

OK let’s keep going ↓

Here’s what you should know:

• Introducing Claude for Education

  A specialized version of Claude tailored for higher education institutions

• OpenAI Academy

  Launched last year in September, OpenAI Academy has now expanded to offer a variety of free resources, including video tutorials, courses, and workshops.

• OpenAI Introduces PaperBench for AI Agent Evaluation

  A benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research

• Large Language Models Pass the Turing Test

  People had a five minute, three-way conversation with another person & an AI. They picked GPT-4.5 (prompted to act human) as the “real” person 73% of time.

• AI masters Minecraft: DeepMind program finds diamonds without being taught

  The Dreamer system reached the milestone by ‘imagining’ the future impact of possible decisions.

• Windsurf Wave 6 is here

  Quick app deployment update

• Nvidia Blackwell Sets AI Performance Records In Benchmark Tests

  The latest MLPerf Inference V5.0 benchmarks

• VidTok introduces compact, efficient tokenization to enhance AI video processing

  The Microsoft Research Team discusses a method that converts video data into smaller, structured units, or tokens.

From Midjourney’s 4/2 office hours (hosted on X), David mentioned the following:

“[Years later] you’ll be able to tell if an image is made by AI because it will look too good”

“2 years later, you’ll look at Midjourney v7 and [think v7 was bad]”

When asked about why they are not doing API:

“In the short term, we don’t want hyper commercialization…when we allow people to build worlds/games…then we’ll have API”

“We just want people to have fun”

David also mentioned personal/private API as a possibility and said that “in some instances, enterprise API could be good”

He also reiterated that v7 “will not have character reference, but in the future will have an omni-reference feature”

FYI, omni-reference, video expansion, 3D innovations, and even hardware have been mentioned in the past (months ago).

The numbers:

• AI coding startup Replit is in talks to raise $200M at a $3B valuation

• AttoTude secures $50M Series B Funding to accelerate deployment of THz interconnect technology

• Chef Robotics raises $43M Series A to scale AI-enabled robotics in meal assembly

  LTV, raised a $5.2M Series A for its AI-driven email and text messaging product

• UltraGreen, AI for surgical applications, is considering raising up to $300M IPO in Singapore

• Actively AI raises $22.5M to offer sales ‘superintelligence,’ says AI SDRs failed

Thought starters:

David Sacks asked this question on X…is Elon right?

Meme of the day:

Thanks for reading,

Eddie

Are we friends on LinkedIn and X? Let’s connect — DM me or hit the follow button.

P.S. If this was valuable, forward it to a friend. If you’re that smart friend, subscribe here.

P.P.S. Interested in reaching other ambitious readers like you? To become a sponsor, reply to this email.