Google Claims Genie 3 is a Step Towards AGI, AI Bubbles, BigAI and Media Citations

Did Generative AI hack our attention? A Media citation report is telling.

Photo: Nathan Laine/Bloomberg/Getty Images

Hey Everyone,

I wanted to share with you some things on my mind in the AI news cycle and generally watching the space in late Summer, 2025.

Stoking an AI Bubble

OpenAI CEO Sam Altman told reporters at a private dinner recently that investors are over excited about AI models. "Someone" will lose a "phenomenal amount of money.” I cannot help but wonder if he is referring to himself, but odd to hear BigAI talking about an AI Bubble.

An MIT report that 95% of AI pilots fail spooked investors.

The study by MIT also brought the AI bubble topic into focus as the stock market finally cooled off after Trump’s “Liberation Day” event in early April. The stock market has been booming in the four months stretch from Spring into Summer of 2025. The NASDAQ 100, the proxy for Tech stocks is up another 10% year to date. Odd, it looks like MIT News was forced to take down the report that was already bizarrely gated by NAND.

Google DeepMind Claims Genie 3 is a Step Towards Embodied AI

Several companies released new WFMs in the last six weeks, but Google was the only one to claim it’s a step towards AGI.

• “Genie 3 is the first real-time interactive general-purpose world model,” Shlomi Fruchter said.

Genie 3 can generate multiple minutes of interactive 3D environments at 720p resolution at 24 frames per second — a significant jump from the 10 to 20 seconds Genie 2 could produce. But it’s no precursor to AGI by any stretch of the imagination. Although if Google DeepMind researchers are making such claims, what does it actually even mean? 🤔

“We think world models are key on the path to AGI, specifically for embodied agents, where simulating real world scenarios is particularly challenging,” Jack Parker-Holder, a research scientist on DeepMind’s open-endedness team

World models have next to nothing to do with even commercial AGI in practice. So it’s not clear what Google DeepMind’s PR team (known for blatant exaggerations) are up to.

OpenAI have gamed Media Citations to make ChatGPT “Go Viral”

The American internet is starting to resemble the Chinese internet, yet instead of having the CCP on top, we have BigTech and BigAI companies in charge.

This article goes into some depth, with infographics and easy to understand sections, opinions and insights around the Big picture of AI and its impact on capitalism.

AI Supremacy also has a new community, come and check it out (invite link) if you want to nerd out on this stuff.

Microsoft B2C CEO of AI

Microsoft’s CEO of AI, Mustafa Suleyman, published a blog post on Tuesday arguing that the study of AI welfare is “both premature, and frankly dangerous.” The essay discusses "Seemingly Conscious AI" (SCAI), which mimics consciousness but isn't actually conscious. This from a company in Microsoft, that funded a competitor and has executed extremely poorly on models for its own products (e.g. Copilot). In a rather confusing statement, he emphasizes that while true AI consciousness remains speculative and unsupported by evidence, the illusion of it is dangerous and avoidable.

Joseph Politano.

Datacenters as the New Office Construction

The US spends almost as much building data centres as it does building offices. 2025 might be remembered for that moment when compute demands began to outstrip civilization’s demand for office spaces as measured by construction spending.

• The office was about concentrating human beings.

• What is the datacenter exactly?

• Data centres are about concentrating the computation needed for AI, which could reduce the need for more labour in the future in capex intensive self-perpetuating cycles that will demand more compute and energy indefinitely.