Keep pulling the thread on Gary Marcus.
People are betting trillions of dollars on the belief that Large Language Models are intelligent in ways they are not.
Large Language Models are inherently unreliable, unpredictable, and cannot be aligned to human values.
The AI industry has low differentiation, with dozens of companies building the same type of LLM, which makes it difficult to generate profit.
A group of attorneys general, led by New York State and involving approximately 46 states, has subpoenaed OpenAI to investigate issues including consumer data handling and model sycophancy.
The release of Anthropic's Mythos model prompted the US government to shift its stance towards needing to regulate AI.
Anthropic's Mythos model is capable of compromising poorly secured systems but cannot break into well-secured systems like those of the US government or major banks.
A new executive order from President Trump asks tech companies to voluntarily give the government oversight over new models before they are released to the public.
The state of Florida sued OpenAI, alleging the company played a role in a mass shooting by not intervening in a mentally ill user's interactions with its model.
President Trump has suggested the US government should acquire stakes in major AI companies, a proposal also backed by Bernie Sanders.
The proposal for the US government to take stakes in AI companies is a "backdoor bailout" for companies that are losing money.
OpenAI lost $21 billion on an operating profitability basis last year.
Approximately 10 studies have shown that most customers are not finding a return on productivity from implementing AI.