[Yet another "philosophizing" post, but one with some actual numbers. --Boaz] Recently there have been many debates on “artificial general intelligence” (AGI) and whether or not we are close to achieving it by scaling up our current AI systems. In this post, I’d like to make this debate a bit more quantitative by trying to … Continue reading Injecting some numbers into the AGI debate

# Category: Philosophizing

# The uneasy relationship between deep learning and (classical) statistics

An often-expressed sentiment is that deep learning (and machine learning in general) is “simply statistics,” in the sense that it uses different words to describe the same concepts statisticians have been studying for decades. In the 1990s, Rob Tibshirani wrote the following tongue-in-cheek “glossary”:: Something about this table resonates with me. In fact, as anyone … Continue reading The uneasy relationship between deep learning and (classical) statistics

# Why I am not a longtermist

[Apologies for yet another "philosophizing" blog post, hope to get back to posts with more equations soon... Should also mention that one response to this piece was that "anyone who writes a piece called "Why I am not a longtermist" is probably more of a longtermist than 90% of the population" 🙂 --Boaz] “Longtermism” is … Continue reading Why I am not a longtermist

# Philosophy of science and the blockchain: A book review

This blog post is a book review of sorts for the following two books: To Explain the World: The Discovery of Modern Science by Steven Weinberg (2016) The Knowledge Machine: How Irrationality Created Modern Science by Michael Strevens (2020). Both books cover (in different proportions) the history and philosophy of science. By the end of … Continue reading Philosophy of science and the blockchain: A book review

# ML Theory with bad drawings

(Next post: Lecture 1 - a blitz through classical statistical learning theory . See also all seminar posts) This semester I am teaching a seminar on the theory of machine learning. For the first lecture, I would like to talk about what is the theory of machine learning. I decided to write this (very rough!) … Continue reading ML Theory with bad drawings

# Physics Envy

There is something cool about physics. Black holes, anti-matter, "God's particle": it all sounds so exciting. While our TCS "mental experiments" typically involve restricting the inputs of constant-depth circuits, physicists talk about jumping into black holes while holding a dictionary. Physicists also have a knack for names: notions such as "uncertainty principle" or "monogamy of … Continue reading Physics Envy

# The different forms of quantum computing skepticism

(see also pdf version) Quantum computing is one of the most exciting developments of computer science in the last decades. But this concept is not without its critics, often known as "quantum computing skeptics" or "skeptics" for short. The debate on quantum computing can sometimes confuse the physical and mathematical aspects of this question, … Continue reading The different forms of quantum computing skepticism

# Bayesianism, frequentism, and the planted clique, or do algorithms believe in unicorns?

(See also pdf version , and these lecture notes) The divide between frequentists and Bayesians in statistics is one of those interesting cases where questions of philosophical outlook have actual practical implications. At the heart of the debate is Bayes’ theorem: $latex \Pr[A|B] = \Pr[A \cap B ]/\Pr[B]\;.$ Both sides agree that it is correct, but they disagree … Continue reading Bayesianism, frequentism, and the planted clique, or do algorithms believe in unicorns?

# Speaking about unspeakables

[Below is the transcript of a talk I gave at the Harvard Lectures that Last event (see video). This was a talk for a general (non technical) audience, but I still thought some people might find it interesting. For some more technical discussion of the notion of "computational Bayesian probabilities" I hint at toward the end, … Continue reading Speaking about unspeakables

# Is computational hardness the rule or the exception?

As a cryptographer, I am used to the viewpoint that computational problems are presumed hard unless proven otherwise. We cryptographers constantly come up with new hardness assumptions, and generally use the heuristic that if a problem doesn't have an obvious algorithm then it must be hard. We've had some interesting failures (see here for a recent example), but this heuristic seems … Continue reading Is computational hardness the rule or the exception?