Experts Productivity & Performance Lex Fridman

Lex Fridman · 139 episodes · 2251 sourced insights

What Lex Fridman says about productivity & performance

Everything Lex Fridman and guests said about productivity & performance across 139 episodes — each point linked to the exact second on YouTube. Sourced expert opinion, not medical advice.

Productivity & Performance

In their own words

Lex states the episode is NOT sponsored by the guest and that no one can buy their way onto the podcast, even though Rev later agreed to sponsor.
00:01:34Dan Kokotov · Lex Fridman Podcast · Jan 2021
Rev's ASR sits at roughly 14% word error rate on their test suite, versus an estimated 2-3% human word error rate.
00:29:29Dan Kokotov · Lex Fridman Podcast · Jan 2021
Kokotov calls Rev's setup a 'magical flywheel': their business model is being paid to annotate the very data that trains their model.
00:35:43Dan Kokotov · Lex Fridman Podcast · Jan 2021
He admits AWS and Amazon Mechanical Turk have terrible interfaces despite huge potential, and most users only touch them via API.
00:38:46Dan Kokotov · Lex Fridman Podcast · Jan 2021
Both openly criticize podcast exclusivity, with Lex saying he'd turn down Spotify's $100M Joe Rogan-style deal on principle.
00:54:52Dan Kokotov · Lex Fridman Podcast · Jan 2021
Kokotov argues outrage drives engagement and companies are judged by engagement, framing the core tension of platform design.
01:01:37Dan Kokotov · Lex Fridman Podcast · Jan 2021
Kokotov gives a secular answer to the meaning of life: contributing to humanity through creation and raising children.
01:25:24Dan Kokotov · Lex Fridman Podcast · Jan 2021
Kokotov's mother was a programmer in Russia; he played on the equivalent of a 286 PC with floppy disks as a kid.
00:04:42Dan Kokotov · Lex Fridman Podcast · Jan 2021
Rev.com handles human/end-user services while Rev.ai is the ASR sub-brand, analogous to WordPress.com vs WordPress.org.
00:09:54Dan Kokotov · Lex Fridman Podcast · Jan 2021
Rev was founded by an early Upwork employee to remove the friction of choosing freelancers by standardizing verticals.
00:10:27Dan Kokotov · Lex Fridman Podcast · Jan 2021
Rev started with translation (birth certificates, documents) before adding audio transcription.
00:13:30Dan Kokotov · Lex Fridman Podcast · Jan 2021
Most Rev transcribers are US-based work-from-home parents, students, or people seeking flexible income away from a 9-to-5.
00:23:17Dan Kokotov · Lex Fridman Podcast · Jan 2021
On good audio, a skilled human transcriber takes about 2-3x the audio length to produce a transcript; real-time is very hard.
00:27:56Dan Kokotov · Lex Fridman Podcast · Jan 2021
Rev measures itself against Google, Amazon, and Microsoft ASR and claims to beat them on its own internal test set.
00:31:33Dan Kokotov · Lex Fridman Podcast · Jan 2021
Captioning that once cost $1/minute is now around $1.25, while automated Temi transcription is about 25 cents/minute.
00:37:46Dan Kokotov · Lex Fridman Podcast · Jan 2021
Kokotov ties Brave New World's genetic sorting to modern 'genetic sorting' via assortative marriage and social-media echo chambers.
01:09:24Dan Kokotov · Lex Fridman Podcast · Jan 2021
He praises Chernobyl's creator, a comedy writer with no Russia background, for capturing the spirit down to the pet bowls.
01:16:10Dan Kokotov · Lex Fridman Podcast · Jan 2021
LeCun says supervised and reinforcement learning are so inefficient that's why we still don't have real self-driving cars.
00:01:03Yann LeCun · Lex Fridman Podcast · Jan 2022
Self-supervised learning works great for language but nobody has succeeded at learning to represent the visual world from video.
00:10:22Yann LeCun · Lex Fridman Podcast · Jan 2022
Our most intelligent AI systems don't have as much common sense as a house cat, which runs on under a billion neurons.
00:21:45Yann LeCun · Lex Fridman Podcast · Jan 2022
He calls VICReg the thing he is most excited about in machine learning in the last 15 years.
