Karen Howe, MIT mechanical engineer turned journalist and author of Empire of AI Dreams and Nightmares in Sam Altman's OpenAI, joins to discuss her investigation into the AI industry's first decade. Having interviewed over 250 people including 90+ current and former OpenAI employees, Howe provides an inside look at how OpenAI evolved from nonprofit to for-profit entity.

The conversation explores the imperial nature of AI companies, examining how they extract resources, exploit labor, and consolidate power while promising utopian futures. Howe draws parallels to Frank Herbert's Dune, comparing AI executives to leaders who step into messianic myths to control public perception while potentially losing themselves in their own mythology.

From Sam Altman's controversial firing and reinstatement to the environmental impact of massive data centers, Howe reveals the human cost of AI development. The discussion covers job displacement, data annotation labor, and the growing resistance movement as 80% of Americans now support AI regulation.

The Origins and Mythology of Artificial Intelligence

AI as a field began in 1956 when John McCarthy named it 'artificial intelligence' at Dartmouth, previously trying 'Automata Studies' - but colleagues worried about pegging the discipline to recreating human intelligence when we have no scientific consensus on what human intelligence actually is.

Companies use 'artificial general intelligence' as a moving target, with Sam Altman defining it differently for each audience: curing cancer for Congress, digital assistants for consumers, $100 billion revenue for Microsoft, and 'systems that outperform humans in most economically valuable work' on OpenAI's website.

The mythology requires both utopian and catastrophic narratives - 'worst case lights out for everyone, best case we cure cancer and solve climate change' - to justify anti-democratic control over AI development by claiming only they can prevent disaster.

Sam Altman's Manipulation and the OpenAI Power Struggle

In 2015, Altman wrote that AI development was 'probably the greatest threat to the continued existence of humanity' specifically to convince Elon Musk to co-found OpenAI, mirroring Musk's exact language about existential risk.

Altman later muscled Musk out by convincing Greg Brockman that Musk was too 'unpredictable' and 'erratic' to control such powerful technology, leading to Musk's departure and current legal vendetta.

The November 2023 board firing occurred after Ilya Sutskever and Mira Murati approached independent board members with concerns that Altman was 'creating too much instability' and 'pitting teams against each other' while building technology they believed could 'make or break the world.'

Board members discovered OpenAI's startup fund was actually 'Altman's startup fund,' not OpenAI's, representing one of several instances where 'there continuously are inconsistencies between the way that Altman is portraying what is being done versus what is actually being done.'

The Imperial Structure of AI Companies

Empire of AI identifies parallels between AI companies and historical empires: claiming resources not their own (data, IP), exploiting massive labor forces, monopolizing knowledge production, and using 'good empire vs evil empire' narratives to justify resource extraction.

Like Dune's houses fighting for spice control, AI executives step into messianic myths they create, with Howe noting: 'they know they're doing the myth-making, but also lose themselves in the myth because they have to live and breathe and embody it day in and day out.'

Companies control research by employing most AI scientists globally - 'if most climate scientists were bankrolled by fossil fuel companies, do you think we would get an accurate picture of the climate crisis?' - and censor inconvenient findings like Google firing Timnit Gebru.

OpenAI subpoenaed critics during its nonprofit-to-profit conversion, serving papers demanding 'every single piece of communication' involving Musk, demonstrating systematic intimidation of watchdog groups.

The Reality of Job Displacement and Data Labor

Klarna reduced from 6,000 to 3,000 employees while doubling revenue, with CEO Sebastian noting that by late 2023, 'even the most skeptical engineers basically said that coding has now been resolved and you don't need to code anymore.'

Data annotation became one of LinkedIn's top 10 fastest-growing jobs, with laid-off professionals including 'award-winning directors in Hollywood secretly doing this data annotation work to put food on the table.'

The New York Magazine piece revealed how data annotation workers live in constant anxiety, with one mother saying: 'I screamed at my child for distracting me because I don't know when the project is going to go, and I need to earn as much money as possible in this window.'

Anthropic's new report shows 40% reduction in entry-level jobs, with their disruption map indicating finance, law, media, and office work face displacement while physical jobs like construction and agriculture remain safe.

Environmental and Community Impact of AI Infrastructure

OpenAI's Texas data center will be the size of Central Park, running 1 million computer chips and consuming over 20% of New York City's power demand as part of the $500 billion Stargate Initiative.

Elon Musk's Memphis supercomputer uses 35 methane gas turbines, with residents discovering the facility when they 'literally smelled what seemed like a gas leak in all of their living rooms' in a predominantly Black community already facing environmental racism.

Meta's Louisiana facility will be one-fifth the size of Manhattan and use half of New York City's average power demand, with communities competing against these facilities for fresh water resources during droughts.

Companies specifically target vulnerable communities for data centers, creating a dynamic where 'if you are in the misfortunate category of being a have-not, you're competing with these facilities for fresh water, they're polluting your air, your bills have increased.'

The Path Forward: Breaking Up AI Empires

80% of Americans now support AI regulation in rare bipartisan consensus, with dozens of protests against data centers successfully stalling projects and banning facilities from localities nationwide.

The solution isn't eliminating AI but building 'bicycles of AI' like DeepMind's AlphaFold, which won the 2024 Nobel Prize in Chemistry using small curated datasets and significantly less computational resources while providing enormous benefit.

Resistance includes artists and writers suing for IP infringement, parents like Megan Garcia suing over AI chatbot harms to children, and communities exercising 'democratic contestation against the ways that the empires are going about their business.'

The goal is preventing companies from operating as empires that 'extract and exploit' without fair value exchange, instead creating 'mechanisms by which the very same capabilities could be developed with much more efficient methods, with much less resource consumption.'