AI with Kyle Daily Update 015
Trump's AI War Plan vs China, Free Speech vs AI Safety, Will China Win Robotics Race
The skinny on what's happening in AI - straight from the previous live session:
Highlights
🇺🇸 Trump Declares AI War on China
Trump's administration released "Winning the AI Race: America's AI Action Plan" - basically a 28-page declaration of war against China in artificial intelligence.
The plan calls for removing all regulation, building massive data centres, and ensuring only "objective" AI models get government contracts. Translation: they want models more like Grok, not ChatGPT or Claude with their "left-leaning liberal bias."
Kyle's take: This is it then - it's the Cold War but with AI instead of nuclear weapons.
The language is aggressive: "counter Chinese influence," "export American AI to allies," "the AI race is America's to win."
They're backing this with serious cash too - £500 billion for the Stargate project alone, plus £200 million Department of Defence contracts for OpenAI, Google, Grok, and Anthropic just last week.
This is going to be the BIG topic for the next few months and beyond.
Source: White House AI Action Plan
🤖 The Problem with "Unbiased" AI Models
The Trump plan demands AI models be "free from top-down ideological bias," but Kyle explained why this is bollocks.
Most modern AIs goes through Reinforcement Learning from Human Feedback (RLHF) - humans literally grade what answers are acceptable. Someone has to decide what's "good" versus "bad" content from the internet's cesspit of hatred, racism, and conspiracy theories. And someone chooses who those humans are - generally the AI company building the model.
There's no such thing as "objective" - it's about whose values you're encoding. It’s still a choice.
Kyle's take: This is the classic right-wing tactic of claiming "objectivity" whilst pushing their own agenda. Most current models lean left because they're built in San Francisco by more liberal and socially accepting tech bros - no surprise there.
But when Grok tried to be "unbiased," it immediately started spouting anti-Semitic nonsense. The question isn't whether to have guardrails, it's whose ethics get baked in. Trump wants models that align with his worldview, not some mythical "neutral" AI that doesn't exist.
Source: Chip Huyen on RLHF
🏭 China Already Won the Robotics Race
Member Question from Aisha: "Is the next level of AI robotics and will China lead the way?”
China will have the ChatGPT moment for robotics, not America.
Whilst Boston Dynamics etc. leads research, they're focused on military and industrial units. China has the manufacturing capability to mass-produce consumer robots cheaply - the same advantage that let them dominate everything else we touch in our homes and lives.
Kyle's take: This race is already over. The West spent decades moving all our manufacturing to China for cheap labour, and now we're surprised they control production?
Building up Western manufacturing capability would take decades, and robotics will have its moment in the next few years. America can research all they want, but China will be the ones putting robots in people's homes.
Member Question: "How do you compare the UK action plan versus the US?"
Kyle's response: The UK doesn't have an action plan. They've got Keir Starmer and Peter Kyle making noises about using AI and have allocated £2bn but we’re not seeing a huge amount of movement.
The numbers are mental - America's throwing around hundreds of billions, nearly trillions. The UK talks about spending £2bn over 4 years. You can't enter this race with pocket change when the price of entry is measured in hundreds of billions. America knows this is the race to win against China. The UK is basically watching from the sidelines. With a cup of tea and a nice digestive biscuit.
This question was discussed at [13:24] during the live session.
Want to submit a question? Drop it below this video and I'll cover it in a future live.
Want the full unfiltered discussion? Join me for the daily AI news live stream where we dig into the stories and you can ask questions directly.