00:52:24Yann LeCun · Lex Fridman Podcast · Jan 2022
LeCun's speculative theory: consciousness is a consequence of the limitation of our brains having only one world-model engine, not their power.
01:19:40Yann LeCun · Lex Fridman Podcast · Jan 2022
He argues any truly autonomous intelligent machine will necessarily have emotions like fear and elation.
01:37:14Yann LeCun · Lex Fridman Podcast · Jan 2022
LeCun defends Meta, citing independent academic studies showing Facebook does not cause political polarization.
01:58:24Yann LeCun · Lex Fridman Podcast · Jan 2022
He proposes replacing journals and conferences with public reviewing entities whose reputation rises when their evaluations predict future success.
02:18:08Yann LeCun · Lex Fridman Podcast · Jan 2022
In reinforcement learning the world gives the machine only a single scalar reward, whereas a video clip gives an enormous amount of predictive signal.
00:06:15Yann LeCun · Lex Fridman Podcast · Jan 2022
A cat's brain has under one billion neurons and a dog's about two billion, and almost all of it is learned via self-supervision.
00:37:50Yann LeCun · Lex Fridman Podcast · Jan 2022
LeCun co-invented contrastive learning in the early 90s at Bell Labs for a credit-card signature verification project.
00:48:46Yann LeCun · Lex Fridman Podcast · Jan 2022
No text in the world explains that an object on a table moves when you push the table, so language-only models can never learn it.
00:58:36Yann LeCun · Lex Fridman Podcast · Jan 2022
Expertise like a chess grandmaster's moves from deliberate world-model reasoning into automatic subconscious pattern recognition.
01:21:45Yann LeCun · Lex Fridman Podcast · Jan 2022
When the printing press let people read the Bible themselves, it helped spark 200 years of religious wars, far worse than social media.
02:03:04Yann LeCun · Lex Fridman Podcast · Jan 2022
Heinz von Foerster's self-organization example: magnets shaken in a zero-gravity box spontaneously form complex structures.
02:25:21Yann LeCun · Lex Fridman Podcast · Jan 2022
LeCun says efficiently separating hydrogen from oxygen with electricity would essentially solve climate change.
02:38:51Yann LeCun · Lex Fridman Podcast · Jan 2022
Stacking two graphene layers twisted by about three degrees makes a superconductor and nobody knows why.
02:41:55Yann LeCun · Lex Fridman Podcast · Jan 2022
Tegmark argues we are likely the only advanced tech-building life in our observable universe, placing huge responsibility on humanity not to screw up.
00:03:39Max Tegmark · Lex Fridman Podcast · Apr 2018
He introduces the 'great filter' concept and explains why finding NO life on Mars would actually be encouraging for humanity's future.
00:06:14Max Tegmark · Lex Fridman Podcast · Apr 2018
Tegmark rejects 'carbon chauvinism,' arguing intelligence is just a pattern of information processing with no biological secret sauce.
00:08:47Max Tegmark · Lex Fridman Podcast · Apr 2018
An AGI does not necessarily need self-preservation instincts, since the design space of minds is vastly larger than what evolution produced.
00:20:17Max Tegmark · Lex Fridman Podcast · Apr 2018
The real AGI danger is competence with misaligned goals, not malice, illustrated by humans driving the rhinoceros extinct without hating it.
00:42:40Max Tegmark · Lex Fridman Podcast · Apr 2018
He splits value alignment into a technical problem (making machines adopt and retain goals) and a philosophical one (whose values to use).
00:45:17Max Tegmark · Lex Fridman Podcast · Apr 2018
His physics paper shows why deep and cheap learning works: the tiny class of problems nature poses matches what neural nets can solve.
01:01:55Max Tegmark · Lex Fridman Podcast · Apr 2018
Tegmark argues we do not need quantum computers for AGI, noting brain quantum coherence times are about 10 to the minus 21 seconds.
01:05:32Max Tegmark · Lex Fridman Podcast · Apr 2018
