Julia Bossmann
Research AI consciousness, sentience, and welfare
CERC AAI and others
Bio
- Neuroscience in Damasio lab at USC (consciousness research)
- AI governance, ethics, and alignment at Mila
- Entrepreneur in Residence (AI focus) at Singularity University
- Patent in knowledge engineering
- World Economic Forum Global Future Council on AI & Robotics
- Contributed to US Senate testimony on AI safety
- Taught about AI to world leaders in Davos and the to the public on a TV show
- Active connections across labs, safety orgs, and research
Mentee must-haves/nice-to-haves
You're excited to execute on a project and produce something. If you're unsure whether you're a good fit, default to applying anyway.
Mentee role
You're the adventurer. You can brainstorm the direction with me, get guidance and support when you're stuck, and co-author the output. If desired, we can have weekly co-working sessions, too.
❓ Sample mentee tasks
- Literature review and synthesis
- Developing and refining ideas, frameworks, and core arguments
- Designing, running, and analyzing experiments or surveys (if empirical)
- Drafting and revising written output
Mentor support
- Guidance
- Feedback
- (Co-)shaping direction
- Removing obstacles
- Connections to people in the space
- Co-authorship where appropriate
Questions for applicants
- Which project interests you most, and why? Or describe your own idea. (Max 300 words)
- What background do you bring, and what do you want to learn? (Max 200 words)
- What kind of public output do you want to create? Please share an example (incl. URL) of similar work you've completed, if any. (Max 200 words)
- What are 1-3 pieces of evidence that you'd be able to do well in this project? (These don't have to be standard credentials!) Please concisely describe them and why they're relevant. (Max 300 words)
- A model produces outputs that humans interpret as expressing "distress." How would you approach determining whether this reflects something welfare-relevant versus trained patterns? (Uncertainty is okay here; share how you would think about this question.) (Max 300 words)
Mentor-led project
Digital consciousness/sentience and welfare
All projects are intended to build on existing research in the field, and to result in publication where appropriate.
Click here for more information on each project (navigate using tabs on left).
1. Experimental: The model unhappiness tax
Investigate whether welfare-informed design produces measurable improvements in model behavior and reliability.
2. Empirical: “What Would Convince You?”
Ask experts and the public: what evidence would actually convince you an AI system is conscious or deserves moral consideration?
3. Position: a) Humanity's selfish case for consciousness research / b) Humanity's selfish case for model welfare
Make the strategic case that consciousness research / model welfare research is far from charity: it's required for humanity's future in a coming post-AGI world.
4. Explorative: The “proof of work” approach to consciousness
Are there outputs that constitute evidence of consciousness because computationally no other explanation suffices? Explore how expressions might serve as evidence of inner life.
**5. First principles: What's welfare, what's projection? **
What could a theory of wellbeing for any possible sentient system look like? How might we determine what's genuinely universal to complex intelligences?