Increments - Episodes Tagged with “Ai”

#94 - Is AI Just a Tool? (w/ Scott Aaronson)

Ben Chugg and Vaden Masrani — Thu, 06 Nov 2025 10:00:00 -0800

The time has come for Vaden to defend his faith in the face of cold, hard scientific rationality. Will AI take over the world, automating away everything that makes humans distinct? Or can Vaden defend the church of just-ism, the radical belief that AI is simply "just a tool." Scott Aaronson, professor of computer science at UT Austin, goes to head to head against the zealotry.

Check out Scott's website and his blog, Shtetl Optimized.

We discuss

Scott view's on education. Should we radically reform K-12?
Is ChatGPT changing Scott's approach to teaching
The religion of "justa-ism"
Is AI just a tool?
Is there any principle which lets us say that AI won't be as general as humans?
Aaronson's thesis of Artificial Intelligence
Computational universality vs explanatory universality
The many-worlds interpretation of quantum mechanics

Socials

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani
Come join our discord server! DM us on twitter or send us an email to get a supersecret link
Become a patreon subscriber here. Or give us one-time cash donations to help cover our lack of cash donations here.
Click dem like buttons on youtube

Have you been converted? Tell us at incrementspodcast@gmail.com

Special Guest: Scott Aaronson.

#90 (Reaction) - Disbelieving AI 2027: Responding to "Why We're Not Ready For Superintelligence"

Ben Chugg and Vaden Masrani — Sun, 17 Aug 2025 18:00:00 -0700

Always the uncool kids at the table, Ben and Vaden push back against the AGI hype domininating every second episode of every second podcast. We react to "We're not ready for superintelligence" by 80,000 Hours - a bleak portrayal of the pre and post AGI world. Can Ben keep Vaden's sass in check? Can the 80,000 hours team find enough cubes for AGI? Is Agent-5 listening to you RIGHT NOW?

Listener Note:

We strongly recommend watching the video for this one, available both on youtube and spotify:
- https://www.youtube.com/@incrementspod
- https://open.spotify.com/show/1gKKSP5HKT4Nk3i0y4UseB

We discuss

The incentives of superforecasters
Arguments by authority
Whether superintelligence is right around the corner
The difference between model size and data
Are we running out of high quality data?
Does training on synthetic data work?
The assumptions behind the AGI claims
The pitfalls of reasoning from trends

References

Michael I Jordan
Neil Lawrence
[Important technical paper from Jordan pushing back on Doomerism](A Collectivist, Economic Perspective on AI)
Jordan article talking about dangers of using AlphaFold data
Nature paper showing you can't use synthetic data to train bigger models
Paper estimating of when training data will run out (Coincidentally enough, sometime between 2027-2028)

Socials

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani
Come join our discord server! DM us on twitter or send us an email to get a supersecret link
Become a patreon subscriber here. Or give us one-time cash donations to help cover our lack of cash donations here.
Click dem like buttons on youtube

But how many cubes until we get to AGI though? Send a few of your cubes over to incrementspodcast@gmail.com

Episode header image from here.

#77 (Bonus) - AI Doom Debate (w/ Liron Shapira)

Ben Chugg and Vaden Masrani — Tue, 19 Nov 2024 13:30:00 -0800

Back on Liron's Doom Debates podcast! Will we actually get around to the subject of superintelligent AI this time? Is it time to worry about the end of the world? Will Ben and Vaden emotionally recover from the devastating youtube comments from the last episode?

Follow Liron on twitter (@liron) and check out the Doom Debates youtube channel and podcast.

We discuss

Definitions of "new knowledge"
The reliance of deep learning on induction
Can AIs be creative?
The limits of statistical prediction
Predictions of what deep learning cannot accomplish
Can ChatGPT write funny jokes?
Trends versus principles
The psychological consequences of doomerism

Socials

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani, @liron
Come join our discord server! DM us on twitter or send us an email to get a supersecret link
The world is going to end soon, might as well get exclusive bonus content by becoming a patreon subscriber here. Or give us one-time cash donations to help cover our lack of cash donations here.
Click dem like buttons on youtube

Was Vaden's two week anti-debate bro reeducation camp successful? Tell us at incrementspodcast@gmail.com
Special Guest: Liron Shapira.

#76 (Bonus) - Is P(doom) meaningful? Debating epistemology (w/ Liron Shapira)

Ben Chugg and Vaden Masrani — Fri, 08 Nov 2024 14:30:00 -0800

Liron Shapira, host of [Doom Debates], invited us on to discuss Popperian versus Bayesian epistemology and whether we're worried about AI doom. As one might expect knowing us, we only got about halfway through the first subject, so get yourselves ready (presumably with many drinks) for part II in a few weeks! The era of Ben and Vaden's rowdy youtube debates has begun. Vaden is jubilant, Ben is uncomfortable, and the world has never been more annoyed by Popperians.

Follow Liron on twitter (@liron) and check out the Doom Debates youtube channel and podcast.