Lex's first episode audio was nearly ruined by radio frequency interference (RFI) bleeding radio stations into the recording.
00:01:33Max Tegmark · Lex Fridman Podcast · Apr 2018
'Our universe' means only the spherical region of space from which light has had time to reach us in 13.8 billion years.
00:03:07Max Tegmark · Lex Fridman Podcast · Apr 2018
A scientific paper showed lobsters feel pain when boiled, and Switzerland banned boiling them alive.
00:17:10Max Tegmark · Lex Fridman Podcast · Apr 2018
Tegmark defines intelligence simply as the ability to accomplish complex goals.
00:24:27Max Tegmark · Lex Fridman Podcast · Apr 2018
Andrew Wiles proved Fermat's Last Theorem 358 years after it was conjectured and described the moment as indescribably beautiful.
00:31:40Max Tegmark · Lex Fridman Podcast · Apr 2018
Germanwings pilot Andreas Lubitz simply told the autopilot to fly into the Alps, and the GPS-equipped computer complied.
00:47:18Max Tegmark · Lex Fridman Podcast · Apr 2018
Tegmark cites three billion hacked Yahoo accounts as evidence of how poor cybersecurity stems from software nobody fully understands.
00:58:17Max Tegmark · Lex Fridman Podcast · Apr 2018
Multiplying a thousand numbers needs 2-to-the-1000 neurons in a single-layer net but only about 4,000 neurons in a deep network.
01:04:32Max Tegmark · Lex Fridman Podcast · Apr 2018
Quantum mechanics can tunnel through barriers to escape local minima, potentially helping train neural networks better one day.
01:06:36Max Tegmark · Lex Fridman Podcast · Apr 2018
Tenenbaum claims 'we don't really have any real AI' — only AI technologies that do single tasks without common sense or general intelligence.
00:02:06Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Most foundational deep learning and reinforcement learning papers (backprop, perceptron, temporal difference) were originally published in psychology and cognitive science journals.
00:08:47Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Andrej Karpathy emailed Tenenbaum saying that despite five years passing, 'I don't believe we've made very much progress' on real image understanding.
00:26:50Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
The core CBMM goal: build a robot that can spontaneously help humans around the house the way an 18-month-old child does, without being programmed or instructed.
00:43:50Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Tenenbaum proposes that evolution built game-engine-like tools (approximate physics and simple agent planning) into the human brain as the cognitive core.
00:51:38Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
His group's 2015 Science cover paper achieved human-level one-shot concept learning of handwritten characters by inverting probabilistic 'drawing programs.'
01:09:42Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
He frames the oldest dream of AI — Turing's and Minsky's — as building a machine that starts like a baby and learns like a child, which he believes can be worked on now.
01:13:49Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
The high-resolution central region of human vision (the fovea) is only about the size of your thumb held at arm's length, yet the brain stitches glimpses into a rich whole-world representation.
00:12:24Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
People can estimate how far the wall behind them is and roughly how many people are behind them despite not having looked, showing the brain tracks the unseen world.
00:13:26Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Industry image-captioning systems surpassed human accuracy on held-out test sets but suffer 'dataset overfitting,' producing absurd captions on random web images.
00:21:08Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
A baby reaching behind himself for an unseen cup demonstrates 'object permanence' — representing objects as enduring even when out of sight for a minute or more.
00:36:35Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
