AI imaginings
The divergence of screen world and the real world looks set to accelerate. ChatGPT and its emerging text-based rivals are reshaping the possibilities of online interactions with their machine learning of natural language using LLMs, large language models. Synthesized voice is also on the cusp of a new generation of software sounding even more like real human voices, proper intonation included. Algorithms are demonstrating some ability to generate visual and musical art on the basis of no more than text descriptions. And most impressive of all, new convincing fakery is coming to instantaneous video, resulting from Google’s Dreamix research.
I’ll just link to the web page from the research with some current examples here. It should work fine on any web browser—there are only still and moving images, no audio and no ads. I’m not quite sure how to embed parts of that particular page here, so I encourage the click-through. Here is the short Tweet version:
Now, I’m no tech person. I’m comfortable enough with the gizmos and software as a general user, but I couldn’t do coding to save my life. So when it comes to the details of how all this works, I know little more than some basic terminology: AI, machine learning, neural networks—all of which make this new technology dynamic rather than static. The bots and algorithms are learning and improving as time marches forward.
While we are still at the beginning of exploring the abilities and limits of chatbots and spontaneous image generation, we can speculate freely about what may come from all of this technology based on the sorts of things we seem to do collectively. To begin with, some of us will be star-struck at the possibilities of AI generated video—or terrified and horrified. The initial reception will be too optimistic and too pessimistic all at once. Since we’re given to the negative, the earliest public policy responses likely will involve demands for a lot of overreach—if true to past patterns: based on moral panics—especially if the technologies threaten whole human job categories, whole livelihoods.
While Ned Ludd may have been a fictional character, weavers and knitters in 18th and 19th century Britain certainly felt their livelihoods were at stake because of knitting machines. There have been episodes of luddites through history as technology threatened to take people’s sources of income. It will be interesting to see whether these artificial intelligences wind up threatening the livelihoods of today’s credentialed upper middle classes. So far, technology has threatened to take the jobs of supposedly unskilled and menial laborers, not of college-educated types who manipulate words and symbols on screens for a living. It remains to be seen how welcoming the population will be that isn’t, per Trump’s description, “the poorly educated.” What will they do in response to technology putting downward pressure on their incomes?
For me, the worrisome thing about products of AI is that because they can seem so real, so slick, so much the fruit of a superior mind, people will take whatever AI tells them (whether or not AI is acknowledged to be involved) as representing a purer form of real--objective--reality.
Whereas in actuality, everything a machine says or does is ultimately traceable back to human programmers, who made certain decisions about what the machines will say or do, and why, and those decisions will have been influenced by human biases and ulterior motives, which humans always have.
It's better to be up front about what goals and perspectives were programmed into these creations. I can evaluate those for myself. If the end result is found to be sensible and useful, then the disclosure will take nothing away from that. And if there isn't that kind of disclosure, I'm going to be skeptical and reserve judgment about any given thing they try to slip past us.
Hey Marque, I got a strange email asking me to text t something...was that some bot on here or something?
I can't text if not...lol...no smart phone
Yep, a ton of them, a bot...I have to fight them on my forum...