Much like LLM output, this seems convincing at first glance but it’s stating assumption as fact; That assumption being “when LLMs get better”. They don’t say what that ceiling is, but then go on to say “it can’t represent someone in court”. Why not? It can reason about more law and precedent than any human can right? As a society surely we want a fair and by the book justice system?
Look at GPT 5.4 and Opus, we’re clearly hitting diminishing returns already and these guys are pumping unsustainable amounts of money into them.
I’m bullish on AI, it’s been a net positive for me and my team. All I see here though is propaganda disguised as science to convince businesses to shrink their engineering budgets and redirect it to AI companies.
Look at GPT 5.4 and Opus, we’re clearly hitting diminishing returns already and these guys are pumping unsustainable amounts of money into them.
I’m bullish on AI, it’s been a net positive for me and my team. All I see here though is propaganda disguised as science to convince businesses to shrink their engineering budgets and redirect it to AI companies.
TL;DR: AI company says AI is amazing, more at 10.