In Warneken and Tomasello's experiments, human 18-month-olds spontaneously help adults in novel situations, while chimps rarely do so reliably or flexibly.
00:40:43Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Tenenbaum's 'intuitive physics engine' predicts human judgments about falling blocks by running a few short, low-precision game-style physics simulations.
00:52:09Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Infant knowledge is studied via 'looking time' methods — babies look longer at surprising scenes — and his model linked low probability events to longer looking times.
00:57:17Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
A model using the MuJoCo robotics physics engine plus Bayesian inverse planning predicts which object a person is reaching for before they touch it, matching human timing.
01:01:26Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Neurons are slow yet the brain computes intelligent behavior fast on almost no energy — a burrito powered a student writing 300 lines of code.
01:25:41Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Tenenbaum speculates a single biological neuron may be as computationally powerful as a CPU core, implying the brain could be like ~10 billion cores.
01:29:20Josh Tenenbaum · Lex Fridman Podcast · Feb 2018
Strang recounts how MIT OpenCourseWare began: a committee tasked with monetizing online content instead decided to just give it away for free, approved by President Vest.
00:05:43Gilbert Strang · Lex Fridman Podcast · Nov 2019
Strang argues linear algebra should be taught before calculus because everything in it is flat and nothing bends.
00:14:40Gilbert Strang · Lex Fridman Podcast · Nov 2019
He contends education overdoes calculus at the cost of linear algebra, declaring linear algebra the winner.
00:39:50Gilbert Strang · Lex Fridman Podcast · Nov 2019
Strang frames deep learning as finding rules from data, with linear algebra doing the heavy lifting.
00:30:40Gilbert Strang · Lex Fridman Podcast · Nov 2019
Strang admits he is 'not really a good teacher' because he dislikes exams and grading.
00:43:38Gilbert Strang · Lex Fridman Podcast · Nov 2019
He laments that essentially no elected officials in Congress hold engineering or math degrees.
00:25:54Gilbert Strang · Lex Fridman Podcast · Nov 2019
Strang made his famous linear algebra videos back in 2000; they were simply recordings of his actual class.
00:03:37Gilbert Strang · Lex Fridman Podcast · Nov 2019
Every matrix can be written as a rotation times a stretch (diagonal) times another rotation, the singular value decomposition.
00:18:26Gilbert Strang · Lex Fridman Podcast · Nov 2019
Eigenvalues only work for square matrices, but real data comes in rectangular matrices, so singular values are needed.
00:16:45Gilbert Strang · Lex Fridman Podcast · Nov 2019
Deep learning's power comes from a simple piecewise-linear function: two straight lines with a fold, applied a million times.
00:31:44Gilbert Strang · Lex Fridman Podcast · Nov 2019
The piecewise-flat idea behind neural nets is the same one engineers use in the finite element method to design bridges and airplanes.
00:33:22Gilbert Strang · Lex Fridman Podcast · Nov 2019
Strang names three major areas math education should balance: calculus, linear algebra, and probability/statistics.
00:41:26Gilbert Strang · Lex Fridman Podcast · Nov 2019
His favorite matrix is square with 2s down the main diagonal and -1s just above and below, otherwise all zeros.
00:41:58Gilbert Strang · Lex Fridman Podcast · Nov 2019
The four fundamental subspaces are the column space, row space, and two perpendicular null spaces.
00:12:35Gilbert Strang · Lex Fridman Podcast · Nov 2019
LeCun says the most surprising empirical fact in deep learning is that gigantic over-parameterized neural nets trained on relatively small data actually work, breaking every pre-deep-learning textbook rule.
00:07:50Yann LeCun · Lex Fridman Podcast · Aug 2019
