“The case for AGI by 2030” by Benjamin Todd

In recent months, the CEOs of leading AI companies have grown increasingly confident about rapid progress:

OpenAI’s Sam Altman: Shifted from saying in November “the rate of progress continues” to declaring in January “we are now confident we know how to build AGI”

Anthropic’s Dario Amodei: Stated in January “I’m more confident than I’ve ever been that we’re close to powerful capabilities… in the next 2-3 years”

Google DeepMind’s Demis Hassabis: Changed from “as soon as 10 years” in autumn to “probably three to five years away” by January.

What explains the shift? Is it just hype? Or could we really have Artificial General Intelligence (AGI)There’s no single point at which a system becomes ‘AGI,’ and the term gets used in many different ways.

More fundamentally, we can classify AI systems based on the (i) strength and (ii) breadth of their capabilities.

‘Narrow’ AI demonstrates strong performance at a small range of tasks (e.g. chess-playing AI). Most technologies have very narrow applications.

‘General’ AI is supposed to have strong capabilities in a wide range of domains, in the same way that humans can learn to do a wide range of jobs. But there’s no single point at which narrow becomes general – it’s just a spectrum.

Narrated by the author.

---

Outline:

(00:00) Introduction

(00:33) The case for AGI by 2030

(04:04) The article in a nutshell

(05:46) Section One: what''s driven recent AI Progress?

(05:52) How we got here: the deep learning era

(07:45) Where are we now: the four key drivers

(08:57) 1: Scaling pretraining

(12:14) Algorithmic efficiency

(14:22) How much further can pretraining scale?

(16:15) 2: Training the models to reason

(22:06) How far can scaling reasoning continue?

(25:01) 3. Increasing how long medels think

(28:00) 4: Building better agents

(33:38) How far can agent improvements continue?

(35:57) Section Two: how good will AI become by 2030?

(37:40) Trend extrapolation of AI capabilities

(39:56) What jobs would these systems help with?

(40:48) Software Engineering

(42:10) Scientific Research

(43:19) AI research

(44:28) What''s the case against this?

(49:16) Additional resources on the skeptical view

(49:48) When do the ''experts'' expect AGI?

(51:04) Section Three: why the next 5 years are crucial

(52:08) Bottlenecks around 2030

(56:00) Two potential futures for AI

(58:02) Conclusion

(59:24) Thanks for listening

---

First published:
March 20th, 2025

Source:
https://80000hours.org/ai/guide/when-will-agi-arrive