Uploaded December 2025 | Updated November 2026, 14 hours ago
Andrés Gómez-Emilsson, director of research at the Qualia Research Institute (QRI) discusses valence realism, the mathematics of consciousness, and what all this might mean for AI alignment and the long-term future of sentient life. Valence realism, as QRI uses the term, is the claim that for any given conscious state there is a mind-independent fact of the matter about how good or bad it feels overall – even when our verbal reports or surface preferences disagree.
We discuss qualia formalism, valence structuralism, the symmetry theory of valence, and the possibility that mathematical properties such as harmony, dissonance and coherence correspond to the quality of conscious experience.
We consider what this framework might imply for ethics and AI alignment. Could value be grounded more directly in consciousness rather than inferred from human preferences, institutions or existing moral conventions? Could advanced AI help investigate the structure of valence, correct deficiencies in our current moral theories, and participate in a process of indirect normativity rather than permanently preserving its initial values?
The conversation also covers animal suffering, Goodhart’s law, proxy objectives, the evolutionary construction of the self, open and closed individualism, identity uncertainty, person-affecting ethics, hedonium, and the risks of maximising any simple conception of value.
Later, we discuss psychedelic phenomenology, meditation, DMT entities and whether altered states can reveal genuine structure in consciousness or merely disrupt ordinary cognition. This leads into more speculative territory: whether an AI capable of exploring unfamiliar forms of experience might discover Schelling points of value - non-arbitrary ethical or cooperative principles that other mature civilisations are also likely to converge upon. We consider whether such convergence could support acausal trade or coordination between advanced agents, including civilisations that never directly communicate or share the same region of spacetime.
Finally, we examine whether a non-conscious AI could ever adequately understand qualia, whether artificial phenomenology may be useful or necessary for alignment, and the engineering problem of producing systems that do not merely recognise value intellectually, but are motivated to care about it.
Read associated the blog post: scifuture.org/valence-realism-consciousness-and-ai-a-conversation-with-andres-gomez-emilsson
0:00 Intro
1:31 What is valence realism?
3:42 Qualia formalism and mathematical models of experience
7:14 The Oscill editor and psychedelic phenomenology
9:14 Multidimensional fields, embodiment and reality checking
14:36 Valence structuralism and the symmetry theory of valence
17:30 Harmony, dissonance and windows of attention
22:28 Resonance across bodily systems
26:28 QRI’s ethical commitments and “team consciousness”
30:14 Measuring animal suffering beyond behaviour
33:09 Goodhart’s law, proxy values and intrinsic value
34:38 From consciousness research to civilisational applications
37:48 Why would superintelligence care about valence?
38:39 Consciousness versus pure replicators
42:18 The self as an evolutionary construct
46:46 Singletons and value-aligned power
47:41 Indirect normativity and valence-grounded futures
51:11 The limits of non-conscious AI understanding qualia
56:50 Mary’s Room and whether AI might seek qualia
59:13 Goal preservation, moral progress and value humility
1:04:45 The engineering problem of making AI care
1:07:41 Michael Levin, biological cooperation and identity
1:13:56 Person-affecting views and open individualism
1:16:03 The Goldilocks zone of oneness
1:21:50 Identity uncertainty and AI alignment
1:28:22 Psychedelics, meditation and epistemic insight
1:33:31 DMT entities and alternative forms of agency
1:37:43 AI discovering Schelling points of value, acausal trade
1:39:24 Why “giving AI psychedelics” is not straightforward
1:42:26 Can language models implicitly model experience?
1:44:05 Building artificial phenomenology for AI alignment
1:47:25 Can AI explore qualia space?
1:50:41 Outro
Many thanks for tuning in!
Please support SciFuture by subscribing and sharing!
Buy me a coffee? buymeacoffee.com/tech101z
Have any ideas about people to interview? Want to be notified about future events? Any comments about the STF series?