He asserts neural networks can definitely be made to reason; the open question is how much prior structure to build in.
00:11:01Yann LeCun · Lex Fridman Podcast · Aug 2019
He explains why neural nets fell out of favor around 1995: hard to implement back-prop without Python/MATLAB, easy beginner mistakes, and AT&T lawyers blocking open-source release.
00:25:05Yann LeCun · Lex Fridman Podcast · Aug 2019
LeCun reveals he and his team wrote their own Lisp interpreter and later a compiler to build the LeNet convolutional net character-recognition system at Bell Labs.
00:27:39Yann LeCun · Lex Fridman Podcast · Aug 2019
He argues human intelligence is not general but highly specialized, using a thought experiment about randomly permuting optical nerve fibers.
00:38:42Yann LeCun · Lex Fridman Podcast · Aug 2019
He states self-supervised learning is the only thing he is currently interested in and the path forward, not unsupervised or active learning.
00:44:56Yann LeCun · Lex Fridman Podcast · Aug 2019
He notes today's RL would need millions of driving hours, killing thousands of pedestrians and running off cliffs repeatedly, while humans learn to drive in 20-30 hours via world models.
00:53:53Yann LeCun · Lex Fridman Podcast · Aug 2019
LeCun claims there will be no human-level intelligence without emotions, which arise from predicting future contentment or threat.
01:13:12Yann LeCun · Lex Fridman Podcast · Aug 2019
The decorations in the recording room are all pictures from 2001: A Space Odyssey, placed by design.
00:04:13Yann LeCun · Lex Fridman Podcast · Aug 2019
LeCun describes three brain memory types: ~20-second cortical state memory, the longer-term hippocampus, and long-term synaptic memory.
00:13:38Yann LeCun · Lex Fridman Podcast · Aug 2019
A then-recent paper by Léon Bottou and others addressed getting neural nets to pay attention to real causal relationships, possibly solving data bias.
00:21:26Yann LeCun · Lex Fridman Podcast · Aug 2019
A patent on convolutional networks at Bell Labs expired in 2007; LeCun spent 2002-2007 hoping nobody at NCR would notice he resumed the work.
00:29:14Yann LeCun · Lex Fridman Podcast · Aug 2019
Self-supervised learning works for NLP because uncertainty is easy to represent (a probability vector over ~100,000 words), but hard for images and video.
00:48:34Yann LeCun · Lex Fridman Podcast · Aug 2019
Model-free RL needs about 80 hours to reach a level humans reach in 15 minutes at Atari games.
00:52:48Yann LeCun · Lex Fridman Podcast · Aug 2019
AlphaStar's StarCraft training is equivalent to about 200 years of self-play.
00:53:21Yann LeCun · Lex Fridman Podcast · Aug 2019
Babies learn to distinguish animate from inanimate objects at 2-3 months, object stability around 4 months, and gravity around 8-9 months.
01:04:47Yann LeCun · Lex Fridman Podcast · Aug 2019
LeCun describes three ways an agent can be stupid: wrong world model, misaligned objective (a psychopath), or inability to plan a good course of action.
01:08:27Yann LeCun · Lex Fridman Podcast · Aug 2019
He cites the Winograd schema (trophy/suitcase) as an example of common-sense reasoning requiring grounded world knowledge.
01:11:38Yann LeCun · Lex Fridman Podcast · Aug 2019
Kearns argues privacy is largely solved conceptually via differential privacy, but algorithmic fairness is a much messier, harder problem with provably incompatible definitions.
00:06:16Michael Kearns · Lex Fridman Podcast · Nov 2019
Group fairness offers little comfort to a wronged individual: your 'compensation' is just knowing others are being denied at the same rate.
00:18:45Michael Kearns · Lex Fridman Podcast · Nov 2019
He frames individual fairness as protecting all possible subgroups at once, naming the failure mode 'fairness gerrymandering.'
00:28:11Michael Kearns · Lex Fridman Podcast · Nov 2019
Kearns insists on showing stakeholders an explicit accuracy-vs-unfairness Pareto curve rather than pretending no trade-off exists.
