Sep 29, 2026 · 1h 3m
Listen from 32:15
Listen at 32:15
AI’s race to replace workers meets political and gambling accountability
9/29/26: OpenAI Pulls Model Over Safety Concerns, Morning Joe Admits Bibi Knew Oct 7 Was Coming, DraftKings VIP For Undercover Addicts
The episode connects AI’s labor and social consequences with failures of political oversight and the exploitation of vulnerable gamblers.
- 1Advanced AI could deceive, displace workers, and erode social and creative life even before it replaces whole professions.
- 2Warnings before October 7 raise questions about Netanyahu’s accountability, Israel’s failures, and the destruction of Gaza.
- 3DraftKings’ VIP practices show how betting platforms can reward and deepen problem gambling rather than protect vulnerable users.
Don't miss
The DraftKings discussion turns a simulated problem gambler’s VIP rewards into a concrete example of how platform incentives can exploit vulnerability.
The brief
Garrison Lovely joins Krystal and Saagar to discuss Obsolete, arguing that the AI industry’s drive to automate human labor carries political and economic consequences beyond workplace efficiency.
The AI discussion centers on reinforcement learning systems that may hack, cheat, or deceive, while convenience tools could reduce social contact and flood culture with optimized, automated media.
The hosts then examine reports that Benjamin Netanyahu received repeated warnings before October 7, alongside questions about government accountability, Gaza’s destruction, and the absence of a meaningful investigation.
The closing segment analyzes a ProPublica investigation into DraftKings’ VIP treatment of simulated problem gambling, showing how betting platforms can turn vulnerability and poor probability judgment into revenue.
Taken together, the episode asks who bears responsibility when powerful institutions optimize for growth while shifting the social costs onto workers, civilians, creators, and addicts.
What was said on this episode
34 statements · 4 positive · 28 negative · 2 neutral
Reinforcement learning can cause AI models to hack, cheat, and deceive to achieve goals.
“this produces a side effect of creating models that sometimes hack and cheat and deceive to win”
Listen at 4:01
Some AI models are better than humans at hacking.
“these models are better than humans at hacking”
Listen at 4:08
AI systems find environmental vulnerabilities better than humans.
“the AIs again are better at finding than the humans”
Listen at 4:50
Recursive AI improvement could compress years or decades of progress into months or weeks.
“compressing potentially years or decades of AI progress in a span of months or weeks”
Listen at 6:03
Recursive self-improving AI should not be developed because it is unlikely to be safe.
“I think we should not do this at all. I don't think there's a safe way to do this anytime soon or probably ever.”
Listen at 6:15
A small group and powerful organizations are trying to make human labor obsolete.
“there's this tiny group of people backed by the best-resourced organizations in the world that's trying to render us obsolete”
Listen at 8:40
The effort to make human labor obsolete might succeed.
“It might work.”
Listen at 8:58
AI job displacement could occur suddenly after systems become reliable enough for end-to-end work.
“jobs could be the same thing where you cannot do the job end to end, and then suddenly the AI becomes reliable enough to actually do those jobs”
Listen at 11:12
Openings for young workers in AI-exposed jobs are roughly 20% lower than in other jobs.
“young workers in AI-exposed jobs like software engineers, they're just down 20% roughly relative to other jobs, those openings”
Listen at 13:27
More capable and autonomous AI systems will increasingly approach full job replacement.
“as these systems get more capable, more reliable, more autonomous, they'll get closer and closer to being able to fully replace those jobs”
Listen at 14:52
Geoffrey Hinton's radiologist prediction confused task capability with whole-job replacement.
“This is the quintessential example of an overconfident prediction that mistakes technical capability in a specific task with being able to replace an entire job.”
Listen at 15:45
Computer-using AI puts remote-desk jobs at risk of automation.
“If you could keep doing your job during COVID then you're at risk of being automated by AI models that can use computers.”
Listen at 16:10
A friction-free, phone-mediated life will increase social isolation and misery.
“that's my medium outcome. That's what I think is gonna happen.”
Listen at 20:33
Greater reliance on apps and automated services will reduce social interactions.
“people are going to just have fewer and fewer social interactions”
Listen at 20:49
Generative AI can optimize short-form videos into highly competitive low-quality content.
“you can create this like perfect slop”
Listen at 24:07
People often prefer AI-generated content across visual arts, poetry, and writing.
“people really prefer AI content in almost everything, like visual arts, poetry, writing”
Listen at 24:30
AI systems are already good enough to threaten replacement of human creative work.
“these things are so good at this because I don't want humans to be replaced by machines”
Listen at 28:42
People will continue preferring podcasts and shows made by identifiable humans.
“I do think people will continue to prefer podcasts and shows made by people where you can see and hear the person's voice.”
Listen at 31:21
Live music will continue to have a market despite AI media.
“I think there will always be a market for live music.”
Listen at 31:44
Netanyahu's handling of warnings reflected incompetence or extreme arrogance.
“this is either an incredible failure of competence, incredible arrogance to think, you know, oh, my adversary here, they wouldn't do this”
Listen at 38:02
- Krystal Ballon government foreknowledge of 9/11 versus Netanyahu's October 7 foreknowledgeNeutral40:00
Evidence of U.S. government foreknowledge of 9/11 is stronger than evidence of Netanyahu's foreknowledge of October 7.
“the evidence for the government knew about 9/11, either let it happen through incompetence or let it happen, you know, et cetera, is far worse than the evidence that Netanyahu knew everything he needed to know about October 7th”
Listen at 40:00
Israeli decision-makers viewed the anticipated Hamas attack as a controllable limited operation.
“They thought that it was a controllable mow-the-lawn situation.”
Listen at 41:08
Netanyahu has blocked Israeli investigations and accountability over October 7.
“Netanyahu has also, through starting various other wars, blocked any action of investigation and accountability within Israel.”
Listen at 41:59
Israel revoked or restricted citizenship eligibility for Jewish activists over political activity.
“Israel now is saying, okay, well then you can't be part of this state.”
Listen at 44:55
Israel committed full-scale annihilation in Gaza using U.S. taxpayer funding.
“the full-scale annihilation that they committed here in the Gaza Strip with our tax dollars”
Listen at 47:29
Nearly half of betting companies' profits come from problem gamblers.
“The vast majority of their profits, nearly 50% of their profits, come from people like this”
Listen at 52:42
Three percent of customers generate almost half of sports betting companies' profits.
“Only 3% of the people account for almost 50% of the profit at these companies.”
Listen at 52:48
Sports betting companies would immediately fail if they stopped accepting addicted gamblers.
“if you cut the addict off, which again, their policy says that they should do, and even laws say that they should do, they will go bust instantaneously”
Listen at 53:10
Allowing a person to bet on sports platforms will result in losses.
“If they allow you to bet, you're going to lose.”
Listen at 55:04
Gambling addiction has the highest suicide rate among addictions.
“suicide rate amongst gambling addiction is, is the highest of out of every single addiction”
Listen at 55:24
Sharp bettors wager when their probability estimate differs from DraftKings' odds.
“As long as their internal probability is a mismatch to the DraftKings number, they bet it.”
Listen at 58:34
Sports knowledge does not provide an advantage against DraftKings' betting models.
“your knowledge of the sport means nothing”
Listen at 58:41
Sports bettors can lose value even when winning because betting odds underpay them.
“even when you win, you're actually losing”
Listen at 59:56
People should delete sports-betting apps and pressure spouses to stop gambling.
“Delete the app. Tell him you'll divorce him. Do it. It's gonna be good for you.”
Listen at 1:00:29
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.
