ne of AI’s biggest proponents is calling attention to its risks — again.
In a follow-up to his technofuturist essay “Machines of Loving Grace,” on Monday Anthropic CEO Dario Amodei published a 20,000-word essay called “The Adolescence of Technology,” laying out the risks that AI presents and how we can overcome them.
Amodei highlights five key areas of risk: autonomy, misuse for destruction, seizing power, economic disruption and the unforeseen risks.
- The first concern involves the AI systems themselves, with Amodei noting that they risk absorbing problematic and unexpected behaviours due to biased training data.
- Then there are the risks that stem from human nature: putting weapons-grade capabilities into the hands of anyone with a motive, from singular bad actors to authoritarian governments.
- Amodei also called attention to the risk that many worry about today: economic disruption, labor displacement and wealth concentration as a result of AI moving faster than humans can adapt.
- And of course, there’s the fear of the unknown, as Amodei dubbed the “black seas of infinity.” This includes unhealthy dependencies on AI and a general sense of nihilism and purposelessness.
Still, for each of these problems, Amodei proposed solutions, including targeted, evidence-based regulation, safety guardrails and interpretability, government transparency requirements, and generally avoiding “doomerism.” More broadly, humanity must learn to expect the unexpected as we navigate through the growing pains of a technological revolution, stated Amodei.
“I have seen enough courage and nobility to believe that we can win—that when put in the darkest circumstances, humanity has a way of gathering, seemingly at the last minute, the strength and wisdom needed to prevail,” Amodei wrote at the end of his essay. “We have no time to lose.”