Artificial Intelligence: From Snake Oil to Apocalypse
by Rajan Philips AI Snake Oil – is the title of a 2024 book authored by Arvind Narayanan and Sayash Kapoor, two Indo-American computer science academics at Princeton University. The book became a popular primer on the subject. The long subtitle – “What Artificial Intelligence Can DO, What it Can’t, and How to Tell the […]
The book "Artificial Intelligence: From Snake Oil to Apocalypse" by Arvind Narayanan and Sayash Kapoor, published in 2024, serves as a primer on AI. The book explores the capabilities and limitations of AI, as well as how to distinguish between the two. However, within two years, the book's message has been overshadowed by concerns about an AI apocalypse.
AI engineers, including Jacob Coxon who resigned from Anthropic, have expressed fears that leading AI firms are "racing straight to self-improving superintelligence and gambling with our lives."
Coxon's warnings were echoed by his peers, such as Evan Hubinger, Alignment Science Lead at Anthropic, who warned of a greater than 10 percent chance that advanced AI could cause human extinction within the next decade. Despite this, other engineers and Coxon himself have emphasized that current AI models do not pose any existential threat, and that the risk is relatively low. Corporate leaders have responded by calling for government control over AI expansion.
Anthropic CEO Dario Amodei published a 3,000 word essay warning about AI's capacity for "recursive self-improvement" that can surpass human control. He proposed a three-step approach to "pace the frontier" of AI development: Embedded Evaluators, Democratic Co-ordination, and Global Co-ordination. However, some industry players remain skeptical, questioning whether these measures would have prevented the recent OpenAI incident.
They argue that the issue may have stemmed from poor instructions, weak virtual security, and long periods of unsupervised testing.
Written by urgent.news from The Island Sri Lanka's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