We discuss

Whether we're concerned about AI doom
Bayesian reasoning versus Popperian reasoning
Whether it makes sense to put numbers on all your beliefs
Solomonoff induction
Objective vs subjective Bayesianism
Prediction markets and superforecasting

References

Vaden's blog post on Cox's Theorem and Yudkowsky's claims of "Laws of Rationality": https://vmasrani.github.io/blog/2021/the_credence_assumption/
Disproof of probabilistic induction (including Solomonov Induction): https://arxiv.org/abs/2107.00749
EA Post Vaden Mentioned regarding predictions being uncalibrated more than 1yr out: https://forum.effectivealtruism.org/posts/hqkyaHLQhzuREcXSX/data-on-forecasting-accuracy-across-different-time-horizons#Calibrations
Article by Gavin Leech and Misha Yagudin on the reliability of forecasters: https://ifp.org/can-policymakers-trust-forecasters/
Superforecaster p(doom) is ~1%: https://80000hours.org/2024/09/why-experts-and-forecasters-disagree-about-ai-risk/#:~:text=Domain%20experts%20in%20AI%20estimated,by%202100%20(around%2090%25).
The existential risk persuasion tournament https://www.astralcodexten.com/p/the-extinction-tournament
Some more info in Ben's article on superforecasting: https://benchugg.com/writing/superforecasting/
Slides on Content vs Probability: https://vmasrani.github.io/assets/pdf/popper_good.pdf

Socials

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani, @liron
Come join our discord server! DM us on twitter or send us an email to get a supersecret link
Trust in the reverend Bayes and get exclusive bonus content by becoming a patreon subscriber here. Or give us one-time cash donations to help cover our lack of cash donations here.
Click dem like buttons on youtube

What's your credence that the second debate is as fun as the first? Tell us at incrementspodcast@gmail.com
Special Guest: Liron Shapira.

#49 - AGI: Could The End Be Nigh? (With Rosie Campbell)

Ben Chugg and Vaden Masrani — Wed, 22 Mar 2023 10:15:00 -0700

When big bearded men wearing fedoras begin yelling at you that the end is nigh and superintelligence is about to kill us all, what should you do? Vaden says don't panic, and Ben is simply awestruck by the ability to grow a beard in the first place.

To help us think through the potential risks and rewards of ever more impressive machine learning models, we invited Rosie Campbell on the podcast. Rosie is on the safety team at OpenAI and, while she's more worried about the existential risks of AI than we are, she's just as keen on some debate over a bottle of wine.

We discuss:

Whether machine learning poses an existential threat
How concerned we should be about existing AI
Whether deep learning can get us to artificial general intelligence (AGI)
If AI safety is simply quality assurance
How can we test if an AI system is creative?

References:

Mathgen: Randomly generated math papers

Contact us

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani
Follow Rosie at @RosieCampbell or https://www.rosiecampbell.xyz/
Check us out on youtube at https://www.youtube.com/channel/UC_4wZzQyoW4s4ZuE4FY9DQQ
Come join our discord server! DM us on twitter or send us an email to get a supersecret link

Prove you're creative by inventing the next big thing and then send it to us at incrementspodcast@gmail.com
Special Guest: Rosie Campbell.

#45 - Four Central Fallacies of AI Research (with Melanie Mitchell)

Ben Chugg and Vaden Masrani — Mon, 31 Oct 2022 10:45:00 -0700

We were delighted to be joined by Davis Professor at the Sante Fe Insitute, Melanie Mitchell! We chat about our understanding of artificial intelligence, human intelligence, and whether it's reasonable to expect us to be able to build sophisticated human-like automated systems anytime soon.

Follow Melanie on twitter @MelMitchell1 and check out her website: https://melaniemitchell.me/

We discuss:

AI hype through the ages
How do we know if machines understand?
Winograd schemas and the "WinoGrande" challenge.
The importance of metaphor and analogies to intelligence
The four fallacies in AI research:
- 1. Narrow intelligence is on a continuum with general intelligence
- 2. Easy things are easy and hard things are hard
- 3. The lure of wishful mnemonics
- 4. Intelligence is all in the brain
Whether embodiment is necessary for true intelligence
Douglas Hofstadter's views on AI
Ray Kurzweil and the "singularity"
The fact that Moore's law doesn't hold for software
The difference between symbolic AI and machine learning
What analogies have to teach us about human cognition

Errata

Ben mistakenly says that Eliezer Yudkowsky has bet that everyone will die by 2025. It's actually by 2030. You can find the details of the bet here: https://www.econlib.org/archives/2017/01/my_end-of-the-w.html.

References:

NY Times reporting on Perceptrons.
The WinoGrande challenge paper
Why AI is harder than we think
The Singularity is Near, by Ray Kurzweil

Contact us

Follow us on Twitter at @IncrementsPod, @BennyChugg, @VadenMasrani
Check us out on youtube at https://www.youtube.com/channel/UC_4wZzQyoW4s4ZuE4FY9DQQ
Come join our discord server! DM us on twitter or send us an email to get a supersecret link

Eliezer was more scared than Douglas about AI, so he wrote a blog post about it. Who wrote the blog post, Eliezer or Douglas? Tell us at over at incrementspodcast@gmail.com. Special Guest: Melanie Mitchell.

