AI optimists envision a future where artificial general intelligence (AGI) surpasses human intelligence, but the path remains riddled with scientific and logistical hurdles.
Identifying vulnerabilities is good for public safety, industry, and the scientists making these models.