The Hidden Complexity of Wishes — LessWrong

The genie knows, but doesn't care — LessWrong

Followup to: The Hidden Complexity of Wishes, Ghosts in the Machine, Truly Part of You …

On technological optimism and technological pragmatism

"There are no guarantees that things are going to turn out very well for anyone.”

The Ratchet: How Preference Standards Erase What They Can't Express - Astral's Blog

The Role of Luck in Life Success Is Far Greater Than We Realized

Are the most successful people in society just the luckiest people?

People Reject Algorithms in Uncertain Decision Domains Because They Have Diminishing Sensitivity to Forecasting Error

Will people use self-driving cars, virtual doctors, and other algorithmic decision-makers if they outperform humans? The answer depends on the uncertainty inherent in the decision domain. We propose that people have diminishing sensitivity to forecasting error and that this preference results in people favoring riskier (and often worse-performing) decision-making methods, such as human judgment, in inherently uncertain domains. In nine studies ( N = 4,820), we found that (a) people have diminishing sensitivity to each marginal unit of error that a forecast produces, (b) people are less likely to use the best possible algorithm in decision domains that are more unpredictable, (c) people choose between decision-making methods on the basis of the perceived likelihood of those methods producing a near-perfect answer, and (d) people prefer methods that exhibit higher variance in performance (all else being equal). To the extent that investing, medical decision-making, and other domains are inherently uncertain, people may be unwilling to use even the best possible algorithm in those domains.

New Nature

Contours of can’t-be-evil futures

Why major life decisions should be approached with ‘epistemic humility’ | Psyche Videos

At a crossroads in life? No matter how much you deliberate, one unknowable remains: the person you will be on the other side

The Human Condition: Second Edition

A classic work of political thought, more relevant today than ever The twenty-first century has seen a resurgence of interest in the political thinker Hannah Arendt, “the theorist of beginnings,” whose work probes the logics underlying unexpected transformations—from totalitarianism to revolution. A work of striking originality, The Human Condition is in many respects more relevant now than when it first appeared in 1958. In her study of the state of modern humanity, Hannah Arendt considers humankind from the perspective of the actions of which it is capable. The problems Arendt identified then—diminishing human agency and political freedom, the paradox that as human powers increase through technological and humanistic inquiry, we are less equipped to control the consequences of our actions—continue to confront us today. This new edition, published to coincide with the sixtieth anniversary of its original publication, contains Margaret Canovan’s 1998 introduction and a new foreword by Danielle Allen. A classic in political and social theory, The Human Condition is a work that has proved both timeless and perpetually timely.

News

To improve anything, we need to dare to imagine and make what doesn’t exist.

A Theory of Response Sampling in LLMs: Part Descriptive and Part Prescriptive

Large Language Models (LLMs) are increasingly utilized in autonomous decision-making, where they sample options from vast action spaces. However, the heuristics that guide this sampling process remain under-explored. We study this sampling behavior and show that this underlying heuristics resembles that of human decision-making: comprising a descriptive component (reflecting statistical norm) and a prescriptive component (implicit ideal encoded in the LLM) of a concept. We show that this deviation of a sample from the statistical norm towards a prescriptive component consistently appears in concepts across diverse real-world domains like public health, and economic trends. To further illustrate the theory, we demonstrate that concept prototypes in LLMs are affected by prescriptive norms, similar to the concept of normality in humans. Through case studies and comparison with human studies, we illustrate that in real-world applications, the shift of samples toward an ideal value in LLMs' outputs can result in significantly biased decision-making, raising ethical concerns.