Oct 22, 2026 · 1h 4m
AI’s promise meets the hard limits of crash prediction
The End Is Nigh. Buy Here?
The episode connects sweeping claims about AI and abundance with a practical investing lesson: uncertainty makes resilience more dependable than prediction.
- 1Extensive backtesting finds no universal crash hedge that works across radically different market crises.
- 2Diversification, risk management, and staying invested offer more practical protection than elaborate timing or hedging systems.
- 3AI could reshape work, healthcare, energy, and scarcity, but its social consequences remain difficult to forecast.
Don't miss
The hosts pivot from a 1960 prediction of the world ending in November 2026 to the idea that alarming forecasts can become solvable problems.
The brief
Tobias Carlisle and Jake Taylor open with the technological singularity, asking whether increasingly capable AI will make the future fundamentally harder to predict.
That question meets extensive investment backtesting: moving averages, CTAs, valuation signals, and hedges all struggle against crashes with different causes and timelines.
The practical conclusion is deliberately unglamorous: sensible diversification, risk management, and the ability to stay invested beat the fantasy of a universal crash-avoidance system.
The conversation then widens to healthcare, autonomous vehicles, local AI, abundant energy, and a future where software and intelligence may become cheap while physical experiences remain scarce.
A humorous 2026 doomsday prediction supplies the episode’s sharpest tonal turn, reframing alarming forecasts as sometimes solvable problems rather than reliable prophecies.
Carlisle’s Crash Test Dummy ties the themes together, arguing that investing means preparing not only for crashes but also for the recoveries and booms that follow.
What was said on this episode
42 statements · 24 positive · 16 negative · 2 mixed
Coding and some mathematical problems provide feedback loops for AI self-improvement.
“coding it does, you know, maybe obscure math problems it might.”
Listen at 2:18
Backtested investment strategies rarely persist across separate data sets.
“it's very hard to find anything that persists. From one set to the next.”
Listen at 3:35
Harry Browne’s simple permanent portfolio performs adequately across historical periods.
“a naive, like no idea of what's going to happen, it, it does, it does fine through the whole thing.”
Listen at 6:59
There is no satisfactory way to avoid market crashes.
“there's basically the answer is there's no way to do it satisfactorily.”
Listen at 8:49
Expensive markets imply lower future returns and somewhat higher volatility.
“the implications of an expensive market of lower returns and a little bit of volatility in the future.”
Listen at 9:44
The equity risk premium does not predict future returns.
“And it's not predictive at all.”
Listen at 11:33
High earnings yields precede better returns, while low yields precede poorer returns.
“High earnings yield leads to pretty good returns. Low earnings yield leads to pretty poor returns across a big data set.”
Listen at 12:05
Starting valuation may be most predictive of returns over a 20-year horizon.
“I thought it was most predictive at 20 years”
Listen at 13:11
The traditional 60/40 portfolio failed in 2022 because bonds and equities both declined.
“Didn't work in '22 because bonds sell off too.”
Listen at 14:53
The 60/40 portfolio’s diversification benefit has worked for only about 20 years.
“60/40, it's only worked for 20 years.”
Listen at 15:00
Moving-average strategies can diversify portfolios but do not solve all investment problems.
“moving averages are trend strategies. And that's an asset class all by itself that you can include, but it doesn't, it's not, it doesn't solve all your problems.”
Listen at 20:49
CTA performance records are distorted by survivorship bias.
“There's a real survivorship bias in those, I think.”
Listen at 21:44
Tobias’s VIX, HYG, and index put hedges expired worthless.
“And in every single instance, they have expired worthless.”
Listen at 22:33
A sensible asset-allocation mix is the most practical crash-management solution.
“I think that the solution is something like you have to have some sort of asset allocation mix”
Listen at 24:40
People should act as though catastrophic events will not end the world.
“you have to proceed as if it's not the end of the world.”
Listen at 26:54
Global fertility declined from 4.9 children per woman to roughly 2.25.
“global fertility went from 4.9 children per woman in 1950 to roughly 2.25 today.”
Listen at 31:10
The Montreal Protocol reduced CFC use by about 99%, putting ozone recovery on track.
“Today about 99% of those chemicals are no longer used and the ozone is on track to recover.”
Listen at 32:07
“In 1820, 1 adult in 10 could read. Today it's more like 9 in 10.”
Listen at 33:08
Access to learning will continue improving.
“you can learn almost anything nowadays and, and it'll keep getting better from here, I think.”
Listen at 33:35
Previously excluded people will produce major discoveries.
“we'll get discoveries from people who previously never had a chance to contribute on a big scale.”
Listen at 34:51
The world will not end on November 13, 2026 because of population overcrowding.
“the world will not end on November 13th due to 76 billion people crushing each other.”
Listen at 35:08
Some countries achieve similar healthcare outcomes at roughly half U.S. spending.
“there are other countries that get roughly equivalent healthcare outcomes that spend like half as much as we do.”
Listen at 39:21
Personalized medicine could substantially reduce healthcare inefficiency.
“personalized medicine can probably help a lot there”
Listen at 39:51
AI may generate many useful solutions by combining previously separate ideas.
“what if it's all just What if there's like a million suitcases with wheels that are just waiting to be— solutions put forth for us, ready to go?”
Listen at 41:51
OpenAI and Anthropic are already creating machines that improve machine intelligence.
“we're living in the singularity right now.”
Listen at 43:06
Autonomous cars are relatively close to widespread realization.
“The autonomous cars, one way or another, it's close-ish”
Listen at 43:49
Tesla remains roughly ten years away from autonomous driving.
“it's still 10-ish years away for Tesla.”
Listen at 44:01
Waymo’s autonomous vehicles appear to perform well on accidents per mile.
“it seems like statistically they're, they're doing well as far as like accidents per mile driven”
Listen at 44:29
Humanoid robots are close to practical deployment.
“they're close.”
Listen at 45:36
Older AI models already perform many needed tasks at little or no cost.
“we're already at the point where older models are quite good at doing a lot of stuff that people are going to need, and you basically get them for free.”
Listen at 47:22
Power, computing capacity, and data-center constraints are solvable.
“all of those problems are like solvable problems that they're not that difficult to solve.”
Listen at 47:49
AI can reconstruct underlying software from binary outputs.
“I think that the AI can take the binary, the output, and then turn it back into the underlying software.”
Listen at 49:08
AI-driven decompilation could make all software effectively crackable.
“All software's cracked.”
Listen at 49:45
Advanced robots and AI could perform physical and cognitive labor continuously without mistakes.
“it's, if you've got, if you've got a robot that can do all the physical labor. Yeah. And you've got this ultra intelligent machine that it's smarter than you are. It doesn't make mistakes. It works 24 hours a day.”
Listen at 49:57
Advanced automation may eliminate the need for human work.
“Maybe we don't have to work anymore.”
Listen at 50:29
The robot revolution has arrived.
“The robot revolution is here.”
Listen at 50:37
Abundant energy would enable greater availability of many other goods.
“If you have actually abundant energy, you can have a lot of other stuff.”
Listen at 53:16
Market crashes occur approximately once every seven years.
“there's a crash about once every 7 years.”
Listen at 54:42
Another market crash is highly likely.
“We're highly likely to get another crash.”
Listen at 54:44
Assets leading one market boom rarely lead the subsequent boom.
“The thing that worked in a preceding boom is rarely the thing that works in the next boom.”
Listen at 55:10
The AI singularity is imminent or may already have begun.
“the singularity is close if it's not already here.”
Listen at 57:07
Advanced AI could produce superabundance.
“abundance, right? Super abundance. That's the claim.”
Listen at 1:02:27
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.