
GARY MARCUS: The conversation Silicon Valley doesn't want you to hear
In this second episode in our AI mini-series I met with Professor Gary Marcus live at the RAID conference in Brussels. Gary has been writing code since he was 10, built a Latin translation program at 16, and became a professor of psychology and neuroscience at NYU.
He's founded AI startups, testified before the US Senate, authored multiple books including his latest: Taming Silicon Valley: How to Protect Our Jobs, Safety, and Society in the Age of AI, meanwhile his Substack has over 80,000 subscribers who rely on him to cut through the hype.
When he warned that AI was heading toward catastrophe, Sam Altman called him a troll. Gary argues that large language models are a glorified autocomplete that hallucinate constantly. He also reveals why "P Doom" (probability of AI ending humanity) is overblown, but "P Dystopia" is approaching 100%. He explains why GPT-5 disappointed everyone, and why he believes we're witnessing the greatest theft of intellectual property in history. This is the conversation Silicon Valley doesn't want you to hear.
LESSONS YOU'LL LEARN FROM GARY:
P Dystopia is far more dangerous than P Doom. Forget AI ending humanity. Focus on the real threat: universal surveillance states, free misinformation, and the collapse of trust in truth itself.
Large language models don't understand the world, they just predict what words come next. That's why they still hallucinate constantly and, in Gary’s opinion, will never achieve AGI.
We're witnessing “the greatest data heist in history”. AI companies are training on all copyrighted material without paying a penny, with the ultimate aim of replacing everyone - including you.
Democracies are under threat from AI-powered misinformation. Generative AI is the "machine gun of disinformation" - making it faster, cheaper, and pitch-perfect.
Critical thinking is the only defense. In a world where misinformation is free to generate, teaching kids to question everything - especially AI output - is the most important skill we can develop.
Taming Silicon Valley: How to Protect Our Jobs, Safety, and Society in the Age of AI
https://www.amazon.co.uk/Taming-Silicon-Valley-Protect-Society/dp/0262551063
His Substack Marcus on AI is available here:
Otros episodios de "Crisis What Crisis?"



No te pierdas ningún episodio de “Crisis What Crisis?”. Síguelo en la aplicación gratuita de GetPodcast.







