The AI Safety Expert: These Are The Only 5 Jobs That Will Remain In 2030! - Dr. Roman Yampolskiy

He argues that any role requiring unique personal experience or creative human perspective may be rare and that most other tasks could be automated; however, niche human preferences might preserve some roles (e.g., certain traditional professions). Timestamp: 810.

He asserts a high confidence in simulation theory and explains that, even if true, the subjective experience of life and relations remains meaningful; the outside-world question matters mainly for curiosity and research. Timestamp: 3442.

No. He argues that superintelligent systems would be distributed, capable of backups, and likely to disable human intervention before we can turn them off. Timestamp: 1824.

He frames superintelligence as a meta-solution: if guided well, it could help with climate and wars; if mismanaged, it could be catastrophic, making the focus on safety crucial. Timestamp: 1787.

He suggests engaging with safety advocates and researchers, learning enough to question claims, and supporting peaceful, legal advocacy while awaiting pragmatic policy influence. Timestamp: 2830.

Narrow AI excels in specific domains; AGI can operate across many domains; superintelligence would surpass humans in all domains. Timestamp: 499.

A likely pathway discussed is the development and deployment of a highly capable biological tool (virus) enabled by AI, with post-deployment misuse or malicious design. Timestamp: 2281.

He emphasizes broad social planning, inclusion of basic income concepts, and exploring fields that leverage AI safely, while questioning retraining as a sole strategy since automation could span nearly all jobs. Timestamp: 3290.

He supports peaceful, legal protest and advocacy, but cautions that policy alone may be insufficient; broad public engagement is needed to shape incentives. Timestamp: 3249.

Yes—he notes longevity research as a major future focus and discusses the idea of longevity escape velocity as a means to maintain human relevance alongside AI progress. Timestamp: 4067.