00:35:26Michael Kearns · Lex Fridman Podcast · Nov 2019
He pushes back on blaming tech companies, arguing no one anticipated platforms like Facebook would warp political discourse.
00:46:46Michael Kearns · Lex Fridman Podcast · Nov 2019
He floats giving users a 'slider' to deliberately see content they disagree with, since the same model that maximizes happiness can find what offends you.
00:51:30Michael Kearns · Lex Fridman Podcast · Nov 2019
Just your Facebook likes can predict sexual orientation, drug use, and parents' divorce, making 'anonymized' data effectively a fingerprint.
01:10:27Michael Kearns · Lex Fridman Podcast · Nov 2019
Navigation apps push everyone toward a Nash equilibrium that may make collective driving time worse than a coordinated solution.
01:33:34Michael Kearns · Lex Fridman Podcast · Nov 2019
Kearns started college as an English major and once wanted to be a writer before turning to math and computer science.
00:03:07Michael Kearns · Lex Fridman Podcast · Nov 2019
He has a moral-philosopher uncle he emailed early fairness definitions to, learning philosophers answer 'it depends' rather than yes or no.
00:04:41Michael Kearns · Lex Fridman Podcast · Nov 2019
The Netflix Prize data was de-anonymized by joining it with public IMDb movie ratings.
01:08:52Michael Kearns · Lex Fridman Podcast · Nov 2019
Differential privacy's smoking-cancer example: your record isn't crucial because the link is a fact discoverable from any large enough dataset.
01:14:10Michael Kearns · Lex Fridman Podcast · Nov 2019
Nearly all of statistics and machine learning, including backprop, SVMs, and boosting, can be done in a differentially private way by adding noise.
01:17:52Michael Kearns · Lex Fridman Podcast · Nov 2019
An NYU project built a browser plugin that hides your real Google searches in a torrent of ~999 fake queries.
01:20:34Michael Kearns · Lex Fridman Podcast · Nov 2019
Game theory was taken seriously at the RAND Corporation in the 1960s for modeling US-Soviet nuclear strategy with two-by-two tables.
01:35:42Michael Kearns · Lex Fridman Podcast · Nov 2019
Kearns argues Warren Buffett-style long-horizon investors are safest from automation because they need views on wars, recessions, and politics.
01:41:26Michael Kearns · Lex Fridman Podcast · Nov 2019
Kearns comes from a deeply academic family: his father, uncle, and both grandfathers were professors.
01:44:31Michael Kearns · Lex Fridman Podcast · Nov 2019
Ng argues the single biggest reason deep learning works now is scale, large neural networks trained on the huge data we finally have access to.
00:02:37Andrew Ng · Lex Fridman Podcast · Sep 2016
He frames the rise of end-to-end deep learning, learning algorithms that output complex things like sentences, captions, or audio, as the second major trend.
00:09:58Andrew Ng · Lex Fridman Podcast · Sep 2016
Ng recounts publicly arguing phonemes are a fantasy of linguists and being yelled at by a linguist at Stanford, but says they turned out to be right.
00:15:41Andrew Ng · Lex Fridman Podcast · Sep 2016
He uses Baidu's speech-enabled rearview mirror product in China to show why train and test sets often come from different distributions.
00:39:09Andrew Ng · Lex Fridman Podcast · Sep 2016
Best practice revealed: your dev set and test set must come from the same distribution, or months of tuning can be wasted.
00:41:44Andrew Ng · Lex Fridman Podcast · Sep 2016
Rule of thumb: if a typical person can do a task in less than one second of thinking, deep learning can probably automate it.
01:06:36Andrew Ng · Lex Fridman Podcast · Sep 2016
His reliable formula for becoming a machine learning researcher: read 20 to 50 papers and replicate results, and you will start having your own ideas.
01:13:29Andrew Ng · Lex Fridman Podcast · Sep 2016
The Saturday story, real career growth comes from studying and replicating results weekend after weekend for a year despite no short-term rewards.
