scripod.com

How DeepMind Plans to Reach AGI Safely

Shownote

DeepMind’s co‑founder and CEO Demis Hassabis lays out a pragmatic path to AGI—why he still sees a 5–10 year window, what current models are missing (reasoning, planning, continual learning), and why world models will likely converge with today’s LLMs. He e...

Highlights

In this episode of 'The Tech Download,' the conversation dives into the cutting-edge developments at Google DeepMind, exploring how artificial intelligence is reshaping not only technology but also the future of science and society. With AI advancing at a breakneck pace, the focus shifts from raw computational power to the deeper challenges of reasoning, control, and real-world integration.
05:38
Current chatbots lack continuous learning and true originality.
13:55
AI could help discover room-temperature superconductor materials
21:55
LLMs alone won't achieve AGI; world models are essential for understanding the physical world.
30:04
Gemini series puts Google DeepMind at the top of the AI leaderboards.
45:25
AlphaFold could lead to major drug breakthroughs through AI-driven protein design
48:20
Apple will use Google's Gemini to power the new Siri.

Chapters

How did Google DeepMind become the engine of Google's AI revolution?
00:00
Are we hitting the limits of AI scaling—and what comes next?
05:38
What will it take to build AGI in the next decade?
11:12
Can we control superintelligent AI before it controls us?
16:34
How did Google go from slow starter to AI frontrunner?
27:14
Who’s winning the AI race: U.S. or China?
32:40
Why 2026 could be the year AI goes truly mainstream?
48:20

Transcript

Steve Kovach: A CNBC original podcast. Arjun Kharpal: Hello and welcome to The Tech Download, a new CNBC original podcast where we unpack the tech stories that matter most. Steve Kovach: Each season, we dive into one big theme and what it means for your ...