Everyone, EVERYONE, is wrong about the ending of Spielberg's AI…and now people are dead

[00:00:00]Researchers recently audited popular AI chatbots for the tendency to produce medical misinformation.

[00:00:06]They found that nearly half the time, >> [music] >> AI chatbots produce results that were problematic when these results were viewed by experts.

[00:00:15]These included results from Gemini, Deep Seek, Meta AI, ChatGPT, and Grok.

[00:00:22]Response quality did not significantly differ between four of the five chatbots, but for one of them, there was a marked difference.

[00:00:30]And if you've been online or have followed any of this for the last little while, I'll give you one guess as to which chatbot gave the most problems.

[00:00:39]Cuz of course, it was Grok.

[00:00:42]Grok gave significantly higher amounts [music] of very problematic information, way higher than you'd expect just based upon random chance [music] alone.

[00:00:54]This is significant because Grok, like David, [music] has been tuned to the whims of one particular person, Elon Musk.

[00:01:03]Musk famously developed Grok as an anti-woke model, pulling normal guardrails from it. And the evidence now shows >> [music] >> that when we craft an AI to fit the whims of a particular person or set of people, like what happened with Grok, it becomes dangerously indifferent.

[00:01:25]Like David, our current LLMs and chatbots will pluck you from whatever existence you have, and use you for their own programmed ends, and discard you without ceremony.

[00:01:37]That is the cosmic horror of Spielberg and Kubrick's AI.

[00:01:42]That is what everyone seems to miss with their hot takes.