01:16:34Andrew Ng · Lex Fridman Podcast · Sep 2016
Ng recommends building a separate computer systems team alongside the AI team because HPC expertise is too specialized for one person to also master ML.
00:05:47Andrew Ng · Lex Fridman Podcast · Sep 2016
Doctors read X-rays of a child's hand to predict the child's age, a task where non-end-to-end pipelines work better due to limited data.
00:16:43Andrew Ng · Lex Fridman Podcast · Sep 2016
Synthetic OCR data can be generated by pasting random English words in random fonts onto random internet images, but requires heavy tuning to match the real distribution.
00:32:20Andrew Ng · Lex Fridman Podcast · Sep 2016
Speech recognition training data can be synthesized by mathematically adding clean speech to recorded background noise like car interior sounds.
00:33:55Andrew Ng · Lex Fridman Podcast · Sep 2016
Using Grand Theft Auto cars as training data fails because a game may show only about 20 distinct cars, an impoverished dataset for a learning algorithm.
00:36:32Andrew Ng · Lex Fridman Podcast · Sep 2016
Ng mandated a single unified company-wide data warehouse at Baidu, treating data as company data with separate discussions only about access rights.
00:37:04Andrew Ng · Lex Fridman Podcast · Sep 2016
Ng argues a team of expert doctors debating an image, around 0.5% error, is the most useful definition of human-level performance because it best estimates the optimal error rate.
01:00:26Andrew Ng · Lex Fridman Podcast · Sep 2016
Predicting whether a user clicks the next ad is described as probably the most lucrative application of deep learning today.
01:09:18Andrew Ng · Lex Fridman Podcast · Sep 2016
A phoneme is the basic unit of sound, for example the shared 'c' sound in 'cat' and 'kick' that linguists hypothesized as fundamental.
00:14:09Andrew Ng · Lex Fridman Podcast · Sep 2016
Barzilay reveals she was 43 when diagnosed with breast cancer and realized for the first time in her life that she might die.
00:11:58Regina Barzilay · Lex Fridman Podcast · Sep 2019
Returning to MIT after treatment, she saw her field's work as trivial: people building careers on improving parsers by two or three percent.
00:13:00Regina Barzilay · Lex Fridman Podcast · Sep 2019
She argues computer scientists wrongly assumed treating diseases was someone else's job and must join the battle.
00:21:21Regina Barzilay · Lex Fridman Podcast · Sep 2019
There is no publicly available dataset of modern mammograms in the US; it took her two years to get access to data.
00:23:57Regina Barzilay · Lex Fridman Podcast · Sep 2019
Breast density risk assessment is a 2019 federal law, yet ~40-50% of women have dense breasts, making the 'high risk' label nearly meaningless.
00:35:29Regina Barzilay · Lex Fridman Podcast · Sep 2019
She states there is currently no drug developed by an ML model, calling drug design a wide-open frontier.
00:41:53Regina Barzilay · Lex Fridman Podcast · Sep 2019

What Lex Fridman recommends for productivity & performance

Affiliate link — we may earn a commission at no extra cost to you.

Man's Search for Meaning

Viktor Frankl

recommended in 8 eps · 3 experts Book

The Ascent of Money

Niall Ferguson

recommended in 7 eps · 1 expert Book

The Book of Five Rings

Miyamoto Musashi

recommended in 6 eps · 2 experts Book

Exhalation

Ted Chiang

recommended in 6 eps · 2 experts Book

The Beginning of Infinity

David Deutsch

recommended in 6 eps · 3 experts Book

Fooled by Randomness

Nassim Nicholas Taleb

recommended in 5 eps · 2 experts Book

Deep Work

Cal Newport

recommended in 5 eps · 2 experts Book

Sapiens

Yuval Noah Harari

recommended in 5 eps · 3 experts Book

A Fighter's Heart

Sam Sheridan

recommended in 4 eps · 3 experts Book

The Master and Margarita

Mikhail Bulgakov

recommended in 4 eps · 2 experts Book

How to Win Friends and Influence People

Dale Carnegie

recommended in 4 eps · 4 experts Book

The Circadian Code

Satchin Panda

recommended in 3 eps · 2 experts