It's over.

Playing with Claude Fable [00:00]

• The speaker has extensively used Claude Fable and finds it to be a very good model, capable of whipping up games quickly.
• Productivity tools are deemed obsolete, with the suggestion to create one's own. The speaker rebuilt their career by making productivity apps and now shifts to game creation.
• Around 3-4 game concepts were developed on a Friday, and Fable handled all tasks thrown at it without issue.

"I moved on to making games and I suggest you do the same."

The Generosity and Generosity of Anthropic's Plans [00:54]

• Anthropic's $200 plan is described as extremely generous, offering an estimated $8,000 worth of usage according to Semi Analysis.
• Anthropic is noted for aggressive strategies, having had an "incredible week" that has now seemingly come to an end.

"Anthropic's $200 plan is extremely generous, man."

The Political Game of AI Giants [01:14]

• The current state of AI has become a political game for industry titans, creating an information asymmetry where "peasants" lack insider knowledge.
• The speaker avoids speculating on the precise reasons behind government or corporate actions due to this information gap.
• The core driver behind actions is likely power and capital, not safety, as evidenced by the reaction to Fable.

"When in doubt, assume it's about power and capital. Power and money. Nothing else has ever mattered."

Anthropic's Dominance and Government Intervention [02:44]

• Anthropic is perceived to have gained too much power and is significantly ahead of other AI labs, which have considerable influence on the government.
• The typical cycle of AI model releases (Anthropic, OpenAI, Google) has been dominated by Anthropic for several cycles, leading to its higher valuation compared to OpenAI.
• The US government's intervention is interpreted as a move to "slow down" Anthropic and allow others a fairer chance, a principle that is hard to question.

"Anthropic has gotten too much power, way too much power. They are kicking everyone's ass."

The Politico Report and the "Wellness Retreat" Narrative [04:43]

• Politico reported a bizarre detail: the US government fabricated that Dario was at a wellness retreat when contacted.
• Anthropic denies this, stating Dario was available and on the phone quickly after being given an alternative contact.
• The speaker believes Anthropic on this point, viewing the wellness retreat story as propaganda.

"And Anthropic denies that. They're like, 'Dude, that is a total blatant lie.'"

Amazon CEO Andy Jassy as the "Snitch" [05:49]

• Initial news suggested Amazon CEO Andy Jassy was the one who alerted the White House about Fable's safety concerns.
• The narrative was that Jassy asked Fable to find bugs in Amazon's codebase, and it succeeded, prompting him to request the model be taken down.
• The speaker finds the idea of an AI finding bugs as a reason for concern to be absurd, as bug-finding is a core function of these models.
• Jassy reportedly demonstrated Fable's bug-finding capability to the White House, which was sufficient for them to act.

"So please take this model down which what the [ __ ] has AI been doing man for the last two years."

The "Game of Thrones" of AI Politics [07:17]

• The situation is likened to "Game of Thrones," where players are engaged in a game of power and propaganda.
• Sam Altman is humorously nominated to play "Littlefinger," known for playing all sides.
• The speaker presents scenarios for the underlying reasons, including a 10% chance of sophisticated collusion, a 45% chance that government individuals sought a stake in Anthropic's potential IPO, and a 45% chance that it's simply people improvising and reacting emotionally.

"It's just a game of power, man. It's a Game of Thrones. That's all it is."

The Role of Intelligence and Human Intent [17:05]

• The speaker emphasizes the continuous human input required for AI models. During their session, they were constantly prompting the model, acting as the "human with the intention."
• The model, by itself, "does nothing" and requires human guidance to assess quality and results.

"The model by itself is is just sitting there. It does nothing. And so it needs me."

Augmentation, Not Replacement [17:34]

• Many companies are shifting their narrative from AI "replacing" humans to "augmenting" them.
• This change is attributed to a sobering realization rather than just backlash; AI is seen as a "really cool calculator" that needs direction, not an entity with agency.
• The speaker expresses skepticism about the idea of AI replacing people, suggesting that while fewer people may do more work, the fundamental nature of the technology empowers humans.

"Fundamentally this technology does not replace humans. It um it empowers them."

A Category Error in Understanding AI [19:20]

• There's a fundamental "category error" in how AI is perceived, with potential for significant disruption and job losses, but not outright replacement of human roles.
• The speaker acknowledges the complexity and admits that even those building the technology don't fully understand the future implications.

"It's all true. It's all true. Like everything is simultaneously true."

Conversational Style and Future Content [20:44]

• The video adopts a conversational style, a format the speaker intends to continue for future content.
• The goal is to create a space for sharing thoughts and discussions on AI, acknowledging that opinions will vary.

"And as you can see, this is a conversational video and and I think this is the kind of style of videos I'll be doing from here on out."