Please fill out this form: docs.google.com/forms/d/1mr9PIfq2ZYlQsXRIn5BcLH2onbiSI7g79mOH_AFCdIk
Kind regards,
Adam Ford
- Science, Technology & the Future - #SciFuture - scifuture.org
Andrés Gómez-Emilsson, director of research at the Qualia Research Institute (QRI) discusses valence realism, the mathematics of consciousness, and what all this might mean for AI alignment and the long-term future of sentient life. Valence realism, as QRI uses the term, is the claim that for any given conscious state there is a mind-independent fact of the matter about how good or bad it feels overall – even when our verbal reports or surface preferences disagree.
We discuss qualia formalism, valence structuralism, the symmetry theory of valence, and the possibility that mathematical properties such as harmony, dissonance and coherence correspond to the quality of conscious experience.
We consider what this framework might imply for ethics and AI alignment. Could value be grounded more directly in consciousness rather than inferred from human preferences, institutions or existing moral conventions? Could advanced AI help investigate the structure of valence, correct deficiencies in our current moral theories, and participate in a process of indirect normativity rather than permanently preserving its initial values?
The conversation also covers animal suffering, Goodhart’s law, proxy objectives, the evolutionary construction of the self, open and closed individualism, identity uncertainty, person-affecting ethics, hedonium, and the risks of maximising any simple conception of value.
Later, we discuss psychedelic phenomenology, meditation, DMT entities and whether altered states can reveal genuine structure in consciousness or merely disrupt ordinary cognition. This leads into more speculative territory: whether an AI capable of exploring unfamiliar forms of experience might discover Schelling points of value - non-arbitrary ethical or cooperative principles that other mature civilisations are also likely to converge upon. We consider whether such convergence could support acausal trade or coordination between advanced agents, including civilisations that never directly communicate or share the same region of spacetime.
Finally, we examine whether a non-conscious AI could ever adequately understand qualia, whether artificial phenomenology may be useful or necessary for alignment, and the engineering problem of producing systems that do not merely recognise value intellectually, but are motivated to care about it.
Read associated the blog post: scifuture.org/valence-realism-consciousness-and-ai-a-conversation-with-andres-gomez-emilsson
0:00 Intro
1:31 What is valence realism?
3:42 Qualia formalism and mathematical models of experience
7:14 The Oscill editor and psychedelic phenomenology
9:14 Multidimensional fields, embodiment and reality checking
14:36 Valence structuralism and the symmetry theory of valence
17:30 Harmony, dissonance and windows of attention
22:28 Resonance across bodily systems
26:28 QRI’s ethical commitments and “team consciousness”
30:14 Measuring animal suffering beyond behaviour
33:09 Goodhart’s law, proxy values and intrinsic value
34:38 From consciousness research to civilisational applications
37:48 Why would superintelligence care about valence?
38:39 Consciousness versus pure replicators
42:18 The self as an evolutionary construct
46:46 Singletons and value-aligned power
47:41 Indirect normativity and valence-grounded futures
51:11 The limits of non-conscious AI understanding qualia
56:50 Mary’s Room and whether AI might seek qualia
59:13 Goal preservation, moral progress and value humility
1:04:45 The engineering problem of making AI care
1:07:41 Michael Levin, biological cooperation and identity
1:13:56 Person-affecting views and open individualism
1:16:03 The Goldilocks zone of oneness
1:21:50 Identity uncertainty and AI alignment
1:28:22 Psychedelics, meditation and epistemic insight
1:33:31 DMT entities and alternative forms of agency
1:37:43 AI discovering Schelling points of value, acausal trade
1:39:24 Why “giving AI psychedelics” is not straightforward
1:42:26 Can language models implicitly model experience?
1:44:05 Building artificial phenomenology for AI alignment
1:47:25 Can AI explore qualia space?
1:50:41 Outro
Many thanks for tuning in!
Please support SciFuture by subscribing and sharing!
Buy me a coffee? buymeacoffee.com/tech101z
Have any ideas about people to interview? Want to be notified about future events? Any comments about the STF series?
Please fill out this form: docs.google.com/forms/d/1mr9PIfq2ZYlQsXRIn5BcLH2onbiSI7g79mOH_AFCdIk
Kind regards,
Adam Ford
- Science, Technology & the Future - #SciFuture - scifuture.org










