Why Control Alignment Fails

Superintelligent AI is rapidly approaching, and humanity (specifically, the labs building it) is still a long way from proving it can be safely controlled. But what if that’s the wrong way of viewing the problem? Is there a way to instill values to the mutual benefit of both humans and machines?

In this episode of the Win-Win Podcast, Liv sits down with Emmett Shear — Twitch co-founder, former interim CEO of OpenAI and founder of Softmax — for a groundbreaking conversation on the future of AI alignment.

Emmett unpacks why the traditional “control alignment” approach, which tries to hard-code pre-determined values into AI, is destined to fail — and why “organic alignment,” inspired by nature, teamwork, and human societies, might be the only path forward for coexistence. From multicellularity to markets, he explains how collective identities form; why betrayal can sometimes be the moral choice; and how we might go about teaching AI to care for us as its own family, rather than see us as masters to obey (or perhaps more accurately, ants to squash).

Hear how Emmett's organisation Softmax is experimenting with small, goal-oriented agents in virtual worlds. Why superintelligence should emerge as societies of diverse AIs instead of a single runaway system. And how our own capacity for alignment as groups of humans may be the most powerful clue for what to do next.

If you’re ready to explore the very foundations of your assumptions around technology, philosophy and the nature of intelligence itself, this episode will be right up your alley...

Chapters:

00:48 - Intro.

01:04 - Organic Alignment.7

09:33 - The Formation Of Emergent Coordination.

16:01 - Techniques For Creating Alignment.

23:43 - What Is The Ideal Environment To Generate Alignment

30:09 - How Much Influence Is Projected On The Ai?

32:49 - The Global Super Organism That Is The Market.

40:41 - Group Questions.

40:50 - The Difference Between Coordination And Alignment.

43:02 - What Is The Win Condition For Softmax?

50:17 - How Do We Maintain The Need For Humans In The Loop?

58:57 - Selecting And Scaling The Right Alignment Model.

1:02:09 - Blowtorch Theory & Black Holes

1:04:48 - Agents Aligning With Each Other Isn't The Same As Aligning With Humans.

1:07:25 - Exploring Ai's Capability For Self Modification And Replication.

1:14:35 - The Difference Between Alignment Capacity And Aligning With Humanity.

1:18:38 - Does Emmett Believe In God?

1:25:23 - Are Humans The Appendix To The Future Super Intelligence?

1:32:17 - Is Training Ai On Human Ethics And Morality Actually A Good Idea?

1:36:17 - Will Ai Be Conscious Or Self Aware?

1:40:48 - One Axiom To Instill Into Ai's Being Created.

Links:

♾️Softmax - https://softmax.com/

♾️Nebulosity and Meaningness - https://meaningness.com/nebulosity♾️Black Holes and our evolving universe - https://theeggandtherock.com/p/in-which-i-tell-you-about-my-next

♾️AI Alignment Forum - https://www.alignmentforum.org/♾️Evolving Intrinsic Motivations for Altruistic Behaviour - https://arxiv.org/pdf/1811.05931

Credits:

♾️ Hosted by Liv Boeree

♾️ Produced by Luca de Vico

The Win-Win Podcast:

Poker champion Liv Boeree takes to the interview chair to tease apart the complexities of one of the most fundamental parts of human nature: competition. Liv is joined by top philosophers, gamers, artists, technologists, CEOs, scientists, politicians and more to understand how competition manifests in their world, and how to change seemingly win-lose systems into Win-Wins.Podcast

links:

♾️ Youtube: https://www.youtube.com/playlist?list=PLWgq0OZMtwtOIyMsVM_vksqdfWcM-b68S

♾️ Spotify: https://open.spotify.com/show/03bGVUaFZmJUmEvSHNDPdI?si=64379cc23696454f

♾️ Apple Podcasts: https://podcasts.apple.com/us/podcast/win-win-with-liv-boeree/id1724791350

♾️ Pocketcast: https://play.pocketcasts.com/podcasts/7f708340-d17c-013b-f46e-0acc26574db2

#winwinpodcast #organicalignment #aialignment #ai #aisafety