If foundation model companies want their government contracts renewed, they are going to have to make sure their AI output aligns with this administration's version of "truth".
This phrasing exactly corresponds to "politically correct" in its original meaning.
> I already am eating from the trashcan all the time. The name of this trashcan is ideology. The material force of ideology - makes me not see what I'm effectively eating. It's not only our reality which enslaves us. The tragedy of our predicament - when we are within ideology, is that - when we think that we escape it into our dreams - at that point we are within ideology.
So like making sure everyone knows that 2+2=5 and that we have always been at war with East Asia?
So no, democracy isn’t the ability to govern. It’s the ability to change those who govern, once every 5 years, i.e once every 4600 laws.
> In the AI context, DEI includes the suppression or distortion of factual information about race or sex; manipulation of racial or sexual representation in model outputs; incorporation of concepts like critical race theory, transgenderism, unconscious bias, intersectionality, and systemic racism; and discrimination on the basis of race or sex. DEI displaces the commitment to truth in favor of preferred outcomes and, as recent history illustrates, poses an existential threat to reliable AI.
https://www.whitehouse.gov/presidential-actions/2025/07/prev...
So... the concept of unconscious bias is verboten to the new regime? Isn't it just a pretty simple truth? We all have unconscious biases because we all work with incomplete information. Isn't this just a normal idea?
Let's see how that shakes out in this particular case.
Right, only what Chomsky works on is true science, unlike the intelligent systems pseudo science bullshit people like Geoff Hinton, Bengio or Demis Hassabis work on...
Personally, I find it somewhat amazing that you put Demis on that list given that he, himself, on very good accounts that I have, explicitly pushed back against natural language processing (and thus large language model) development at DeepMind for the longest of times and they had to play major catch up once it became obvious that their primarily reinforcement learning-oriented and "foundational" approaches were not showing as much promise as what OpenAI and Facebook were producing. Do not get me wrong, what he has accomplished is utterly amazing, but he certainly is not a father of large language models.
I have not, but I have watched him talk about this things many times and he always seemed too sure of himself and too dismissive of LLMs, I now believe he's simply wrong.