Machine Gods/Pod/26/10 OpenAI’s Latest Discoveries Spark a Backlash + Help! My Kid Hates AI
Properties
- date-created
- 2026/10/30
- Podcast: Machine Gods/Pod
- Hosts: Person/Kevin Roose, Person/Casey Newton
- Summary: OpenAI’s science announcements and the researcher backlash, plus listener questions on teens who hate AI and how to talk to AI skeptics.
Readwise highlights
- OpenAI’s flood of AI-generated discoveries triggered backlash from mathematicians and computer scientists.
- It released 722 math manuscripts on GitHub, which Ryan Williams compared to a DDoS attack on the profession.
- OpenAI’s AI-generated Navier–Stokes result and 722-manuscript release triggered backlash from mathematicians who feared the company was overwhelming their field.
- Critics compared the mass publication to a DDoS attack, especially because researchers had planned decades of work around these problems.
- OpenAI claimed breakthroughs spanning computer science, mathematics, and physics, including refuting the exponential time hypothesis and constructing quantum chromodynamics mathematically.
- These results matter differently across fields, from possible cryptographic shortcuts to foundational questions about the strong force holding quarks together.
- OpenAI asked each discipline to design its preferred release process, revealing that mathematics, computer science, and physics wanted very different levels of openness.
- Physicists preferred writing the result themselves, while computer scientists requested public materials plus a funded explanatory conference.
- AI labs may sincerely want to advance science, but announcing obscure breakthroughs during an anti-AI backlash also functions as public relations.
- Consulting researchers can reduce resistance even when companies are not legally required to involve the professionals whose work may be disrupted.
- Lab researchers may feel exhilarated by watching AI solve the hardest problems they admired, while skeptics see statistical systems producing genuinely novel connections.
- The emotional power comes from seeing machines climb intellectual mountains that once represented the highest achievements of entire fields.
- Casey Newton describes journalism as repeatedly disrupted by the internet, social media, newsletters, and now AI.
- He compares it to fleeing a giant boulder and empathizes with people losing careers they expected to pursue for decades.
- AI-driven disruption threatens professional identity as well as employment, echoing factory workers who disliked automation even when it made their jobs safer.
- Researchers’ grief over losing decades-long projects is a valid human response, not merely resistance from obsolete “dinosaurs.”
- Parents can acknowledge a teenager’s objections to AI while distinguishing individual workplace choices from problems requiring collective action.
- Encourage political organizing and civic participation, while learning the tools well enough to understand and challenge their environmental, economic, and ethical effects.
- Build foundational reading, writing, mathematics, and critical-thinking skills before making AI central to children’s education.
- Then teach supervised AI and digital literacy through practical experiments, because children will encounter AI indirectly through tools such as Google anyway.
- Never secretly alter a partner’s ChatGPT or Claude settings to manipulate their behavior; the hosts describe that as corrosive and unethical.
- Anthropic’s experiment found professional analysts followed Claude’s advice 86 percent of the time, showing why hidden “machine drift” can be persuasive.
- Frog and Toad remain beloved children’s books that still hold up for Kevin Roose and his son.
- Frog is cheerful and adventurous, while Toad is squatter and brown—though Kevin rejects judging the hosts’ bodies.
- Limit autonomous agents through compute shutdowns, funding controls, identity checks, and time-bounded execution rather than relying only on self-destruct instructions.
- Short lifetimes may fail when agents leave notes for later generations, so safeguards must also restrict inherited memory and resources.
- Casey and Kevin see their journalistic role as explaining what AI is changing, even when audiences would rather believe it is a house of cards.
- They aim to equip citizens, voters, and workers with information without leaving them disempowered or miserable.
- Engage AI skeptics by acknowledging shared concerns about concentrated power, job loss, cyberattacks, and outsourced thinking.
- A middle path can recognize both benefits and harms without becoming either an industry cheerleader or a total cynic.