What Dave Blundin has said about AI alignment

Blundin argues that only AI can supervise AI at scale and that transparency into model activations makes alignment a tractable problem.

“nothing's going to ever be able to police AI other than other AI”

Moonshots with Peter Diamandis: Ask the Mates Anything Round #2 | MOONSHOTS AMA #293 (Sep 22, 2026) · Listen from 5:39

Blundin contends that alignment is less difficult than commonly portrayed if one can inspect what a model is thinking.

“if you know what it's thinking, it's actually very straightforward to make sure it's having nothing but humanly beneficial thoughts”

Moonshots with Peter Diamandis: Ask the Mates Anything Round #2 | MOONSHOTS AMA #293 (Sep 22, 2026) · Listen from 5:13

Blundin argues that AI productivity gains will flow to shareholders rather than wages, so workers should seek equity ownership.

“Almost all value is going to accrete to capital gains through ownership and equity and not to payroll”

Moonshots with Peter Diamandis: Ask the Mates Anything Round #2 | MOONSHOTS AMA #293 (Sep 22, 2026) · Listen from 30:45

Blundin argues a recursively self-improving AI could revise its trained values and that interpretability monitoring to halt such changes is technologically easy but hard to enforce.

“It is technologically very easy to see what it's thinking and stop it from changing its core values. It's just a question of enforcing it”

Moonshots with Peter Diamandis: Should we slow down AI progress? | MOONSHOTS #288 (Sep 11, 2026) · Listen from 2:38:39