- cross-posted to:
- technology@hexbear.net
- technology
- cross-posted to:
- technology@hexbear.net
- technology
The echoes of Y2K resonate in today’s AI landscape as executives flock to embrace the promise of cost reduction through outsourcing to language models.
However, history is poised to repeat itself with a similar outcome of chaos and disillusionment. The misguided belief that language models can replace the human workforce will yield hilarious yet unfortunate results.
I agree that machine learning will keep improving, and that there are jobs already being replaced. I’m just pointing out that it’s currently very far from being able to do what people are claiming. In particular, it’s pretty decent at tasks like content generation, or searching through information to pull up relevant details. So, jobs in these areas will be impacted. However, it’s simply a non starter for any work where there’s a specific correct result required. Stuff like self driving cars is a good example of this. Main problem is that the models are just trained on a bunch of text and images, and then they make inferences based on that. They don’t have any understanding in a human sense of what they’re doing because they don’t have a model of the world that we have.
In my opinion, really interesting AI developments will start happening once we start seeing models trained using embodiment, where they start encoding rules of the physical world internally through experience the way children do, and then they’re taught language within that context.
With regards to capitalism, it’s definitely going to be interesting to see how the system grapples with this. Ultimately, one of the fundamental contradictions is that companies need consumers with disposable income, and if you start doing mass automation of jobs, then you’re also eliminating your consumer demographic.