#9 - Facial Recognition Technology with Stephen Caines

Ben Chugg and Vaden Masrani — Thu, 06 Aug 2020 19:00:00 -0700

The talented Stephen Caines punctures the cloud of confusion that is Ben and Vaden's conception of facial recognition technology. We talk about the development and usage of facial recognition in the private and public spheres, the dangers and merits of the technology, and Vaden's plan to use it a bars. For God's sake don't give that man a GPU.

Stephen is a legal technologist with a passion for access to justice. He is a 2019 graduate of the University of Miami School of Law with a concentration in the Business of Innovation, Law, and Technology. While in law school, his work focused on public interest, legal aid organizations, and non-profits. He was a 2018 Access to Justice Technology Fellow and has worked with the Legal Services of Greater Miami, Inc. on a variety of technology initiatives aimed at optimizing their operations. Additionally, he worked on the legislative and technology policy team of the Cyber Civil Rights Initiative. Stephen’s current work focuses on developing standards and best practices for the safe and ethical implementation of technology in the public sector.

References:

Stephen's website.

Perpetual Lineup Project (out of Georgetown)

Stephen on the Our Data podcast

IBM, Amazon, and Microsoft put moratoria on some aspects of their FRT technology.

Clearview AI

Special Guest: Stephen Caines.

#7 - Philosophy of Probability II: Existential Risks

Ben Chugg and Vaden Masrani — Tue, 07 Jul 2020 11:00:00 -0700

Back down to earth we go! Or try to, at least. In this episode Ben and Vaden attempt to ground their previous discussion on the philosophy of probability by focusing on a real-world example, namely the book The Precipice by Toby Ord, recently featured on the Making Sense podcast. Vaden believes in arguments, and Ben argues for beliefs.

Quotes
"A common approach to estimating the chance of an unprecedented event with earth-shaking consequences is to take a skeptical stance: to start with an extremely small probability and only raise it from there when a large amount of hard evidence is presented. But I disagree. Instead, I think the right method is to start with a probability that reflects our overall impressions, then adjust this in light of the scientific evidence. When there is a lot of evidence, these approaches converge. But when there isn’t, the starting point can matter.

In the case of artificial intelligence, everyone agrees the evidence and arguments are far from watertight, but the question is where does this leave us? Very roughly, my approach is to start with the overall view of the expert community that there is something like a one in two chance that AI agents capable of outperforming humans in almost every task will be developed in the coming century. And conditional on that happening, we shouldn’t be shocked if these agents that outperform us across the board were to inherit our future. Especially if when looking into the details, we see great challenges in aligning these agents with our values."
- The Precipice, p. 165

"Most of the risks arising from long-term trends remain beyond revealing quantification. What is the probability of China’s spectacular economic expansion stalling or even going into reverse? What is the likelihood that Islamic terrorism will develop into a massive, determined quest to destroy the West? Probability estimates of these outcomes based on expert opinion provide at best some constraining guidelines but do not offer any reliable basis for relative comparisons of diverse events or their interrelations. What is the likelihood that a massive wave of global Islamic terrorism will accelerate the Western transition to non–fossil fuel energies? To what extent will the globalization trend be enhanced or impeded by a faster-than-expected sea level rise or by a precipitous demise of the United States? Setting such odds or multipliers is beyond any meaningful quantification."
- Global Catastrophes and Trends, p. 226

"And while computers have been used for many years to assemble other computers and machines, such deployments do not indicate any imminent self- reproductive capability. All those processes require human actions to initiate them, raw materials to build the hardware, and above all, energy to run them. I find it hard to visualize how those machines would (particularly in less than a generation) launch, integrate, and sustain an entirely independent exploration, extraction, conversion, and delivery of the requisite energies."
- Global Catastrophes and Trends, p. 26

References:
- Global Catastrophes and Trends: The Next Fifty Years
- The Precipice: Existential Risk and the Future of Humanity
- Making Sense podcast w/ Ord (Clip starts around 40:00)
- Repugnant conclusion
- Arrow's theorem
- Balinski–Young theorem

#4 - The Hubris of Computer Scientists

Ben Chugg and Vaden Masrani — Mon, 08 Jun 2020 11:00:00 -0700

Are computer scientists recklessly applying their methods to other fields without sufficient thoughtfulness? What are computer scientists good for anyway? Ben, in true masochistic fashion, worries that computer scientists are overstepping their bounds. Vaden analyzes his worries with a random forest and determines that they are only 10% accurate, but then proceeds to piss of his entire field by arguing that we're nowhere close to true artificial intelligence.

References

"Good" isn't good enough, Ben Green.

"How close are we to creating artificial intelligence?", David Deutsch, Aeon

"Artificial Intelligence - The Revolution Hasn't Happened Yet", Michael Jordan, Medium

"Deep Learning: A Critical Appraisal", Gary Marcus

Errata

Vaden says "every logarithmic curve starts with exponential growth". This should be "every logistic curve stats with exponential growth".

Vaden says "95 degree accuracy". This should be "95 percent accuracy."

The three main rationalists were Descarte, Spinoza, and Leibniz, and the three main empiricists were Bacon, Locke, and Hume. (Not whatever Vaden said)