James Corbett on Technocratic Control and the Dangers of AI
I’ve often marveled at the effectiveness of modern propaganda. Part of what makes it so effective is the availability of technology, from social media and search engines to large language model artificial intelligence. OpenAI’s ChatGPT has taken the world by storm and companies across a range of industries are already talking about replacing large numbers of white collar workers with AI.
This, even though there are serious problems with this technology. For example, we’re finding chatbots have a tendency to lie and fantasize. Researchers are calling these instances “hallucinations.” Basically, the AI is concocting a fantasy based on the information available and reciting it as fact. And that’s in addition to the bias that can be built in by programmers. So, while it’s an incredibly exciting technology, we cannot be naïve about its risks.
One obvious risk is that state-endorsed propaganda can become the only information available to people, as this technology starts monopolizing online searches and virtual assistants.
There won’t be a multitude of answers anymore. There will only be one, and he who controls the AI will have the power to control the beliefs of the entire world. Of course, yet another risk is that no one will be able to control it and the AI will control itself. I don’t know which might be worse. Corbett comments:
“You introduced this topic with the concept of propaganda and potential uses of large language models for propagandistic purposes. We should go back to the man who wrote the book on propaganda called ‘Propaganda,’ Edward Bernays, who [said]:
‘The conscious and intelligent manipulation of the organized habits and opinions of the masses is an important element in democratic society. Those who manipulate this unseen mechanism of society constitute an invisible government, which is the true ruling power of our country.’
That was Edward Bernays in 1928. His words are as true today as they were then, perhaps even more so. And the true ruling power of the country, of the world at this point perhaps, are those who can most effectively, consciously and intelligently manipulate the organized habits and opinions of the masses.
And I don’t think enough people have really stopped to cogitate on the fact that these large language models already starting to produce material that really cannot be distinguished from human-written material …
You don’t have to be a crystal ball prognosticator to see how this will extend out in the foreseeable future … [to] the point where you can have entire conversations, entire fields of interest and study that will be completely populated by artificial-created conversation …
A large language model that is able to accurately and without much prompting be able to populate botnets to flood social media and other places will essentially be able to dominate that conversation, [and] will consciously and intelligently manipulate the habits and opinions of the masses. At that point, you are talking about the ultim
Article from LewRockwell